The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity 0198802609, 9780198802600

One of the books most central to late-antique religious life was the four-gospel codex, containing the Gospels of Matthe

407 28 8MB

English Pages 400 [384] Year 2019

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity
 0198802609, 9780198802600

Citation preview

OXFORD EARLY CHRISTIAN STUDIES General Editors GILLIAN CLARK ANDREW LOUTH

THE OXFORD EARLY CHRISTIAN STUDIES series includes scholarly volumes on the thought and history of the early Christian centuries. Covering a wide range of Greek, Latin, and Oriental sources, the books are of interest to theologians, ancient historians, and specialists in the classical and Jewish worlds. Titles in the series include: Liturgy and Byzantinization in Jerusalem Daniel Galadza (2017) The Roman Martyrs Introduction, Translations, and Commentary Michael Lapidge (2017) Philo of Alexandria and the Construction of Jewishness in Early Christian Writings Jennifer Otto (2018) St Theodore the Studite’s Defence of the Icons Theology and Philosophy in Ninth Century Byzantium Torstein Theodor Tollefsen (2018) Gregory of Nyssa’s Doctrinal Works A Literary Study Andrew Radde Gallwitz (2018) The Donatist Church in an Apocalyptic Age Jesse A. Hoover (2018) The Minor Prophets as Christian Scripture in the Commentaries of Theodore of Mopsuestia and Cyril of Alexandria Hauna T. Ondrey (2018) Preaching Christology in the Roman Near East A Study of Jacob of Serugh Philip Michael Forness (2018) Augustine’s Early Thought on the Redemptive Function of Divine Judgment Bart van Egmond (2018) God and Christ in Irenaeus Anthony Briggman (2018) The Idea Of Nicaea In The Early Church Councils, AD 431 451 Mark S. Smith (2018)

The Eusebian Canon Tables Ordering Textual Knowledge in Late Antiquity

M A T T HEW R . C R AW FO RD

1

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Matthew R. Crawford 2019 The moral rights of the author have been asserted First Edition published in 2019 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2018957123 ISBN 978 0 19 880260 0 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

Acknowledgements Eusebius was able to create his Canon Tables only because of the prior intellectual labour of Ammonius of Alexandria, and the ingenuity apparent in his creation owes a great deal to the stimulating environment of the Caesarean library. The present project is similarly indebted to a number of scholars who have offered insightful suggestions and criticisms along the way, and was similarly nurtured by two intellectually fertile institutional settings. I began writing this book while employed as a postdoctoral researcher in the Department of Theology and Religion at Durham University and completed it after making a transition to the Institute for Religion and Critical Inquiry at Australian Catholic University. The initial research was supported by a grant from the UK Arts and Humanities Research Council for the project ‘The Fourfold Gospel and its Rivals’, whose chief investigator was Francis Watson of Durham. The book owes more to Francis than to any other single person, since he was the one who introduced me to the Canon Tables and convinced me that it was a worthy object of scholarly scrutiny. His influence on my thinking on this matter extends well beyond the citations to his own scholarship in the pages that follow. The first presentation I gave on the Canon Tables was to the New Testament research seminar at Durham in early 2013, and shortly thereafter the Reverend Canon Rosalind Brown kindly invited me to give a public lecture on the topic to a lunchtime crowd at Durham Cathedral, in connection with the Lindisfarne Gospels exhibit in summer 2013. One of the surprising outcomes of both those events (surprising to me at least) was the degree of interest shown in the topic, an experience that has been replicated many times over in the subsequent four and a half years. In fact, I never would have imagined I would write a book on the Eusebian Canon Tables, but after repeatedly seeing the enthusiasm the material generated among scholarly audiences I finally realized that to do so was imperative. Papers that eventually became chapters were presented at the annual meetings of the North American Patristics Society, the Society of Biblical Literature, the Australasian Association for Byzantine Studies, and the Asia-Pacific Early Christian Studies Society, and I am grateful for the feedback I received on each of those occasions. An earlier version of chapter two was published as an article in New Testament Studies in 2015, and an earlier version of chapter six was published in 2017 in an edited collection entitled Producing Christian Culture: Medieval Exegesis and Its Interpretative Genres, edited by Giles Gasper, Francis Watson, and myself. Alessandro Bausi generously invited me to give a talk on Canon Tables at a conference on multiple-text manuscripts organized by the

vi

Acknowledgements

Centre for the Study of Manuscript Cultures in November 2016, which prodded me to delve into the world of manuscript studies, and he invited me to return in May 2018 for a stimulating conference entirely devoted to Canon Tables. In October 2017 I was able to return to Durham and, thanks to the invitation of Jane Heath, present some of the material in more developed form to the New Testament research seminar, where I was once again reminded that there are few places that are as collegial and intellectually engaging. In addition to these public events, chapters or portions of chapters in various stages were read by Lewis Ayres, Jeremiah Coogan, Ben Edsall, Jaś Elsner, Brian Gronewoller, Will Kynes, Margaret Mitchell, Judith McKenzie, Dawn LaValle Norman, Adam Ployd, Robert Thomson, and Jonathan Zecher, all of whom offered comments that have improved the final form of the book. Other scholars provided assistance in a variety of ways. Stephen Carlson helped me on more than one occasion with conversations about the Synoptic Problem and tracking down and citing Greek manuscripts. Michael Papazian answered multiple queries about Armenian grammar. Andrew Riggsby and Nathan Sidoli shared pre-publication versions of their work which turned out to be crucial for certain stages of the argument, and Peter Williams introduced me to the fascinating Codex Climaci Rescriptus, which I discuss in Appendix 3. For pressing me to consider the grander implications of the topic and for introducing me to the scholarship of Mary Carruthers (cited in two of the following chapters), I am particularly grateful to T. J. Lang, whom I had the pleasure of having as a colleague during my final year at Durham. The isolated hallway in the top level of the Dun Cow Cottage where we had our offices witnessed many conversations about paratexts, memory, and interpretation, which resulted in a co-authored article and provided the context within which the shape for the present book emerged. Four research assistants have helped in a variety of ways at differing stages of the project: Clift Ward, Carolyn Alsen, Ed Jeremiah, and Jon Simons. I should also thank the Research Office of ACU for providing the funds to cover costs associated with securing copyright permissions and printing the images included in the book, as well as funds for research trips to UCLA in April 2017 and to the British Library in October 2017 during which I was able to examine several Greek, Latin, Syriac, and Armenian gospelbooks related to this project. This financial support came as part of a five-year grant project titled ‘Modes of Knowing and the Ordering of Knowledge in Early Christianity’ (2017–21), a fitting theme since the present volume draws attention to one of the most innovative attempts at ordering knowledge in late antiquity. Finally, I should mention the libraries and individuals who have assisted me with finding the images that appear in the book and arranging the permissions to publish them here, including Judith McKenzie (Faculty of Oriental Studies, Oxford), Simon Elliott (Charles E. Young Research Library, UCLA), Julia Rodwell (National Gallery of Victoria, Melbourne), Sandra Powlette (British

Acknowledgements

vii

Library, London), Alexander Devine and Anne McLaughlin (The Parker Library, Corpus Christi College, Cambridge), Carl Graves (Egypt Exploration Society, London), Alessandro Moro (Shylock E-Solutions, Venice, who provided images of materials in the Biblioteca Nazionale Marciana), Gevorg Ter-Vardanian (Matenadaran, Erevan), Michael Gervers, Mikheil Tsereteli (Georgian National Museum, Tbilisi), Simone Verde (Complesso Monumentale della Pilotta, Parma), Florent Palluault (Médiathèque François-Mitterrand, Poitiers), Kerstin Herzog (Universitätsbibliothek, Augsburg), Mary Haegert (Houghton Library, Harvard), Sharon Sutton (The Library of Trinity College Dublin), Benedicta Erny (Universitätsbibliothek, Basel), Stefano Grigolato and Maddalena Piotti (Biblioteca Queriniana, Brescia), and Anna Rita Fantoni and Eugenia Antonucci (Biblioteca Medicea Laurenziana). This book is dedicated to two people who will probably never read it, but without whose friendship and kindness it would never have been completed: Hamish and Andy.

List of Illustrations 1. Mk §64 (in Greek ξΔ = Mk 6:35ff.) in The Gospel Book of Theophanes (second quarter of the twelfth c.)

4

National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 94r.

2. Canon I in The Gospel Book of Theophanes

5

National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 3r.

3. Mt §147 (in Greek ρμζ = Mt 14:14ff.) in The Gospel Book of Theophanes

6

National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 40r.

4. Lk §93 (in Greek ϘΓ = Lk 9:12ff.) in The Gospel Book of Theophanes

6

National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 152v.

5. Jn §49 (in Greek μθ = Jn 6:5ff.) in The Gospel Book of Theophanes

6

National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 211r.

6. The beginning of the gospel of John in Codex Sinaiticus (mid fourth c.)

8

London, British Library, Add MS 43725, fol. 247r (© The British Library Board).

7. Lk §305–9 (= Lk 23:10ff.) in the St Augustine Gospels (sixth c.)

9

The Parker Library, Corpus Christi College, Cambridge, MS 286, fol. 199v.

8. Canon IV in Peshitta Gospels (sixth c.)

10

Paris, Bibliothèque nationale de France, syr. 33, fol. 5v.

9. Lk §215–18 (= Lk 18:14ff.) in Codex Argenteus (sixth c.)

11

Uppsala, Uppsala University Library, MS DG 1, fol. 170r.

10. Canons VI and VII in the Gladzor Gospels (c.1300)

12

Gladzor Gospels, Library Special Collections, Charles E. Young Research Library, UCLA, Armenian MS 1, p. 16.

11. Canon I in The Bert’ay Gospels (tenth c.) Cambridge, MA, Harvard University, Houghton Library, MS Georgian 1, fol. 3v.

13

xiv

List of Illustrations

12. Canon II in Abba Garima I (sixth–seventh c.)

15

Ethiopia, Abba Garima Monastery, AG I, fol. 12r (© Michael Gervers, 2004).

13. Canon VIII, IX, and XMt of the London Canon Tables (sixth–seventh c.)

26

London, British Library, Add. 5111/1, fol. 11r (© The British Library Board).

14. Fragment of Greek Canon Tables from a papyrus codex found amidst the ruins of a monastery in Thebes (sixth–seventh c.)

36

New York, Metropolitan Museum of Art, accession no. X.455, P. Mon. Epiph. 584.

15. Canon IX in Codex Brixianus (sixth c.)

37

Brescia, Biblioteca civica Queriniana, ManoscrittoPurpureo, fol. 11v.

16. P. Oxy. 4168, side b. Fragments of a codex containing Ptolemy’s Handy Tables, dated to the fourth century

47

Courtesy of The Egypt Exploration Society and the University of Oxford Imaging Papyri Project.

17. P. Oxy. 4169, side a. Fragments of a codex containing Ptolemy’s Handy Tables, dated to the third century

48

Courtesy of The Egypt Exploration Society and the University of Oxford Imaging Papyri Project.

18. Portrait of Eusebius above the beginning of his Letter to Carpianus in the Gladzor Gospels (c.1300)

58

Gladzor Gospels, Library Special Collections, Charles E. Young Research Library, UCLA, Armenian MS 1, p. 4.

19. Portraits of Ammonius (right) and Eusebius (left) in the Rabbula Gospels (sixth c.)

88

Florence, Biblioteca Medicea Laurenziana, MS. Plut. 1.56, fol. 2r.

20. Image of Eusebius (top left), Carpianus (top right), and Ammonius (bottom) in the Parma Gospelbook (latter half of eleventh c.)

89

Parma, Biblioteca Palatina, MS. gr. 5, fol. 12v (by permission of the Ministry of Cultural Heritage and Activities and Tourism).

21. Mk §156 (Mk §ρνς = Mk 14:1a) in Codex Basilensis A.N.III.12 (ninth c.)

115

Basel, Universitätsbibliothek, A.N.III.12, fol. 143r.

22. Mt §211 (Mt §CCXI = Mt 21:12–13) in the Lindisfarne Gospels (eighth c.) London, British Library, Cotton MS Nero D IV, fol. 68v (© The British Library Board).

116

List of Illustrations

xv

23. Lk §94 (Lk §ϘΔ = Lk 9:18–20) in The Gospel Book of Theophanes (second quarter of the twelfth c.) 118 National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 153r.

24. The Johannine version of Peter’s commission in The Lindisfarne Gospels (eighth c.)

120

London, British Library, Cotton MS Nero D IV, fol. 258v (© The British Library Board).

25. First half of the Letter to Carpianus in the Rabbula Gospels (sixth c.)

160

Florence, Biblioteca Medicea Laurenziana, MS. Plut. 1.56, fol. 2v.

26. Mt §343–7 (= Mt 26:35ff.) in the Rabbula Gospels (sixth c.)

166

Florence, Biblioteca Medicea Laurenziana, MS. Plut. 1.56, fol. 84r.

27. Mt §354–5 in Codex Sangallensis 1395 (first half of fifth c.)

169

St Gall, Stiftsbibliothek 1395, p. 132.

28. Jn §55–8 (Greek numerals ΝΕ, ΝϚ, ΝΖ, ΝΗ) in Codex Basilensis A.N.III.12 (ninth c.)

170

Basel, Universitätsbibliothek, A.N.III.12, fol. 269r.

29. The poem Canon Evangeliorum by Ailerán of Clonard in the Augsburg Gospels (eighth c.)

201

Augsburg, Universitätsbibliothek, Cod. I.2.4°.2, fol. 1v.

30. Canon II in The Book of Kells (c.800)

202

Dublin, Trinity College Library, MS 58, fol. 2v. (© The Board of Trinity College Dublin).

31. A word square facing Ailerán’s poem in the Augsburg Gospels (eighth c.)

206

Augsburg, Universitätsbibliothek, Cod. I.2.4°.2, fol. 2r.

32. Christ in Majesty in Livre d’Evangiles de l’abbaye Sainte-Croix (eighth c.)

208

Poitiers, Médiathèque François Mitterrand, MS 17 (65), fol. 31r.

33. Canons V, VI, VII, and VIII in The Book of Durrow (seventh c.)

232

Dublin, Trinity College Library, MS 57, fol. 9v (© The Board of Trinity College Dublin).

34. Canons XLk and XJn in Abba Garima III (fifth–seventh c.)

234

Ethiopia, Abba Garima Monastery, AG II, fol. 260v (© Michael Gervers, 2004).

35. Canon V in the Lindisfarne Gospels (eighth c.) London, British Library, Cotton MS Nero D IV, fol. 15r (© The British Library Board).

235

xvi

List of Illustrations

36. Canon XJn in The Gospel Book of Theophanes (second quarter of the twelfth c.)

236

National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 7v.

37. Tholos image in Abba Garima I (sixth–seventh c.)

238

Ethiopia, Abba Garima Monastery, AG II, fol. 258v (© Michael Gervers, 2004).

38. Fountain of life image in the Gospels of St Médard de Soissons (before 827)

238

Paris, Bibliothèque nationale de France, MS lat. 8850, fol. 6v.

39. Tholos with a hypothesis inscription, serving as a frontispiece for the sequence of Canon Tables that follow in a Greek gospelbook (ninth c.)

239

Venice, Biblioteca Nazionale Marciana, MS gr. I 8, fol. 3r.

40. Tholos image in the Adishi Gospels (897), fol. 5v

239

Mestia, Svaneti Museum, Georgian National Museum.

41. Canon II in the Gospels of St Médard de Soissons (before 827)

245

Paris, Bibliothèque nationale de France, MS lat. 8850, fol. 9r.

42. Portraits of Matthew (right) and John (left) alongside Canon VII in the Rabbula Gospels (sixth c.)

247

Florence, Biblioteca Medicea Laurenziana, MS. Plut. 1.56, fol. 9v.

43. First xoran in the Etchmiadzin Gospels (989)

261

Erevan, Matenadaran, MS 2374, fol. 1r.

44. Fourth xoran in the Etchmiadzin Gospels (989)

262

Erevan, Matenadaran, MS 2374, fol. 2v.

45. Seventh xoran in the Etchmiadzin Gospels (989)

263

Erevan, Matenadaran, MS 2374, fol. 4r.

46. A pair of doves in the Etchmiadzin Gospels (989)

264

Erevan, Matenadaran, MS 2374, fol. 6v.

47. Partridges perched atop the fourth xoran in the Etchmiadzin Gospels (989)

264

Erevan, Matenadaran, MS 2374, fol. 2v.

48. Cocks on top of xoran eight in the Gladzor Gospels (c.1300)

265

Gladzor Gospels, Library Special Collections, Charles E. Young Research Library, UCLA, Armenian MS 1, p. 17.

49. Herons perched atop the eighth xoran in the Etchmiadzin Gospels (989) Erevan, Matenadaran, MS 2374, fol. 4v.

265

List of Illustrations 50. Peacocks above the first xoran in the Etchmiadzin Gospels (989)

xvii 266

Erevan, Matenadaran, MS 2374, fol. 1r.

51. Pomegranates growing above the third xoran in the Etchmiadzin Gospels (989)

266

Erevan, Matenadaran, MS 2374, fol. 2r.

52. Date palm growing alongside xoran ten in the Gladzor Gospels (c.1300) Gladzor Gospels, Library Special Collections, Charles E. Young Research Library, UCLA, Armenian MS 1, p. 21.

267

Abbreviations BO

Biblica et Orientalia

CCSL

Corpus Christianorum Series Latina

CSCO

Corpus Scriptorum Christianorum Orientalium

CSEL

Corpus Scriptorum Ecclesiasticorum Latinorum

GCS

Die griechischen christlichen Schriftsteller

LCL

Loeb Classical Library

LSJ

Liddell, Scott, Jones, A Greek English Lexicon (Oxford, 1843; 9th edn 1940)

LXX

Septuagint

OECS

Oxford Early Christian Studies

OECT

Oxford Early Christian Texts

PG

Patrologia Cursus Completus: Series Graeca

PL

Patrologia Cursus Completus: Series Latina

PTS

Patristische Texte und Studien

SC

Sources Chrétiennes

ST

Studi e testi

STAC

Studien und Texte zu Antike und Christentum

TU

Texte und Untersuchungen zur Geschichte der altchristlichen Literatur

VCSup

Supplements to Vigiliae Christianae

Introduction When modern readers open up nearly any edition of the Bible available for purchase today, they encounter much more than the bare text of the ancient documents that comprise the canon of Christian scripture. Any modern reader of the Bible will instead be confronted with a penumbra of additional material intended by editors as aids for the reading or interpretation of the main text. In some cases, such as with study Bibles, this secondary material perhaps outweighs the primary text, with the inclusion of tables of contents, prefaces, explanatory notes, alternative translations or textual traditions, section headings, cross-references, and maps—not to mention chapters and verses. Most Bibles contain less than this, but almost all have at a minimum cross-references to guide the reader to other pertinent passages, or perhaps indexes to allow a reader to find with ease passages that speak to a given topic. So common are these reading aids that they hardly seem remarkable, just another mundane feature of modern life that one can count on always being present, just as one can assume that the local coffee shop will have Wi-Fi available for its customers. Yet it was not always so. Just as the technology that allows a laptop to connect to the wider world through the Internet has a distinct history of its development and use, so also the technology of reading aids—for that is what they are, a technology—did not always exist in the ‘taken for granted’ category that we now perceive them to be. Rather, they too have a certain background out of which they emerged, and, again, just like the Internet, the personal computer, or moveable type, they were a technological advance that created hitherto unforeseen possibilities. This book is about one specific form of reading technology that arose in the early fourth century CE and has influenced the reading of the Christian scriptures ever since. The most prolific Christian author of this period, who experimented with an astonishingly diverse range of genres, topics, and technologies, was Eusebius Pamphili, who lived through the great persecution under Diocletian, then became bishop of Caesarea Maritima, was present at the Council of Nicaea in 325, and delivered an oration for the thirtieth anniversary of Constantine’s reign in 335. Famous as the author of the earliest surviving ecclesiastical history, Eusebius also left to posterity a world The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0001

2

The Eusebian Canon Tables

chronicle, commentaries on biblical books, theological tractates, apologetic works, and a geographical handbook, while also managing the most significant Christian library then in existence and spearheading the most ambitious production of biblical manuscripts of his time.¹ Yet the most successful of all his literary endeavours was the cross-referencing system he developed for the church’s fourfold gospel canon, commonly known as the ‘Canon Tables’. By the time Eusebius became a bishop, there was little doubt that at the centre of the emerging Christian canon there sat the four accounts of Jesus’ life traditionally attributed to Matthew, Mark, Luke, and John, the first and last being apostles while the second and third were assumed to have been followers of the apostles Peter and Paul respectively. Despite the fact that each of these books presents itself as a self-contained and sufficient account of Jesus’ story, Christians from the late second century onwards made the rather ecumenical decision to preserve all four. This choice had complications for users of these texts, for, as even modern readers of the gospels can attest, the four accounts are so similar and yet so distinct that it is often difficult to recall which of the gospels contains a certain story or teaching, or how the parallel stories in multiple gospels differ from one another. Did Jesus pronounce a blessing upon the ‘poor’ without qualification or was the saying more specifically for the ‘poor in spirit’? Does the parable of the good Samaritan occur in only one gospel or in multiple gospels? Don’t all the gospels report the feeding of the 5,000? Or was it the feeding of the 4,000? Such difficulties are only compounded if the edition of the gospels one uses is devoid of any chapters and verses and the other aids modern readers enjoy. Navigation within and among these four narratives becomes a time-consuming process of relying on often-faulty memory accompanied by much scanning of pages. It was this problem that Eusebius’ Canon Tables sought to address. His solution was a marginal apparatus that comprised three elements. First, at the beginning of the four gospels he placed his Letter to Carpianus, a sort of brief instruction manual in which he explained to the reader why he created the system and how it operated.² Second, within each gospel itself, Eusebius divided the text into numbered sections, beginning with 1 and continuing on to the end of each gospel.³ These sections were demarcated on the basis of ¹ For an introduction to Eusebius, see Aaron P. Johnson, Eusebius, Understanding Classics (London: I. B. Tauris, 2014). Two recent collections of essays illustrate the lines of research currently under way on this important figure: Sabrina Inowlocki and Claudio Zamagni, eds, Reconsidering Eusebius: Collected Papers on Literary, Historical, and Theological Issues, VCSup 107 (Leiden: Brill, 2011); Aaron Johnson and Jeremy Schott, eds, Eusebius of Caesarea: Tradition and Innovations, Hellenic Studies 60 (Washington, DC: Center for Hellenic Studies, 2013). ² For a translation of the letter, see Appendix 1. ³ For the following study I will be relying on the edition of the Greek Canon Tables found in the Nestle Aland 28th edition. In this version, Matthew has 355 sections, Mark has 233, Luke has 342, and John has 232. For the biblical citations that follow, I have included both the modern chapter and verse enumeration alongside the Eusebian section number, using the symbol § to

Introduction

3

whether or not they had parallels in other gospels, and, if so, how many other gospels, with the result that some sections are longer than a single chapter in today’s reckoning, while others are merely half a verse. Finally, Eusebius collated the numbers for each of the gospels in ten tables, called ‘Canons’, which he placed at the beginning of the fourfold gospel, following his Letter to Carpianus.⁴ Each of these Canons represents a different set of passages that could be classified together on the basis of which gospels they appeared in.⁵ So, for example, Canon I presented the numbers for those passages that occurred in all four gospels, with the numbers for each set of parallel passages grouped together, placed alongside one another in a given row of the table. He continued on in this fashion as follows for the remaining tables:⁶ Canon II Canon III Canon IV Canon V Canon VI Canon VII Canon VIII Canon IX Canon XMt Canon XMk Canon XLk Canon XJn

Matthew Mark Luke Matthew Luke John Matthew Mark John Matthew Luke Matthew Mark Matthew John Luke Mark Luke John Matthew alone Mark alone Luke alone John alone

The rationale of the system is that by reading horizontally across the rows within the first nine tables one can find passages that are similar in another mark the latter. A useful conversion table between the two reference systems can be found at https://danielbwallace.files.wordpress.com/2014/04/eusebian canons conversion table.pdf [accessed on 21 Dec 2017]. ⁴ Although strictly speaking only these prefatory tables are the ‘Canon Tables’, in keeping with prior scholarly convention I will on occasion use the phrase to refer to the entire paratextual system, comprising the ten tables, the marginal notation throughout the gospels, and the Letter to Carpianus. ⁵ In keeping with the convention of the NA²⁸, I will always give the section number in Arabic numerals and Canon numbers in Roman numerals. Eusebius made a similar typographic distinction, writing the section number in black and the Canon number below it in red (both of course as Greek numerals), a convention that he describes in his Letter to Carpianus. pace Thomas O’Loughlin, who asserted that the colour distinction and placement of one numeral above the other in the margin were innovations introduced by Jerome in his Latin translation: Thomas O’Loughlin, ‘The Eusebian Apparatus in Some Vulgate Gospel Books’, Peritia 13 (1999): p. 4; Thomas O’Loughlin, ‘Harmonizing the Truth: Eusebius and the Problem of the Four Gospels’, Traditio 65 (2010): p. 14; Thomas O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels: Ailerán’s Kanon Euangeliorum as a Lens for Its Appreciation’, in The Lindisfarne Gospels: New Perspectives, ed. Richard Gameson (Leiden: Brill, 2017), p. 99. ⁶ The reader will observe that the tenth and final Canon actually comprises four separate tables. On this aspect of the system, see the discussion in chapter two.

4

The Eusebian Canon Tables

gospel or gospels, or, by reading the passages listed in the tenth Canon, find passages unique to each gospel. A simple example will illustrate the use of the apparatus. Suppose you are reading along in the Gospel of Mark and come across Mark’s telling of the feeding of the 5,000 (Mark 6:35-44 in modern reckoning). ‘Doesn’t this also occur in one of the other gospels?’ you wonder. Eusebius’ system is designed to answer this question for you. In an edition of the gospels equipped with his apparatus, if you look alongside the passage, in the margin of the page, you will see two numbers. The first number, in black, is the number for this section within Mark’s gospel, which happens to be §64 (in Greek ξΔ) (see fig. 1).

Fig. 1. Mk §64 (in Greek ξΔ = Mk 6:35ff.) in The Gospel Book of Theophanes (second quarter of the twelfth c.). National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710-5), fol. 94r

Beneath or alongside the black number is another number in red that tells you in which of the tables you should look to find this passage. Here the red digit is α (= 1), so you turn to the first Canon, which reports parallels among Matthew, Mark, Luke, and John. Immediately, then, you realize that this is a story that occurs in all four gospels. Now, within Canon I, if you scan down the column that lists numbers for Mark, you should eventually come to the number 64 (see fig. 2). For the sake of illustration, here is the row that includes this number, with the preceding and subsequent rows included for comparison: Mt

Mk

Lk

Jn

142 147 166

51 64 82

21 93 94

35 49 17

Once you have located Mark section 64, you can read horizontally, to the left and to the right, to find that that the feeding of the 5,000 is §147 in Matthew, §93 in Luke, and §49 in John. Then, if you were to turn to those numbered passages in the other gospels, you would find the parallel accounts

Introduction

5

Fig. 2. Canon I in The Gospel Book of Theophanes. Each column contains numbers for passages within a given gospel, as indicated by the two letter abbreviations at the top of each column. The second column is for Mark and the first number listed on the right side of the page is ξδ (= Mk §64). (Note that the scribe has here written a lowercase delta, whereas in the page of Mark shown previously he has used an uppercase delta.) National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 3r

6

The Eusebian Canon Tables

of the feeding of the 5,000, and by flipping back and forth between gospels it would be easy to compare the differing versions (see figs 3–5).⁷

Fig. 3. Mt §147 (in Greek ρμζ = Mt 14:14ff.) in The Gospel Book of Theophanes. National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710-5), fol. 40r

Fig. 4. Lk §93 (in Greek ϘΓ = Lk 9:12ff.) in The Gospel Book of Theophanes. National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710-5), fol. 152v

Fig. 5. Jn §49 (in Greek μθ = Jn 6:5ff.) in The Gospel Book of Theophanes. National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710-5), fol.211r

Eusebius’ system, therefore, had the same effect as the cross-references that often occur today in the margins of printed Bibles, though he had to devise a method that worked in the absence of chapter and verse divisions, and, indeed, ⁷ The preceding paragraph is largely a paraphrase of the latter half of the Letter to Carpianus.

Introduction

7

in the absence of any comparable system of numerical cross-references for a corpus of literature that he might have used as a model. Though obvious enough to modern readers, this was an innovation of remarkable proportions for its day. Accordingly, the aim of the present monograph is to highlight the ingenuity that went into its creation and the way in which readers interacted with it in the eight centuries thereafter. Its influence may be seen simply by taking stock of the surviving gospel manuscripts in the various languages of late antiquity. Three of the four early Greek uncials from the fourth and fifth centuries contain traces of the Eusebian apparatus in the margins, though none of them preserves the prefatory Canons: Codex Sinaiticus, Codex Alexandrinus, and Codex Ephraemi Rescriptus.⁸ The oldest of these, Codex Sinaiticus, was probably written only a few decades at most after Eusebius’ death (see fig. 6). Moving from the world of Greek into Latin, we find the Canons in the majority of Vulgate gospelbooks, including most of the well-known ones, such as Codex Fuldensis (6th c.), the St. Augustine Gospels (6th c.), and the Lindisfarne Gospels (early 8th c.),⁹ not surprisingly, since Jerome incorporated a Latin version of the apparatus in the new Vulgate edition he created for Pope Damasus in the early 380s (see fig. 7). At some point in the century after Jerome finished his Vulgate translation, an unknown scribe writing in Syriac produced a revised version of Eusebius’ apparatus to accompany his new translation of the gospels, known today as the Peshitta, and as a result the majority of Peshitta manuscripts likewise contain the marginal system, including well-known examples such as the Rabbula Gospels, copied in the year 586 (see figs 19, 25, 26, 42),¹⁰ and a sixth-century copy of the gospels in Paris (see fig. 8). ⁸ For an overview of these manuscripts, see D. C. Parker, An Introduction to the New Testament Manuscripts and Their Texts (Cambridge: Cambridge University Press, 2008), pp. 71 4, and for specific studies see Dirk Jongkind, Scribal Habits in Codex Sinaiticus, Texts and Studies, Third Series, Volume 5 (Piscataway, NJ: Gorgias Press, 2007), pp. 109 20; W. Andrew Smith, A Study of the Gospels in Codex Alexandrinus: Codicology, Palaeography, and Scribal Hands, New Testament Tools, Studies and Documents 48 (Leiden: Brill, 2014), pp. 139 56. ⁹ Of these three, the Canon Tables in the Lindisfarne Gospels have received the most amount of attention. See especially Michelle Brown, The Lindisfarne Gospels: Society, Spirituality and the Scribe (London: The British Library, 2003), pp. 179 82, 300 5. On the Canon Tables in the St. Augustine Gospels, see Francis Watson, The Fourfold Gospel: A Theological Reading of the New Testament Portraits of Jesus (Grand Rapids: Baker Academic, 2016), pp. 106 13, who argued that the miniature images that surround the portrait of the evangelist Luke all depict scenes unique to this gospel, information gleaned from the Eusebian apparatus. ¹⁰ Cf. Massimo Bernabò, Il Tetravangelo di Rabbula. Firenze, Biblioteca medicea laurenziana, Plut. 1.56: L’illustrazione del Nuovo Testamento nella Siria del VI secolo, Folia Picta 1 (Rome: Edizioni di Storia e Letteratura, 2008); Massimo Bernabò, ‘The Miniatures in the Rabbula Gospels: Postscripta to a Recent Book’, Dumbarton Oaks Papers 68 (2014): pp. 343 58, who argued that only the pages of gospel text, which include the Eusebian apparatus in the margin, were copied in 586, and that the folios containing the set of famous illustrated Canon Tables were added centuries later from a separate sixth century Syriac mansucript.

8

The Eusebian Canon Tables

Fig. 6. The beginning of the gospel of John in Codex Sinaiticus (mid fourth c.). The text of the Gospel of John begins at the top of the left column, and, in the margin next to the text, one can see an Α, indicating Jn §1 (= Jn 1:1ff.) and beneath it a Γ, showing that this is a Canon III passage. This specific parallel, which combines the Johannine prologue with the synoptic genealogies, is discussed in chapter three. London, British Library, Add MS 43725, fol. 247r (© The British Library Board)

Introduction

9

Fig. 7. Lk §305 9 (= Lk 23:10ff.) in the St Augustine Gospels (sixth c.). The Parker Library, Corpus Christi College, Cambridge, MS 286, fol. 199v

The earliest copy of the gospels in Gothic, the manuscript known as Codex Argenteus, which was probably made in Ostrogothic Italy in the sixth century for King Theodoric the Great, also contains the Eusebian system, though the ten prefatory tables are now lost (see fig. 9).¹¹

¹¹ On Codex Argenteus, see further Carla Falluomini, The Gothic Version of the Gospels and Pauline Epistles: Cultural Background, Transmission and Character (Berlin: De Gruyter, 2015), pp. 32 4, and on Canon Tables in Gothic, see Carl Nordenfalk, Die spätantiken Kanontafeln, 2 vols (Göteborg: O. Isacsons boktryckeri a. b., 1938), pp. 261 9; Falluomini, The Gothic Version, pp. 53 4.

10

The Eusebian Canon Tables

Fig. 8. Canon IV in Peshitta Gospels (sixth c.), showing parallels in Matthew, Mark, and John (moving from right to left). Paris, Bibliothèque nationale de France, syr. 33, fol. 5v

Introduction

11

Fig. 9. Lk §215 18 (= Lk 18:14ff.) in Codex Argenteus (sixth c.). The parchment page has been dyed purple and the text has been written in silver, with gold being used for the incipit of each Eusebian section. The architectural frames in the bottom margin list the Lukan passages on the page and the parallels they have in the other gospels, in the sequence Luke, Mark, Matthew, John. For a discussion of this feature, see the section ‘Marginal Concordance Tables in Syriac Manuscripts’ in chapter five. Uppsala, Uppsala University Library, MS DG 1, fol. 170r

12

The Eusebian Canon Tables

Although the earliest surviving Armenian gospelbooks are considerably later, dated to the tenth century at the earliest, they regularly also include the Eusebian Canon Tables, and the uniformity of their presentation suggests that the apparatus must have entered the Armenian tradition at an early stage (see fig. 10).

Fig. 10. Canons VI and VII in the Gladzor Gospels (c.1300). Gladzor Gospels, Library Special Collections, Charles E. Young Research Library, UCLA, Armenian MS 1, p. 16

Introduction

13

Similarly, the earliest surviving Georgian gospelbook, the Adishi Gospels, dated to 897, contains the Eusebian apparatus (see fig. 40 in chapter 7), as does a tenth-century copy of the gospels at Harvard (see fig. 11).

Fig. 11. Canon I in The Bert’ay Gospels (tenth c.). Cambridge, MA, Harvard University, Houghton Library, MS Georgian 1, fol. 3v

14

The Eusebian Canon Tables

Finds made only in the last decade have continued to add to this list. Two fascinating gospelbooks from the ancient kingdom of Axum (in modern-day Ethiopia) preserve a version of the system in Ge‘ez translation. These manuscripts, known as Abba Garima I and Abba Garima III, were once thought to be medieval, but have recently been redated to the late antique period (see fig. 12).¹² Slightly later but unquestionably unique are the remnants of a copy of the Gospel of John written in Caucasian Albanian that were preserved as a palimpsest in the library of Saint Catherine’s Monastery at Mount Sinai. The fragments of this manuscript, dated by modern editors to between the seventh and tenth centuries but reflecting a translation that must have occurred much earlier, contain the Eusebian apparatus in a language that has almost entirely perished from the written record.¹³ Hence, by the seventh century, whether one was reading a gospelbook written in Greek, Latin, Gothic, Syriac, Armenian, Georgian, Ge’ez, or even Caucasian Albanian, there was a good chance that it would be equipped with Eusebius’ numeric cross-referencing system.¹⁴ In view of this widespread diffusion, it would not be an exaggeration to describe his Letter to Carpianus as the most widely translated and copied non-biblical Christian text in the entire late antique period.¹⁵ Moreover, the influence of Eusebius’ system continues today. The latest editio critica minor of the Greek ¹² Cf. Judith McKenzie and Francis Watson, The Garima Gospels: Early Illuminated Gospel Books From Ethiopia (Oxford: Manar al Athar, University of Oxford, 2016). ¹³ Cf. Jost Gippert et al., eds, The Caucasian Albanian Palimpsests of Mt. Sinai, 2 vols, Monumenta Palaeographica Medii Aevi. Series Ibero Caucasica (Turnhout: Brepols, 2008). The remains of the two palimpsests discussed in this volume, one a lectionary and one a copy of the Gospel of John, are the only surviving manuscripts containing text written in Caucasian Albanian. I am grateful to Philip Forness for drawing my attention to them. ¹⁴ Some readers will no doubt notice that Coptic is missing from this list. The earliest surviving evidence for the transmission of the Eusebian apparatus in Coptic comes from the twelfth and thirteenth centuries, and it is unclear when this Coptic translation was made. See G. Horner, The Coptic Version of the New Testament in the Northern Dialect (Oxford: Clarendon Press, 1898), pp. xli xliv, lviii lxi, lxxii lxxiv; Harold H. Oliver, ‘The Epistle of Eusebius to Carpianus: Textual Tradition and Translation’, Novum Testamentum 3 (1959): p. 142; Carl Nordenfalk, ‘Canon Tables on Papyrus’, Dumbarton Oaks Papers 36 (1982): p. 30. The Coptic version of the Letter to Carpianus, based on one of these later manuscripts (Vat.Copt. 9), was published in Ad. Hebbelynck, ‘Les κεϕάλαια et les τίτλοι des évangiles’, Le Muséon 41 (1928): pp. 114 18. Fragments from a Sahidic gospelbook containing traces of the Eusebian apparatus were mentioned by J. B. Lightfoot in Frederick Henry Ambrose Scrivener, A Plain Introduction to the Criticism of the New Testament for the Use of Biblical Students, 3rd edn (Cambridge: Deighton, Bell and Co., 1883), pp. 397 9, who described them as ‘ancient’, but did not give any more precise date. On the Old Church Slavonic version of the Canon Tables found in Codex Zographensis, see Leszek Moszynski, ‘Kanony Euzebiusza w głagolskim rękopisie kodeksu Zografskiego’, Slovo 25 6 (1976): pp. 77 119. I have been unable to find any information about Canon Tables in Arabic manuscripts. e.g., no mention is made of their existence in Hikmat Kashouh, The Arabic Versions of the Gospels: The Manuscripts and Their Families (Berlin: De Gruyter, 2012). ¹⁵ Cf. Martin Wallraff, Kodex und Kanon: Das Buch im frühen Christentum (Berlin: De Gruyter, 2013), p. 34, who pointed out that, after the Bible itself, the apparatus is ‘den bestüberlieferten Text der Antike’.

Introduction

Fig. 12. Canon II in Abba Garima I (sixth seventh c.). Ethiopia, Abba Garima Monastery, AG I, fol. 12r (© Michael Gervers, 2004)

15

16

The Eusebian Canon Tables

New Testament, the Nestle-Aland 28th edition, contains the Eusebian apparatus, as has every edition in the series since the seventh, published in the early twentieth century.¹⁶ Given the evident popularity of Eusebius’ apparatus in the centuries after its creation, the paucity of attention it has garnered in modern scholarship is surprising. The only monograph devoted to the topic is an influential tome from 1938 which is limited to an art-historical analysis,¹⁷ and this has since been joined by a handful of articles, again mostly concerned with the artwork that often accompanies the presentation of the tables in late antique and early medieval gospelbooks. The present study aims to fill this lacuna not simply because it is a hole in existing scholarship, but also because the story of the reception of the Eusebian Canon Tables is paradigmatic for the history of late antiquity as a whole. As I will argue in what follows, this innovative reading technology grew out of the long tradition of literary scholarship practised above all in Alexandria, Egypt, and was then transmitted, via Caesarea, to the diverse Christian traditions that emerged from the fourth century onwards throughout the Mediterranean world and beyond. This is, therefore, a story about the transformation of classical antiquity as Hellenic learning was appropriated and put to new uses by Christian scribes and readers in a diversity of languages and cultures—a development that would have been unimaginable to any of the scholars of the famed Alexandrian library and Mouseion. Chapter 1 lays the theoretical guidelines for the work that follows, presenting Eusebius’ system in the light of the history and theory of paratexts and information visualization and arguing that the apparatus is a paratext designed to order the textual content of the fourfold gospel. Chapter 2 examines the immediate origins of the system in a joint scholarly effort, stretching over a century or more, from the Alexandrian Ammonius to Eusebius in Caesarea. The hermeneutical implications of the Canon Tables are explored in chapter 3, where it is argued that Eusebius’ paratextual system created intertextual links across the fourfold gospel and encouraged an open-ended, hypertextual reading of this corpus. Chapter 4 transitions to the reception history of the apparatus, beginning with Augustine of Hippo, who tacitly used Eusebius’ system for the research he carried out to write his highly influential treatise De consensu evangelistarum, where he developed the first ever general theory to account for the literary phenomena presented by the fourfold gospel. Chapter 5 shifts the focus back eastwards, with an analysis of the revision of the system incorporated into the Syriac Peshitta version of the gospels, which ¹⁶ For a discussion of its inclusion in the seventh edition, see E. Nestle, ‘Die Eusebianische Evangeliensynopse’, Neue kirchliche Zeitschrift 19 (1908): pp. 40 51, 93 114, 219 32. The version of Canon Tables printed in subsequent editions, right up to NA²⁸, is unchanged from that of Nestle’s seventh edition, though Prof. Martin Wallraff of the University of Munich is currently preparing a new critical edition to be published by Mohr Siebeck. ¹⁷ Nordenfalk, Die spätantiken Kanontafeln.

Introduction

17

I am calling Canon Tables 2.0—in other words, the new and improved version of the original, in which the paratext is more finely tuned to accord with the text to which it relates. In chapter 6 I highlight little-studied literature produced by Irish or Irish-influenced scholars between the seventh and the ninth centuries, in which Canon Tables figure prominently as a symbolic but also utilitarian means of engaging with the text of the gospels and analysing the relationship between parallel passages. Finally, carrying forward the theme of the symbolism of the Canon Tables, chapter 7 focuses on other little-studied texts, in this instance, Armenian commentaries from the eighth century and later, which explore how the rich decorative motifs that grew up around the Canon Tables (e.g., architectural frames, birds, plants) can be used as the basis for a programme of contemplative meditation understood as a preparatory (and so paratextual) task required before the reading of the fourfold gospel. This is by no means a comprehensive survey of the reception history of the Eusebian apparatus from late antiquity into the Middle Ages. Given how little scholarly attention has been devoted to this topic, too many questions remain unanswered and too many sources unstudied (and even unpublished) to attempt such a project. Instead, the four traditions considered in the latter portion of this book have been chosen because they represent four different modes of using the Canon Tables as a paratext and so illustrate the potential inherent in the Eusebian apparatus for engaging with the fourfold gospel in a variety of ways, from the philological to the theological to the visual.

1 Eusebius’ Canon Tables as a Paratext for Ordering Textual Knowledge Today we are in a better position to analyse and appreciate Eusebius’ Canon Tables system, thanks to recent advances in two other areas of scholarly enquiry: first, the notion of paratexts, which was formulated three decades ago in French literary criticism; and second, the study of information visualization, which has emerged as a discipline in the wake of the revolution brought about by computers and the Internet. The aim of the present chapter is to survey these two areas and glean from them insights that will allow me to give a more precise definition of exactly ‘what’ Eusebius’ Canon Tables are, so that, having defined my object of enquiry, I can, in the subsequent chapters, explore its more immediate context, the function it was designed to perform, and its use by later readers. The argument, in brief, is that Eusebius issued his system of Canon Tables as a component in a new edition of the fourfold gospel, in which his marginal apparatus related to the text of the gospels as a paratext. This paratext both served as the entryway into the text of the gospels and also ordered the disparate textual content contained therein, and in so doing shaped the reader’s interaction with this text (or rather texts). For both paratextuality and tabular presentation, I first examine the topic from a more theoretical perspective and then turn to a historical survey.

CANON TABLES AS A P ARATEXT

Defining Paratextuality The term ‘paratext’ was coined three decades ago by the literary theorist Gérard Genette and it came eventually to refer to one of the five types of textual transcendence he outlined in his poetics.¹ By ‘textual transcendence’, ¹ Note that Genette’s usage of the term underwent development. As he points out, what he referred to as ‘paratextuality’ in the first volume of his trilogy, he would later term

The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0002

22

The Eusebian Canon Tables

Genette meant ‘everything that brings [the text] into relation (manifest or hidden) with other texts’.² These textual relations, according to Genette, may be classified into the following categories, arranged in increasing order of abstraction: 1) intertextuality—quotation of or allusion to a text in another text; 2) paratextuality (defined below); 3) metatextuality—a text commenting on another text; 4) hypertextuality—a text superimposed upon another text, by imitation or other means; 5) and architextuality—texts grouped together as a genre.³ The first volume of Genette’s poetics was devoted to architextuality,⁴ the second to hypertextuality (under the metaphor of the ‘palimpsest’),⁵ and the third to paratextuality.⁶ We will have occasion in chapter three to revisit Genette’s concept of hypertextuality, since this term has relevance for the style of reading the Eusebian apparatus encourages, but in the present chapter it is paratextuality that is of the greatest relevance. As suggested by the prefix para-, paratextuality refers to what is alongside or surrounding a text, including such commonplace features as titles, prefaces, dedications, notes, indices, and so on.⁷ In order to gather together the wide range of practices and discourses encompassed by the term ‘paratexts’, Genette identified a central function of paratextuality, which he highlighted in the original French title of his volume (Seuils): [Paratexts] surround [the text] and extend it, precisely in order to present it, in the usual sense of this verb but also in the strongest sense: to make present, to ensure the text’s presence in the world, its “reception” and consumption in the form (nowadays, at least) of a book . . . the paratext is what enables a text to become a book and to be offered as such to its readers and, more generally, to the public. More than a boundary or a sealed border, the paratext is, rather, a threshold, or . . . a “vestibule” that offers the world at large the possibility of either stepping inside or turning back. It is an “undefined zone” between the inside and the outside, a zone without any hard and fast boundary on either the inward side (turned toward the text) or the outward side (turned toward the world’s discourse ‘hypertextuality’ (Gérard Genette, The Architext: An Introduction, trans. Jane E. Levin (Berkeley, CA: University of California Press, 1992), p. 82 (originally published as Gérard Genette, Introduction à l’architexte (Paris: Editions du Seuil, 1979)); Gérard Genette, Palimpsests: Litera ture in the Second Degree, trans. Channa Newman and Claude Doubinsky (Lincoln, NE: University of Nebraska Press, 1997), 1 (originally published as Gérard Genette, Palimpsestes: la littérature au second degré (Paris: Editions du Seuil, 1982))). ² Genette, The Architext, p. 81. ³ Genette, Palimpsests, pp. 1 7. See also the brief summary of the five categories in Richard Macksey’s foreword to Gérard Genette, Paratexts: Thresholds of Interpretation, trans. Jane E. Lewin, Literature, Culture, Theory 20 (Cambridge: Cambridge University Press, 1997), pp. xviii xix (originally published as Gérard Genette, Seuils (Paris: Editions du Seuil, 1987)). ⁴ Genette, The Architext. ⁵ Genette, Palimpsests. ⁶ Genette, Paratexts. ⁷ Note that both Genette and his later followers point out that the prefix para denotes not simply proximity but also liminality. Cf. Genette, Paratexts, p. 1; Laura Jansen, ‘Introduction: Approaches to Roman Paratextuality’, in The Roman Paratext: Frame, Texts, Readers, ed. Laura Jansen (Cambridge: Cambridge University Press, 2014), p. 5.

Eusebius’ Canon Tables as a Paratext

23

about the text) . . . [The paratext] constitutes a zone between text and off text, a zone not only of transition but also of transaction: a privileged place of a pragmatics and a strategy, of an influence on the public, an influence that whether well or poorly understood and achieved is at the service of a better reception for the text and a more pertinent reading of it.⁸

As suggested by this passage, Genette defines a paratext largely on functional grounds: ‘the paratextual element is always subordinate to “its” text, and this functionality determines the essence of its appeal and its existence’.⁹ In keeping with his description of his work as a synchronic study rather than a ‘history of the paratext’,¹⁰ Genette’s Seuils is an encyclopedic survey that provides a typology of the diverse kinds of paratexts, covering material from antiquity to the present, though focusing primarily on the modern novel. Befitting this focus, he largely limited his exploration of paratexts to that which stood in close proximity to the original production or publication of a text: ‘By definition, something is not a paratext unless the author or one of his associates accepts responsibility for it’.¹¹ According to this definition, much pre-modern material that might otherwise be paratextual would be discounted, since in many cases such features were added to the text at a later point of its transmission. Subsequent studies, however, have expanded Genette’s fairly restrictive view in order to take into account other, non-authorial features that relate to texts in a similar way. For example, a collection of essays published in 2016 emphasizes that not just texts but also ‘their carriers’—that is, manuscripts—have paratexts, and that these paratexts in manuscripts ‘have at least three main functions, namely (1) structuring (e.g., offering navigation aids that guide the reader, such as tables of contents), (2) commenting (e.g., glosses and annotations that offer interpretations and explanations of a text), and (3) documenting’.¹² Focusing on the latter function, the volume demonstrates how paratextual features such as colophons can be used to track manuscripts through space and time. In the fields most relevant to the present study—classics and biblical studies—a number of publications have recently appeared that expand and deepen Genette’s analysis by focusing on paratextual features in specific

⁸ Genette, Paratexts, pp. 1 2. ⁹ Genette, Paratexts, p. 12. Genette subdivided the category of ‘paratexts’ into two further classifications: ‘peritexts’ refers to paratexts that exist within the same material artefact as the text itself, while ‘epitexts’ are more ‘distanced elements’ that are ‘outside the book’, such as interviews or conversations about the text (Genette, Paratexts, pp. 4 5). ¹⁰ Genette, Paratexts, p. 13. ¹¹ Genette, Paratexts, p. 9. See Genette’s further comments about this authorial restriction on pp. 408 10. ¹² Giovanni Ciotti and Hang Lin, eds, Tracing Manuscripts in Time and Space Through Paratexts, Studies in Manuscript Cultures 7 (Berlin: De Gruyter, 2016), VII.

24

The Eusebian Canon Tables

corpora of literature. Several studies on ancient book titles have appeared,¹³ and a 2011 conference at the University of St. Andrews resulted in an edited collection of papers that explore the role of paratexts in a variety of Latin works.¹⁴ In her introduction to the volume, Laura Jansen emphasized the ‘liminal’ nature of paratexts as neither within nor without the text, and argued that this liminal status is what gives paratexts their hermeneutical potency: paratextual analysis ‘suggests itself as a dynamic, indeed multidirectional, approach to both the ways in which a work frames its meanings through the lens of its paratexts and the complexities behind our own interpretive strategy’.¹⁵ In its focus on the role of the paratext as a mediator between text and reader, Jansen’s comment continues Genette’s functional approach to paratextuality. The study of the great mass of biblical paratexts from antiquity may have been somewhat hindered by a Protestant tendency to view all such material as later accretions that had to be removed to recover the original text, the true site of revelation and locus of authority,¹⁶ but they at least have not

¹³ See, e.g., Jean Claude Fredouille et al., eds, Titres et articulations du texte dans les oeuvres antiques (Paris: Institut d’études Augustiniennes, 1997); Bianca Jeanette Schröder, Titel und Text: Zur Entwicklung lateinischer Gedichtüberschriften. Mit Untersuchungen zu lateinischen Buchtiteln, Inhaltsverzeichnissen und anderen Gliederungsmitteln, Untersuchungen zur antiken Literatur und Geschichte 54 (Berlin: Walter de Gruyter, 1999); Menico Caroli, Il titolo iniziale nel rotolo librario greco egizio: con un catalogo delle testimonianze iconografiche greche e di area vesuviana (Bari: Levante, 2007); Francesca Schironi, To Mega Biblion: Book Ends, End Titles, and Coronides in Papyri with Hexametric Poetry, American Studies in Papyrology 48 (Durham, NC: American Society of Papyrologists, 2010); Gianluca del Mastro, Titoli e annotazioni bibliologiche nei papiri greci di Ercolano (Naples: Centro internazionale per lo studio dei papiri ercolanensi, 2014). ¹⁴ Laura Jansen, ed., The Roman Paratext: Frame, Texts, Readers (Cambridge: Cambridge University Press, 2014). See also Andrew M. Riggsby, ‘Guides to the Wor(l)d’, in Ordering Knowledge in the Roman Empire, ed. Jason König and Tim Whitmarsh (Cambridge: Cambridge University Press, 2007), pp. 88 107; C. T. Mallan, ‘The Book Indices in the Manuscripts of Cassius Dio’, The Classical Quarterly 66 (2017), pp. 705 23; Aaron Pelttari, ‘Speaking From the Margins: Paratexts in Greek and Latin Poetry’, in Walking the Wire: Latin and Greek Poetry in Dialogue, ed. Tine Scheijnen and Berenice Verhelst (Cambridge: Cambridge University Press, forthcoming). ¹⁵ Jansen, ‘Introduction’, p. 2. ¹⁶ So Thomas O’Loughlin in ‘De Bruyne’s Sommaires on its Centenary: Has its Value for Biblical Scholars Increased?’, in Donatien De Bruyne, Pierre Maurice Bogaert, and Thomas O’Loughlin, Summaries, Divisions and Rubrics of the Latin Bible, Studia Traditionis Theologiae 18 (Turnhout: Brepols, 2015), pp. xix xxvi. According to O’Loughlin, the situation was different for Roman Catholic scholars committed to the Vulgate, for whom ‘the sacred text, for all its problems, did not come alone but was embedded within a web of other material: various ways of gathering books together, lists of chapter headings, a variety of division systems, cross referencing systems, along with aids to readers which, starting with the work of Eusebius of Caesarea, seemed to have been added to by every generation until the time of printing’ (p. xx). This focus on such later material was thought to be appropriate because ‘the tradition was part of the work of the Spirit speaking in the Church and it was to be respected as such . . . By culture, training, and temperament the Vulgate editors were inclined to value everything they found in a codex: it was, in its totality, part of the tradition’ (p. xxi). This impulse, however, is not restricted to modern scholarly treatments of biblical texts. Shane Butler has argued that ‘the most

Eusebius’ Canon Tables as a Paratext

25

been completely ignored. A century ago the great New Testament scholar Hermann von Soden provided some analysis of these paratexts in Greek,¹⁷ and shortly thereafter Donatien De Bruyne published collections of paratexts from the Latin biblical tradition,¹⁸ but further study of them was scarce throughout the twentieth century. The trend has obviously changed in the past two decades, no doubt partly under inspiration from Genette, as evidenced in a recent spate of studies on biblical paratexts from late antiquity into the Middle Ages.¹⁹ Although Genette’s notion of paratextuality provides an obvious avenue into the investigation of Canon Tables, thus far Eusebius’ apparatus has not been explored from the angle of paratextual theory. If one follows Genette’s most basic definition of what constitutes a ‘paratext’, Canon Tables clearly fall into this category, since they are a text that exists entirely in the service of another text, in this case the fourfold gospel. A set of Canon Tables on its own would be useless, since the ten numeric tables are indecipherable in the absence of a corresponding edition of the text of the four gospels. In an imaginary world in which the prefatory tables had survived to the present day but no copy of the gospels with the sectional enumeration had been preserved, scholars would be hard pressed to make much sense of them. A useful case in point is the fragmentary set of Canon Tables held in the British Library, commonly known as the London Canon Tables (Add. MS 5111/1; see fig. 13).

important (and illuminating) reason for the oversight [of the capita for Cicero’s works] surely has been the view, as old as the Renaissance, of the work of philology as that of rescuing a classical logos from the encumbrances and corruptions of its medieval flesh, the former being the “text,” seen as a sequence of letters and words, and the latter necessarily including everything else, including, as fate would have it, capitulation’ (Shane Butler, ‘Cicero’s Capita’, in The Roman Paratext: Frame, Texts, Readers, ed. Laura Jansen (Cambridge: Cambridge University Press, 2014), p. 97). ¹⁷ See the section titled ‘D. Einteilungen des Textes der Schriften des NTs’ in Hermann von Soden, Die Schriften des Neuen Testaments in ihrer ältesten erreichbaren Textgestalt (Berlin: Verlag von Alexander Duncker, 1902 13), pp. 388 485. ¹⁸ Donatien De Bruyne, Sommaires, divisions et rubriques de la Bible latine (Namur: Godenne, 1914); Donatien De Bruyne, Préfaces de la Bible latine (Namur: A. Godenne, 1920). ¹⁹ See, e.g., A. A. den. Hollander, Ulrich Schmid, and Willem F. Smelik, eds, Paratext and Megatext as Channels of Jewish and Christian Traditions: The Textual Markers of Contextual ization, Jewish and Christian Perspectives Series 6 (Leiden: Brill, 2003); Louis Charles Willard, A Critical Study of the Euthalian Apparatus, Arbeiten zur neutestamentlichen Textforschung 41 (Berlin: Walter de Gruyter, 2009); Philip S. Alexander, Armin Lange, and Renate Pillinger, In the Second Degree: Paratextual Literature in Ancient Near Eastern and Ancient Mediterranean Culture and Its Reflections in Medieval Literature (Leiden: Brill, 2010); Simon J. Gathercole, ‘The Titles of the Gospels in the Earliest New Testament Manuscripts’, Zeitschrift für die Neutestamentliche Wissenschaft 104 (2013): pp. 33 76; Eric W. Scherbenske, Canonizing Paul: Ancient Editorial Practice and the Corpus Paulinum (New York: Oxford University Press, 2013); Martin Wallraff and Patrick Andrist, ‘Paratexts of the Bible: A New Research Project on Greek Textual Transmission’, Early Christianity 6 (2015): pp. 237 43; Vemund Blomkvist, Euthalian Traditions: Text, Translation and Commentary, TU 170 (Berlin: De Gruyter, 2012). Though not on biblical texts per se, see also David Lincicum, ‘The Paratextual Invention of the Term “Apostolic Fathers” ’, Journal of Theological Studies 66 (2015): pp. 139 48.

26

The Eusebian Canon Tables

Fig. 13. Canon VIII, IX, and XMt of the London Canon Tables (sixth seventh c.). Note that Canon VIII (the two left hand columns) should contain numbers for Luke and Mark, but the header row here reads Matthew and Mark. London, British Library, Add. 5111/1, fol. 11r (© The British Library Board)

These two partially preserved folios from the sixth or seventh century, of likely Constantinopolitan origin, survived solely because they were attached to a much later manuscript, and no trace of the original text to which they belonged has come to light. What makes them noteworthy is that the numbers contained in them diverge widely from the standard set, leading Carl

Eusebius’ Canon Tables as a Paratext

27

Nordenfalk to hypothesize that they represent a later revision of Eusebius’ original system.²⁰ Nordenfalk’s hypothesis might be correct, but unfortunately we have no way to confirm it because we do not have the text to which the numbers refer. Simply put, the Canon Tables take on meaning only when they are viewed in relation to the text for whose service they exist, and as such they are well suited to paratextual analysis. In fact, the present study of the origin and reception of the Eusebian apparatus has something to offer to the wider sub-discipline of paratextual studies, since few paratexts in the history of western literature have proven as successful as Eusebius’ invention. It is possible to study classes of paratexts as they developed over time—the history of the footnote would be one example²¹—but few individual paratexts were transmitted over a long enough period or a wide enough territory to provide sufficient material in their own right to write a book-length history. Of course the reason why we have such material at our disposal is that Eusebius’ paratext was attached to the most widely transmitted text during the centuries in question, the four gospels. As a result, when analysing Canon Tables, one can take a diachronic approach, revealing the developments in a single paratext over time and thereby complementing Genette’s synchronic study. Moreover, among those paratexts studied by Genette and others, Eusebius’ apparatus stands out as a particularly complex example, since it is a system with multiple interlinking parts each of which is necessary for the others to function. The marginal enumeration within each gospel only works in coordination with the preceding tables, and, as Eusebius realized, the system is novel and complex enough that a prefatory set of instructions was required, which he provided in the form of his Letter to Carpianus. Studying the origin and reception of the Canon Tables therefore promises to enrich our understanding of the potential of paratexts to develop over time, as later users find in them possibilities not envisioned by their creators. So Genette provides us with an analytic category in which to place Eusebius’ apparatus, one that helpfully highlights its most fundamental defining feature, namely its dependence upon the fourfold gospel. In one further respect Genette’s work can provide a methodological orientation for the present study. Recall that he approached the study of paratexts from a functional perspective, focusing upon the role for which their authors intended them and the influence they exerted, or at least sought to exert, upon readers. It is precisely such a focus that has been almost completely absent in the few studies of the Eusebian apparatus that have thus far been undertaken. Carl Nordenfalk’s seminal book from eighty years ago is an outstanding work of scholarship

²⁰ Nordenfalk, Die spätantiken Kanontafeln pp. 127 46. ²¹ Anthony Grafton, The Footnote: A Curious History, rev. ed. (Cambridge, MA: Harvard University Press, 1997).

28

The Eusebian Canon Tables

that answered many of the fundamental questions about Eusebius’ original design, but it remains true that he was only really interested in Canon Tables for their relevance to art history. Yet Canon Tables, as much as or even more than other forms of paratexts, perform a distinct function with respect to the texts to which they are attached and have influenced the reception of those texts over the course of time. Of the specific functions of paratexts already mentioned, one that is clearly at work in the Eusebian apparatus is the structuring of the reader’s experience of the text of the gospels: Canon Tables provide multiple access points to this corpus of literature and bind together otherwise disparate elements into a contextual web of meaning.²² Moreover, Laura Jansen’s argument about Latin literature is equally true for Eusebius’ paratext. She has pointed out that paratexts are ‘sites of reception where readers or viewers are prompted to (re)negotiate trajectories of plotting meaning or (re)consider their own construction as audiences’.²³ To use the Genettian metaphor, by passing through the threshold of the paratext one emerges on the other side as a particular kind of reader of this particular text. So too the Canon Tables exert a hermeneutical pressure upon their users, forcing readers (and indeed viewers) to consider their own role in relation to the text and the wider imaginative world within which the text is situated. I will elaborate on each of these claims later in the book, but enough has been said at present by way of orientation.

The Development of Paratexts in Greco-Roman Book Culture up to the Fourth Century CE The creativity of Eusebius’ achievement with his Canon Tables becomes more apparent when they are viewed against the broader history of paratexts from antiquity. Although paratexts were not completely absent from manuscripts of the Hellenistic and imperial periods, they were decidedly more restrained than what is found in a modern printed book, usually appearing only at the beginning or end of a text. In his recent study of the papyri from Herculaneum, George W. Houston points out that many of the bookrolls recovered from the Villa of Papyri included paratextual material, but that this usually consisted of ‘an end title written by the scribe who copied the text and providing the name of the author, the title of the work, and, where appropriate, the number of the book’.²⁴ In the 30 per cent of cases in which the ²² Similarly, Andrew M. Riggsby argues that tables of contents perform two functions: reference and segmentation (Riggsby, ‘Guides to the Wor(l)d’, p. 89). ²³ Jansen, ‘Introduction’, p. 7. ²⁴ George W. Houston, Inside Roman Libraries: Book Collections and Their Management in Antiquity (Chapel Hill, NC: The University of North Carolina Press, 2014), pp. 111 20, at 111 12. The limited navigational and finding aids provided in ancient bookrolls is also

Eusebius’ Canon Tables as a Paratext

29

subscriptiones go beyond this basic content, the additional information often consists of a subtitle to the individual book in the roll or stichometric counts. The relative simplicity of such paratextual material is partly because more complex devices simply had not yet been invented, but it is also has to do with certain cultural factors specific to the ancient world. William A. Johnson has called attention to the ‘radically unencumbered stream of letters’ evident in the bookroll, which lacked not only the sort of subheadings that are standard in books today but also spacing between words, and argued that this design is intrinsic to its status as an elite cultural object.²⁵ Because paratextual features obviously disrupt this ‘unencumbered stream of letters’, the reigning aesthetic might have been something of a hindrance to the development of paratextual features, at least within high-status books. Andrew M. Riggsby has offered a slightly different but nevertheless concordant hypothesis to explain the scarcity of paratextual features specifically within Latin texts: ‘Even in their most literary moments, Romans preferred imagining texts (at least potentially) as speech acts. This makes many informational devices (tables of contents, section numeration, tables, illustrations) problematic, insofar as they are inherently paratextual’.²⁶ The reason for the ‘problematic’ status of such paratextual features is at least twofold, according to Riggsby. First, imagining text as speech means that paratexts are even more ephemeral in relation to text than they otherwise would be, and, secondly, many forms of paratexts create discontinuous text, which disrupts the imagined continuous oral performance.²⁷ In keeping with the paucity and simplicity of paratextual features in manuscripts from the centuries prior to Eusebius, there was, so far as we know, no analogous system of numeric cross-referencing designed for any other corpus of literature in the Greco-Roman world. That is, Eusebius did not simply adapt for the gospels a technique that had been previously developed for emphasized in Jocelyn Penny Small, Wax Tablets of the Mind: Cognitive Studies of Memory and Literacy in Classical Antiquity (New York: Routledge, 1997), pp. 11 25, 53 71; Christian Vandendorpe, From Papyrus to Hypertext: Toward the Universal Digital Library, trans. Phyllis Aronoff and Howard Scott, Topics in the Digital Humanities (Urbana, IL: University of Illinois Press, 2009), pp. 123 4. ²⁵ William A. Johnson, Readers and Reading Culture in the High Roman Empire: A Study of Elite Communities, Classical Culture and Society (New York: Oxford University Press, 2010), 20. See also Johnson’s comments about the ‘aesthetic’ of the bookroll and how it differs from that of the codex in William A. Johnson, Bookrolls and Scribes in Oxyrhynchus (Toronto: University of Toronto Press, 2004), pp. 85 6. ²⁶ Andrew M. Riggsby, Mosaics of Knowledge: Representing Information in the Roman World, Classical Culture and Society (Oxford: Oxford University Press, 2019), see the section ‘A Brief Orientation’. ²⁷ On the latter point Riggsby noted the parallel argument with respect to Greek literary works made by Reviel Netz, ‘Authorial Presence in the Ancient Exact Sciences’, in Writing Science: Medical and Mathematical Authorship in Ancient Greece, ed. Markus Asper, Science, Technology, and Medicine in Ancient Cultures 1 (Berlin: De Gruyter, 2013), pp. 237 40.

30

The Eusebian Canon Tables

the study of, say, Homer, but instead invented a new paratextual device fit for the purpose of reading the fourfold canon.²⁸ However, in one respect there was a paratextual precedent for Eusebius’ invention, namely the use of book titles and subtitles for textual division and demarcation, which, as just noted, represents one of the few paratexts that do regularly appear in manuscripts from the imperial period. Although the exact origin of this practice is unclear, the broad outlines of its emergence can be stated. As has been demonstrated in an illuminating article by Carolyn Higbie, the manner in which ancient authors referred to other texts underwent development through the Hellenistic and Roman periods. In the oldest form, the usual way to cite a passage from a given author was ‘by their name, by the name of their works, if they had written more than one, or by identifying a scene or section of a work’.²⁹ The references to Homer in Herodotus’ Histories 2.116 are a representative example: The priests said that such was the arrival of Helen at Proteus’. Homer seems to me also to have known this story, but it was not as suited for epic as the one which he used, so he chose not to include it, while showing that he knew the story. This is clear, for he describes in the Iliad (and at no other point does he retrace his steps) Alexander’s wanderings, how, as he was leading Helen away, he was carried off course, how he wandered here and there and how he came to Sidon in Phoenicia. He makes mention of this in the aristeia of Diomedes. The verses run as follows [Il. 6.289 92]. He also makes mention of it in the Odyssey in these verses [Od. 4.227 30] and Menelaos says the following to Telemachus [Od. 4.351 2]. In these verses he shows that he knew about the wanderings of Alexander to Egypt.³⁰

If he had been able to avail himself of book and line numbering, Herodotus could have made his point with greater economy and precision. However, his text of the Iliad seemingly was not even divided into books, so he had to describe a scene and then cite the relevant lines himself. Even with respect to crossreferences within his own work, Herodotus obviously lacks a system of precise reference but instead can only point readers to the λόγος or λόγοι on a given topic, without any enumeration or other navigational instructions.³¹ These are clearly ad hoc references that should not be understood as traces of a larger organizational structure provided to the reader as an overall navigation system. ²⁸ There is, therefore, no evidence for the claim that ‘Eusebius may have derived the idea [of the Canon Tables] from pagan scholarship’ (Timothy D. Barnes, Constantine and Eusebius (Cambridge, MA: Harvard University Press, 1981), p. 122). Barnes himself admitted, in his endnote to this sentence, that ‘no model can be identified’ (p. 344 n. 122). Cf. Michael J. Hollerich, ‘Eusebius’, in The New Cambridge History of the Bible. Volume 1: From the Beginnings to 600, ed. James Carleton Paget and Joachim Schaper (Cambridge: Cambridge University Press, 2013), pp. 637 8, who noted that the Canon Tables represent ‘something genuinely new and without demonstrated pagan precedent’. ²⁹ Carolyn Higbie, ‘Divide and Edit: A Brief History of Book Divisions’, Harvard Studies in Classical Philology 105 (2010): pp. 3 4. ³⁰ Herodotus 2.116, cited in Higbie, ‘Divide and Edit’, p. 4. ³¹ See 2.161.3 and 5.36.4, cited and discussed in Higbie, ‘Divide and Edit’, pp. 5 6.

Eusebius’ Canon Tables as a Paratext

31

In Thucydides’ two references to the Iliad (1.9.4; 1.10.4), he, like Herodotus, refers to portions of the text by describing the relevant scene rather than citing a book number.³² In fact, the earliest surviving citation of either Homer or Herodotus by book number dates to centuries later and comes from an inscription rather than a literary source. Written in 99 BCE, the Lindian Chronicle, which exhibits an unusual concern for documenting its claims about the importance of the town of Lindos on Rhodes, referred to a certain gift from the Pharaoh Amasis, ‘about which Herodotus the Thurian testifies in the second book of his Investigations’ (περὶ οὗ μ[αρτ]υρεῖ Ἡρόδοτος [ὁ Θ]ούριος ἐν τᾶι Β τᾶν ἱστο[ρι]ᾶ[ν . . . ).³³ Eight citations of other authors follow this reference to Herodotus, and all of them refer to the author’s name, the title of his work, and the relevant book number. The first reference to Homer by book number is a first-century CE papyrus (P. Mil. Vogl. 1 19) that says the second-century BCE scholar Apollodorus of Athens wrote a commentary on book 14 of the Iliad,³⁴ implying that by the time Apollodorus was writing, the text of the Iliad had been broken up into the numbered books that are familiar to us today. The first author who we can confidently say used book division as a compositional principle was the historian Ephorus (4th c. BCE), and in this he was followed by his fellow historian Polybius (2nd c. BCE).³⁵ From this point onward, book division became increasingly common in philosophical, literary, mathematical, and medical works, and accordingly authors began citing other works by book number.³⁶ Cicero and Quintilian are prominent examples among Latin authors who structured the content of their works by book, but the author who made the fullest use of this device was Pliny the Elder, who devoted the entirety of the first book of his thirty-six-book Natural History to a table of contents for the work that follows.³⁷ His adopted nephew, Pliny the Younger, may have been deliberately following his uncle when he prepared his edition of his letters, for he too began each of the nine books of his collection

³² Cf. Higbie, ‘Divide and Edit’, pp. 6 7. ³³ Cited in Higbie, ‘Divide and Edit’, 9. ³⁴ Achille Vogliano, Papiri della R. Università di Milano, Vol. 1 (Milan: Hoepli, 1937), pp. 174 5. Cf. Higbie, ‘Divide and Edit’, p. 10. Vogliano dated the papyrus to the second century CE, but the Leuven Database of Ancient Books now places it between 1 and 75 CE (LDAB ID: 242). ³⁵ Cf. Higbie, ‘Divide and Edit’, pp. 16 19. See also the brief overview of tables of contents in antiquity in Günther Zuntz, The Ancestry of the Harklean New Testament, British Academy Supplemental Papers 7 (London: Oxford University Press, 1945), pp. 80 2. ³⁶ Cf. Higbie, ‘Divide and Edit’, pp. 20 4. William A. Johnson has pointed out that for longer prose works divided into ‘books’, we should ‘usually’ assume that this structure initially indicated a physical division of the work into ‘separate short papyrus rolls’ (Johnson, Bookrolls and Scribes, pp. 143 52, at 145). Though the equation ‘one book equals one bookroll’ does not always hold, it often does, which is a helpful reminder of the physical constraints that the book form imposes upon a text. ³⁷ Cf. Schröder, Titel und Text, 50 6; Riggsby, ‘Guides to the Wor(l)d’, pp. 93 8; Higbie, ‘Divide and Edit’, pp. 26 7.

32

The Eusebian Canon Tables

with a table of contents listing the addressees and incipits for each letter.³⁸ By the third century CE, citation by book number was more frequent, as seen, for example, in Athenaeus, who ‘cites author after author, not just Homer and Herodotus, by book number’.³⁹ This shift in citation practice is no doubt in part due to shifting social conditions, with the emergence of ‘an increasingly literate world, in which writers consult texts and provide citations for their readers’.⁴⁰ The development of a distinctly Christian book culture over the first four centuries of the common era both mirrored and probably contributed to this shift in citational practices and book design. It goes without saying that none of the authors of the books that became the New Testament structured their works by demarcating the text, but it did not take long for later editors to retrofit these books with various paratextual features. The first biblical paratexts that we can confidently say existed are those that Marcion devised in the mid-second century for his edition of the corpus Paulinum, including his Antitheses as a likely prologue to the whole, and brief argumenta to introduce individual letters.⁴¹ Although the precise dating is uncertain, standard chapter divisions eventually emerged for the books of the New Testament, known in Greek as κεϕάλαια and in Latin as capitula.⁴² The fourth century seems to have been a period of particularly intense experimentation with this genre. In addition to Eusebius’ Canon Tables, to this century also belong the earliest components of the so-called Euthalian apparatus, the most comprehensive attempt to supply all of the New Testament texts, apart from the gospels, with paratextual content in Greek. The Euthalian edition included for each book (Acts, catholic epistles, and Pauline epistles) a prologue, a hypothesis, a list of κεϕάλαια, a list of citations, and various other biographical material about Paul especially.⁴³ In the late fourth century, the Latin bishop Priscillian and his followers seem to have been similarly devoted to developing paratexts for biblical literature. Inspired by Eusebius, they devised a collection of Canons for the letters of Paul, along with prologues to the gospels that were widely transmitted in the later Vulgate tradition.⁴⁴ ³⁸ Roy Gibson, ‘Starting With the Index in Pliny’, in The Roman Paratext: Frame, Texts, Readers, ed. Laura Jansen (Cambridge: Cambridge University Press, 2014), pp. 33 55. See further Riggsby, Mosaics of Knowledge, Chapter 1. ³⁹ Higbie, ‘Divide and Edit’, p. 27. ⁴⁰ Higbie, ‘Divide and Edit’, p. 27. ⁴¹ Scherbenske, Canonizing Paul, chapter 2. ⁴² Cf. von Soden, Die Schriften des Neuen Testaments, pp. 402 32; Harvey K. McArthur, ‘The Earliest Divisions of the Gospels’, in Studia Evangelica, Vol. III, ed. F. L. Cross, TU 88 (Berlin: Akademie Verlag, 1964); Christian B. Amphoux, ‘La division du texte grec des Évangiles dans l’Antiquité’, in Titres et articulations du texte dans les oeuvres antiques, ed. Jean Claude Fredouille et al., Collection des études augustiniennes, Série Antiquité 152 (Paris: Institut d’Études Augusti niennes, 1997), p. 311; James R. Edwards, ‘The Hermeneutical Significance of Chapter Divisions in Ancient Gospel Manuscripts’, New Testament Studies 56 (2010): pp. 413 26. ⁴³ Cf. Willard, A Critical Study of the Euthalian Apparatus; Scherbenske, Canonizing Paul, chapter 3. ⁴⁴ Although they were long known as the Prologues of the Monarchians, the scholarly consensus for the past century has been that the four anonymous prefaces to the gospels that

Eusebius’ Canon Tables as a Paratext

33

Hence, Eusebius’ invention of the Canon Tables is in keeping with the increasing use of paratextual material in both Christian and non-Christian texts in the third and fourth centuries. Certainly by his day it was commonplace for authors to structure a large prose text by dividing it into constituent parts, a practice that readers had probably come to expect. Eusebius himself, of course, divided his own works into numbered βίβλοι, as can be seen in the ten volumes of his Ecclesiastical History, each of which is subdivided into smaller sections that are also enumerated. Moreover, Eusebius also devised at least one other paratextual aid for biblical literature, which will be discussed at greater length in the next chapter. Yet even though the title and subtitle as structuring paratexts predate Eusebius, his Canon Tables apparatus goes far beyond any precedent in its degree of granularity. Some of his sections of text are quite large, as much as a chapter in today’s reckoning, but most of them are smaller than a paragraph and some are only sentence fragments. In other words, Eusebius’ paratext structured the text of the fourfold gospel with a much greater degree of precision than had been attempted for any prior work of literature. Moreover, owing to its creation of cross-references within the fourgospel collection, the Eusebian apparatus established for this corpus of literature not only the kind of serial progression characteristic of titles and subtitles but also a highly reticulated internal structure that was entirely without precedent. This cross-referencing function was accomplished through the use of a tabular matrix, and to continue the rough sketch of the functionality of the Canon Tables this first chapter is aiming to provide, it is now necessary to turn to the theory of tabular presentation and its use in antiquity.

CANON TABLES AND TABULAR PRESENTATION

Tables as Information Visualization The second context within which to situate Eusebius’ Canon Tables is the theory and history of tabular presentation. The information revolution brought about by the computer and Internet have led to the emergence of a new discipline known as visualization, which considers, among other things, the way in appear in over a hundred Vulgate manuscripts are in fact to be traced back, if not to Priscillian himself, at least to his later followers. Cf. John Chapman, Notes on the Early History of the Vulgate Gospels (Oxford: Clarendon Press, 1908), pp. 217 88; Henry Chadwick, Priscillian of Avila: The Occult and the Charismatic in the Early Church (Oxford: Clarendon Press, 1976), pp. 102 9; with text and translation of the Prisc. prol. in Conti, Priscillian of Avila, pp. 250 7, 314 18. On the can. Paul., see T. J. Lang and Matthew R. Crawford, ‘The Origins of Pauline Theology: Paratexts and Priscillian of Avila’s Canons on the Letters of the Apostle Paul ’, New Testament Studies 63 (2017): pp. 125 45, with text and translation in Conti, Priscillian of Avila, pp. 164 209.

34

The Eusebian Canon Tables

which information or knowledge is presented in visual form.⁴⁵ Although most of the work done in this field is, as one would expect, focused on contemporary problems and applications, it has also sparked an interest in the history of information visualization.⁴⁶ Francis T. Marchese has briefly noted Canon Tables as an important milestone in this history,⁴⁷ and it is hoped that the fuller exploration of Eusebius’ paratext in the present volume will contribute tangentially to the study of information visualization in its own right. Visualizing some set of information is typically undertaken in order to further the viewer’s understanding of the content being represented, an exercise that works on the insight that ‘humans are very good at seeing things’.⁴⁸ More specifically, schematic diagrams like tables have at least three broad advantages: they 1) ‘simplify complex situations’; 2) ‘make abstract concepts more concrete’; and 3) substitute easier perceptual inferences for more computationally intensive search processes and sentential deductive inferences’.⁴⁹ In other words, when well designed, a visual display of information has the potential to reveal patterns or concepts that might otherwise take much longer to intuit, or even remain completely hidden. Definitional clarity on what we mean by ‘tables’ will help to highlight the functionality and purpose of Eusebius’ information visualization. At a minimum, a table consists of ‘a compactly organized, gridded structure’,⁵⁰ composed of intersecting rows and columns, which is sometimes also called a ‘matrix’.⁵¹ This minimal definition is a purely descriptive statement about the presentation of some information in tabular form. However, following the work of Andrew M. Riggsby on tables in the Roman world, I want to insist on a more restrictive definition—it is the fact that meaningful information is represented along both a horizontal and a vertical axis that distinguishes a table from a mere list. That is to say, information may be represented in tabular or gridlike form, but this alone is insufficient to consider it a table in the strict sense. An example might be a list of names in which first names, middle names, and ⁴⁵ For an orientation, see Francis T. Marchese and Ebad Banissi, eds, Knowledge Visualization Currents: From Text to Art to Culture (London: Springer, 2013); Ebad Banissi et al., eds, Information Visualisation: Techniques, Usability and Evaluation (Newcastle upon Tyne: Cam bridge Scholars Publishing, 2014). According to the preface in Information Visualisation, p. vii, the field of visualization was first proposed in the late 1980s in a report commissioned by the National Science Foundation in the United States. It has since been divided into three subfields: scientific visualization, information visualization, and knowledge visualization. ⁴⁶ Cf. Francis T. Marchese, ‘The Beginnings of Medieval Information Visualization’, in Information Visualisation: Techniques, Usability and Evaluation (Newcastle upon Tyne: Cam bridge Scholars Publishing, 2014), pp. 2 23. ⁴⁷ Francis T. Marchese, ‘Tables and Early Information Visualization’, in Knowledge Visual ization Currents: From Text to Art to Culture, ed. Francis T. Marchese and Ebad Banissi (London: Springer Verlag, 2013), pp. 41 5. ⁴⁸ Howard Wainer, ‘Understanding Graphs and Tables’, Educational Researcher 21.1 (1992): p. 15. ⁴⁹ Laura R. Novick and Sean M. Hurley, ‘To Matrix, Network, or Hierarchy: That is the Question’, Cognitive Psychology 42 (2001): p. 159. ⁵⁰ Marchese, ‘Tables and Early Information Visualization’, p. 35. ⁵¹ Novick and Hurley, ‘To Matrix, Network, or Hierarchy’.

Eusebius’ Canon Tables as a Paratext

35

surnames are listed in descending columns, with one full name per row. This tabular format of presentation might make it easier to read the list of names, but it is in fact merely a matter of display rather than function, since the matrixstructure contributes nothing to the meaning of the information. As a test of whether or not a list is truly a table, one can ask the question whether it would be possible to label each row and column to designate the information contained therein.⁵² In the example just given, one could provide a header for each column (‘First Name’, ‘Middle Name’, ‘Surname’) but it is not obvious what one might label the rows since the data they contain exhaust their meaning.⁵³ According to these definitional constraints, the first nine of Eusebius’ thirteen prefatory Canons are clearly tables, in that they consist of a gridded structure in which each box denotes that the number contained within it belongs both to the class of information represented by the vertical column and the class of information represented by the horizontal row. Moreover, Eusebius’ tables satisfy the test just adumbrated: Eusebius himself provided the columns with headers— most likely abbreviated forms of the evangelists’ names—and, although he did not provide labels for the rows, one could easily do so. Alongside the numbers representing the passages from whichever gospels are relevant, it would be sensible to provide a title describing the passages in question. For example, Eusebius’ parallel Mt §147, Mk §64, Lk §93, Jn §49 could be represented as:

Pericope Feeding of the 5,000

Mt 147

Mk 64

Lk 93

Jn 49

In fact, at least two later users of Eusebius’ apparatus undertook precisely this project. Several fragments of a sixth- or seventh-century papyrus gospel book were recovered in the early twentieth century from the monastery of Epiphanius at Thebes, and they include portions of Eusebius’ tables (P. Mon. Epiph. 584). Unexpectedly, alongside some of the numbers in the tables are brief marginal descriptions of the content of the adjacent sections (see fig. 14).⁵⁴ For example, in the margin next to the Canon V parallel containing Mt §3 (Mt 1:18) and Lk §2 (Lk 1:35) is written ‘On the Birth of the Lord’ (ΕΙϹ Τ[ΗΝ] ΓΕΝ[ΗϹΙΝ] ΤΟΥ [ΚΥ]). ⁵² Cf. Riggsby, Mosaics of Knowledge, chapter 2: ‘I want to look not just at the shape of objects, but whether they are actually being used to take advantage of at least some of the things tables have been found to be good for, especially the cross cutting of categories. . . . Formally, this means tables should not only have visible rows and columns, but rows and columns that themselves convey information. They should be equipped, or at least equippable, with row and column headers explaining what is to be found in each’. ⁵³ This example is adapted from Riggsby’s analysis of a list of the names of the sponsors of games at Caere in Etruria. ⁵⁴ See black and white images and analysis in Nordenfalk, ‘Canon Tables on Papyrus’, Dumbarton Oaks Papers 36 (1982): pp. 29 38. The editio princeps is W. E. Crum and H. G. Evelyn White, The Monastery of Epiphanius at Thebes: Part II (New York: Metropolitan Museum of Art, 1926), pp. 122 3, 302 5.

36

The Eusebian Canon Tables

Fig. 14. Fragment of Greek Canon Tables from a papyrus codex found amidst the ruins of a monastery in Thebes (sixth seventh c.). At the top of the large fragment in the centre, to the left of the column of numbers are brief statements summarizing the content of certain sections. New York, Metropolitan Museum of Art, accession no. X.455, P. Mon. Epiph. 584

This, however, was by no means a systematic treatment, and it seems to have been limited to a few passages that were to be read at certain times in the ecclesiastical year.⁵⁵ A more thorough attempt was made in Codex Brixianus, a sixth-century purple manuscript, in which each parallel in all ten Canons was accompanied by the opening words of the passage, an addition that expanded the entire sequence to over 150 pages, of which seventy still survive (see fig. 15).⁵⁶

⁵⁵ Nordenfalk, ‘Canon Tables on Papyrus’, pp. 36 7. ⁵⁶ Nordenfalk, Die spätantiken Kanontafeln, p. 269; Nordenfalk, ‘Canon Tables on Papyrus’, p. 37. On Codex Brixianus, see also H. A. G. Houghton, The Latin New Testament: A Guide to Its Early History, Texts, and Manuscripts (Oxford: Oxford University Press, 2016), pp. 53, 216, who pointed out that the opening pages with the Canon Tables, which have an Old Latin affinity, seemingly have a different origin from that of the rest of the manuscript. Two later Latin manuscripts from the eighth and ninth centuries (Poitiers, Médiathèque François Mitterrand, MS 17 (65); Laon, Bibliothèque municipale, 473 bis) also have expanded sets of Canon Tables with incipits, which seem to be related to those in Brixianus (Houghton, The Latin New Testament, pp. 229, 231). Cf. P. Minard, ‘Témoins inédits de la vieille version latine des Évangiles. Les canons à initia des évangéliaires de Sainte Croix de Poitiers et de la Trinité de Vendôme’, Revue Bénédictine 56 (1945): pp. 58 92.

Eusebius’ Canon Tables as a Paratext

37

Fig. 15. Canon IX in Codex Brixianus (sixth c.). On the left side of the page are section numbers for passages in John and Luke, and on the right side are incipits for each of the parallels. The parchment has been dyed purple and the writing done with silver and gold ink. The incipits listed have an Old Latin text type, unlike the rest of the manuscript, which mainly represents Jerome’s Vulgate. Brescia, Biblioteca civica Queriniana, ManoscrittoPurpureo, fol. 11v

38

The Eusebian Canon Tables

The not insignificant expansion required for this revised version of the tables decreases the overall efficiency and utility of the apparatus, which is surely why it was not attempted more often. Still, for our purposes, it illustrates that Eusebius’ first nine tables satisfy the conditions for a tabular matrix, in which both the horizontal and vertical axes encode meaningful information and are navigable by the user. The final four tables, together comprising Eusebius’ Canon X, are not, however, tables, but more properly lists, since each records passages contained within only a single gospel and so consists of merely a single column of numbers. Nevertheless, throughout this study I will refer to Eusebius’ ten ‘tables’ for the sake of simplicity, though with a tacit acknowledgement of this difference between the first nine and the final one. The reason for insisting on this definitional precision is that it brings into focus the message communicated by the visual format in which Eusebius presented his paratext, that is, the information that his tabular design made apparent that would otherwise have been more difficult to apprehend. The specific kind of information that tables are particularly well suited to display is the relationship that exists between certain classes of things. In the words of Laura R. Novick, a table serves to store static information about the kind of relation that exists between pairs of items in different sets. The cells provide a location where the relational information can be stored, and the row and column indices provide the means for accessing this information when it is needed.⁵⁷

Similarly, Francis T. Marchese asserted that the ‘table’s strength as a visualization medium’ is due to the fact that it has ‘a format that promotes associations among diverse data elements, and facilitates exploration of relationships among them’.⁵⁸ Once a set of data has been visually rendered in a tabular form, it is open to a variety of uses. Novick and Marchese have already highlighted two possibilities—the storage and exploration of data—and at least two others have also been listed in the literature: the communication of data to an audience and decoration.⁵⁹

⁵⁷ Laura R. Novick, ‘Understanding Spatial Diagram Structure: An Analysis of Hierarchies, Matrices, and Networks’, The Quarterly Journal of Experimental Psychology 59 (2006): p. 1851. Cf. Sean M. Hurley and Laura R. Novick, ‘Solving Problems Using Matrix, Network, and Hierarchy Diagrams: The Consequences of Violating Construction Conventions’, The Quarterly Journal of Experimental Psychology 63 (2010): p. 278. ⁵⁸ Marchese, ‘Tables and Early Information Visualization’, p. 35. ⁵⁹ See Howard Wainer, ‘Improving Tabular Displays, with NAEP Tables as Examples and Inspirations’, Journal of Educational and Behavioral Statistics 22 (1997): p. 3, who is followed by Marchese, ‘Tables and Early Information Visualization’, pp. 56 7. Cf. M. Campbell Kelly et al., The History of Mathematical Tables: From Sumer to Spreadsheets (Oxford: Oxford University Press, 2003), p. 3: ‘Tables facilitate the selection, categorization, calculation, checking, and extraction of data’.

Eusebius’ Canon Tables as a Paratext

39

Eusebius’ paratext exemplifies this theoretical work on tables insofar as his Canon Tables reveal to the viewer or reader the relationships that exist amongst the various passages in the gospels, and indeed amongst the gospels as a whole. His Canon Tables therefore order the textual content of the fourfold gospel in a two-tiered structure. First, within each of the first nine tables, he highlights for the viewer the relationship that a passage in a given gospel has with one or more passages in other gospels. In his design, it is specifically the rows that are the bearers of new information.⁶⁰ The juxtaposition of a number in one column next to a number in an adjacent column means that these two passages are somehow related and so belong in the same class. Moreover, at a second-order level of abstraction, Eusebius’ Canon Tables reveal the range of relationships that exist amongst the four gospels, information that is indicated by the fact that he created ten tables in succession rather than simply one. Again, this is information that would not otherwise be readily apparent. It might have been possible, for example, that Luke and John had nothing in common that was not also contained in the other gospels, but the fact that Eusebius records uniquely Luke–John parallels in Canon IX reveals that these two gospels do in fact have a distinct relationship to each another. Hence, the placement of a number in a given position in one of the Canon Tables means: 1) that this passage belongs to the same class as the passages represented by the number(s) to its right and/or left and so is in some way related to those other passages; and 2) that this passage belongs to a larger category of passages that appear in however many gospels the current table records. By presenting this information in the form of tabular matrices, the Canon Tables simplify the complex empirical data of the fourfold gospel, make the abstract notion of gospel relationships more concrete, and allow the user to grasp this information quickly, at a glance, without having to undertake the inductive analysis that would otherwise be required. With this information visually encoded within the tables, Eusebius’ paratext is thus amenable to all four of the uses listed above. First, it efficiently stored the information that Eusebius discovered in his painstaking research into gospel relationships so that this valuable data would not be lost. Secondly, it provided a means for the open-ended exploration of this information about the gospels, creating new ways to engage with the textual content of the fourfold canon and enabling users to arrive at insights that otherwise would have been largely unattainable. Thirdly, by being copied in gospel books over subsequent centuries, the Canon Tables communicated this information to a wide audience and allowed later readers to explore the potentiality inherent in this ⁶⁰ That is to say, the information represented by the columns is not noteworthy or hard to come by because it simply indicates that a set of passages belongs in a particular gospel, which someone could discover simply by reading the given gospel.

40

The Eusebian Canon Tables

information technology. Fourthly, Eusebius’ matrices functioned as a decoration in many gospel books, and in this role were interpreted by later users as saturated with symbolic significance going far beyond the mere juxtaposition of parallel passages.

Tabular Presentation in Babylon and Rome Just as in the first half of this chapter we looked first at the theory of paratexts and then turned to their use in antiquity, so now we turn from a theoretical treatment of information visualization to the history of tables in the centuries leading up to Eusebius’ day. Although the format of the table or matrix is ubiquitous today, being employed in every conceivable area of professional expertise and intellectual investigation, tables were not always so obvious a means of visualizing information. Rather, like other information technologies, it took a moment of creative insight and ingenuity to create the tabular matrix, and once it had been invented, it required a number of other similar moments of insight for its applicability across other domains to be recognized. The earliest collection of tables that have survived are found on cuneiform tablets from the twenty-first century BCE in Mesopotamia, and they were clearly designed for a very specific purpose—the recording of economic transactions. Thus these tables do not appear until more than a millennium after the invention of writing and their emergence in the historical record coincides with the rise of the kingdom of Ur as a large territorial state seeking to unify the Mesopotamian sociopolitical structure. The administrative demands occasioned by the ambitious expansion of Ur were no doubt what led to the creation of this new technology.⁶¹ Once the table had been invented as a device for accounting purposes, over the next two millennia it came to be used in other areas as well, most notably arithmetic, grammar, and astronomy.⁶² However, in the words of Eleanor Robson, who has studied this topic more thoroughly than anyone else, the historical record reveals that the invention of tables ‘was not a one-off event, with clearly traceable consequences across an ever-widening arena of functions, contexts, and cultures. Rather, even within the single cultural milieu of ancient Mesopotamia [there was] a fitful pattern of ⁶¹ Cf. Eleanor Robson, ‘Tables and Tabular Formatting in Sumer, Babylonia, and Assyria, 2500 BCE 50 CE’, in The History of Mathematical Tables, pp. 20 7. Robson later on noted that the earliest surviving mathematical table dates to 2600 BCE (p. 28), predating the tables from Ur by half a millennium. However, this seems to be a rare example, since there apparently was ‘a strong disinclination towards the truly tabular format in later school arithmetic, where one might expect to find tables galore’ (p. 30). Robson pointed out that many cuneiform tablets that are often adduced as examples of mathematical tables are in fact ‘prosaic lists’ rather than true tables (p. 31), a point echoed by Andrew Riggsby with respect to tables in Latin, as discussed below. ⁶² Cf. Robson, ‘Tables and Tabular Formatting’, pp. 27 41.

Eusebius’ Canon Tables as a Paratext

41

invention, partial adoption, disappearance, and re-invention time after time over the course of some two and a half millennia’.⁶³ A tabular form might be employed by a given scribe because he found it made his work easier, but often, it seems, the insight died with him or shortly thereafter, with the result that this information technology never reached a saturation level sufficient for it to become the ‘obvious’ way of recording data regardless of domain. The details of this Mesopotamian history need not detain us. What is important to note are the fits-and-starts history of tabular presentation that Robson has highlighted and the relatively restricted domains in which the device was employed,⁶⁴ since these observations help to unsettle our ingrained assumption that tables are the obvious means of representing the relationship between two categories of items or values. Moreover, in keeping with the theoretical analysis of tables in the last section, Robson similarly drew attention to the possibilities created by the use of tables in Mesopotamia: ‘the new format enabled numerical data and relationships to be seen and explored in ways hitherto unimaginable. The material objects themselves facilitated conceptual advances in quantitative thinking’.⁶⁵ The last datable cuneiform table is found on a tablet that records an Ephemeris predicting lunar and solar eclipses for the years 12 BCE to 43 CE,⁶⁶ and the astronomical content of this table provides a fitting segue to the tradition of tables in Greek, which provides the more proximate context for Eusebius. Before, however, considering tabular presentation in Greek sources, I want to jump across the eastern Mediterranean to the classical Roman world, because a survey of this cultural tradition also highlights how non-intuitive the table can appear to persons unaccustomed to it. In his recent survey of this material, Andrew M. Riggsby has argued that prior to the fourth century CE tables in Latin are ‘vanishingly rare’.⁶⁷ In fact, for Latin speakers tables were so rare that they might even have been ‘an obstacle to understanding’, a ⁶³ Robson, ‘Tables and Tabular Formatting’, p. 20. Cf. Eleanor Robson, ‘Accounting for Change: The Development of Tabular Book Keeping in Early Mesopotamia’, in Creating Eco nomic Order: Record Keeping, Standardization, and the Development of Accounting in the Ancient Near East, ed. Michael Hudson and Cornelia Wunsch (Bethesda, MD: CDL Press, 2004); Marchese, ‘Tables and Early Information Visualization’, pp. 37 9. ⁶⁴ Cf. Robson, ‘Tables and Tabular Formatting’, p. 41: ‘Only rarely in Mesopotamia were tables a mainstream document format. . . . Even in the heyday of Mesopotamian tables in the first half of the eighteenth century BCE, they account for only 1 or 2 per cent of all administrative documents and scribes continued to prefer simpler linear or prosaic methods of managing information’. Elsewhere she examined a set of accounting tables that come from a few decades in the nineteenth century BCE and concluded, ‘Tables are so rare in the cuneiform record we should consider their adoption a matter of individual choice even if those individuals are anonymous to us rather than the outcome of large scale, impersonal forces’ (Robson, ‘Accounting for Change’, p. 126). ⁶⁵ Robson, ‘Tables and Tabular Formatting’, p. 41. ⁶⁶ Robson, ‘Tables and Tabular Formatting’, pp. 40 1. ⁶⁷ Riggsby, Mosaics of Knowledge, chapter 2.

42

The Eusebian Canon Tables

conclusion that Riggsby reached based on a close reading of a passage in Quintilian’s Institutio Oratoria in which the rhetorician seemingly avoids explaining a solution to a legal problem because in his view its tabular format would be too complicated for his students.⁶⁸ In his survey of Latin sources up to the fourth century, Riggsby has found only a handful of true tables—tables, that is, in the sense articulated above in which the horizontal and vertical axes contribute to the meaning of the information displayed.⁶⁹ The first is centuriation, the Roman practice of dividing territory in new colonies into a gridded pattern for settlement. The tabular grid was literally drawn on the land and then represented in small scale on stone or bronze for record keeping. Second, the first-century BCE grammarian Marcus Terentius Varro, at one point in his account of the Latin language, provides instructions for the reader to construct a basic table with three columns showing a series of grammatical cases across three rows corresponding to the three genders. Riggsby’s third example is Jerome’s Chronicle, which was adapted from Eusebius’ earlier Chronicle and translated into Latin in the late fourth century. The fact that this Latin table ultimately comes from Eusebius himself is not without significance and will be explored further in the next chapter. The fourth Latin example consists of two duty rosters from Egypt that show the assignments for soldiers over a series of days. Riggsby also noted a handful of other ambiguous examples that might or might not have been true tables, but even if one concluded that these unclear cases are in fact tabular matrices, we would still be left with a very short list. The brevity of this list might initially seem surprising, and for those familiar with Roman history a number of other possible tables might come to mind. Riggsby, however, has analysed all other tabular displays that have been claimed as Roman ‘tables’ and argued persuasively that in each instance what we are dealing with is in fact a list rather than a true table. These include fasti, laterculi, and other ‘tables’ used for grammar, mathematics, accounting, military rosters, and grain and water distribution.⁷⁰ This raises the obvious question—obvious to us anyway—of why tables were so rare in Roman sources, and in reply Riggsby emphasized the difficulty of imagining and understanding the tabular form for people and cultures that are not already saturated with them, and he also proposed that Romans were more comfortable with constitutive rather than representational reasoning, the latter cognitive process being intrinsic to more complex tables.⁷¹ The important point ⁶⁸ Quintilian, inst. 3.6.66; 3.6.70 1. ⁶⁹ For the examples that follow, see Riggsby, Mosaics of Knowledge, chapter 2. ⁷⁰ Riggsby, Mosaics of Knowledge, chapter 2. ⁷¹ On the latter point, Riggsby comments: ‘I would suggest that Romans are more comfortable in informational contexts with records that are constitutive rather than representational. That is, they record facts called into being by the table itself or as part of the same project as the creation of the table; they avoid registering independently existing facts. So, for instance, the centuriation records coordinates and land assignments that are made by the state and its surveyors, not, say,

Eusebius’ Canon Tables as a Paratext

43

for our purposes is to observe that tables are even rarer in Roman contexts than they are in cuneiform tablets. In Babylonian culture, at least a handful of disciplines on occasion made use of tables and some, like astronomy, did so regularly. However, the same is not true for the Latin world, whether one looks at the practical necessities of administering an empire, the day-to-day world of economic transaction, or the more theoretical world of scholarship.

Tabular Presentation in Ptolemy and Eusebius The purpose of the preceding section has been to demonstrate the rarity of tabular visualization and functionality in the ancient world—or at least in two cultures in which this topic has been investigated at length—in order to throw into sharper relief the startling creativity of Eusebius’ paratext. A comprehensive survey of tables in Greek sources has not, to my knowledge, been undertaken, and such an endeavour is beyond the scope of the present project.⁷² Fortunately, however, this is not necessary, because the most important Greek precursor that we need to consider in this chapter has already been identified by Carl Nordenfalk, and has recently been the subject of specialist scholarship. In the following chapter I will examine in detail three works employing a tabular matrix that are undoubtedly related to Eusebius’ Canon Tables: one by Origen of Alexandria, one from a certain Ammonius of Alexandria, and, finally, Eusebius’ own Chronicle. Two of these three presented passages of text in a tabular format and the third is a combination of numbers and text. Eusebius’ Canon Tables, however, are a purely numerical table, aside from the headers above the columns, and so are more abstract than these other three examples. The closest analogies to the purely numeric table that Eusebius devised for the gospels are in fact Greek astronomical tables that survive from Egypt. Picking up where the cuneiform tablets left off, astronomers in Hellenistic Egypt apparently learned the skill of tabular calculation from their Mesopotamian forebears, and by the second century CE this method had become a standard mode of astronomical practice.⁷³ Numerous examples topographic features. The rosters give assignments for soldiers, not their height and weight. Even the parallel texts give what the writer stipulates is a translation, not an independently identifiable feature like a line number’ (chapter 2). ⁷² The closest thing to a general survey is to be found in J. Mansfeld and D. T. Runia, Aëtiana: The Method and Intellectual Context of a Doxographer. Volume 1: The Sources, Philosophia Antiqua 73 (Leiden: E. J. Brill, 1997), pp. 111 16, which is admittedly brief but full of important citations to the primary literature. I am grateful to Stephen Carlson for drawing this source to my attention. ⁷³ Cf. Mathieu Ossendrijver et al., ‘Tables in Antiquity’, in History of Numerical Tables, ed. D. Tournès (New York: Springer, forthcoming): ‘During the Hellenistic period, Greek math ematicians must have developed an appreciation of these tabular methods and begun to incorporate them into their own work, so that by the time Ptolemy began to write the use of

44

The Eusebian Canon Tables

of such tables survive in the papyri.⁷⁴ In fact, more astronomical tables survive from Oxyrhynchus than astronomical texts, suggesting that ‘[a]stronomy, and perhaps astrology, in Greco-Roman Egypt seems to have been a tabular pursuit’.⁷⁵ Without doubt the most famous of these are the tables devised by Ptolemy in the second century CE, which seem to have been so successful that they displaced virtually all other competing models. As a result, Ptolemy’s astronomical tables ‘can justifiably claim to be the first mass produced mathematical table’.⁷⁶ In his search for the origins of the arches (‘Kanonbögen’) by which Eusbeius’ numeric grids are typically framed, Nordenfalk pointed out the similarity of Eusebius’ Canon Tables with the gridded numeric matrices of Ptolemy.⁷⁷ Later, Japp Mansfeld and David T. Runia highlighted Ptolemy’s tables as ‘an entirely neglected but highly important parallel’ for tabular presentation in Greek sources more broadly, arguing specifically that ‘the formal similarity between a preliminary form of the synoptic presentation of texts and tabular chronography and astronomy etc. is undeniable’.⁷⁸ However, neither of these prior studies have commented on this parallel at any length. The goal of the present section is, therefore, to compare these two works both to strengthen the case that Eusebius was inspired by such astronomical tables, and to use this comparison to highlight some otherwise unnoticed aspects of the function of his information technology. Ptolemy’s astronomical tables and Eusebius’ Canon Tables present analogies in at least four categories: 1) the terminology used to describe them; 2) certain formal characteristics of the way in which each is presented; 3) the way each set of tables implicitly encodes a certain theory about the source material being represented; and 4) the way in which these tables functioned for the end user.

tables was a well established part of mathematical practice’. See also Nathan Sidoli, ‘Mathemat ical Tables in Ptolemy’s Almagest’, Historia Mathematica 41 (2014): pp. 13 14. ⁷⁴ Cf. Alexander Jones, Astronomical Papyri from Oxyrhynchus (P. Oxy. 4133 4300a), Vol umes I and II, Memoirs of the American Philosophical Society 233 (Philadelphia, PA: American Philosophical Society, 1999); Alexander Jones, ‘A Classification of Astronomical Tables on Papyrus’, in Ancient Astronomy and Celestial Divination, ed. N. M. Swerdlow (Cambridge, MA: The MIT Press, 2000). ⁷⁵ Ossendrijver et al., ‘Tables in Antiquity’. ⁷⁶ Campbell Kelly et al., The History of Mathematical Tables, p. 6. ⁷⁷ Nordenfalk, Die spätantiken Kanontafeln, pp. 117 18: ‘Wir begegnen hier auf mehreren Seiten chronographischen Übersichtstabellen in einem linearen Rahmengerüst, das die schla gendste Verwandschaft mit dem Rechtecknetz der eusebianischen Kanontafeln aufweist (Abb. 6). Da die Priorität der ptolemäischen Tabellen feststeht, ergibt sich mit Sicherheit, daß die Schreibschule von Caesarea, als sie die Kanones in ein solches Netz einfasste, ein der älteren Buchkunst geläufiges Mittel für die Einrahmung von Tabellen benutzte’. Cf. Nordenfalk, ‘Canon Tables on Papyrus’, p. 33: ‘Eusebius must have been familiar with [the layout of numbers in a grid] from Classical astronomic tables’. ⁷⁸ Mansfeld and Runia, Aëtiana I, pp. 113, 116.

Eusebius’ Canon Tables as a Paratext

45

First, then, the terminology.⁷⁹ From the surviving Greek sources that discuss tables, it is clear that the standard term used for them was κανών, and in the Almagest, in which Ptolemy laid out his table-based astronomical theory, he used a variety of terms derived from this root. His usual word for referring to mathematical tables is κανόνιον, though more rarely he uses κανών for this purpose. For a row of a table he uses στίχος and for a column σελίδιον. These are both words adapted from existing writing conventions, since the former was used to refer to a line of text, whether in poetry or prose, and the latter referred to a column of writing in a papyrus bookroll. When he wants to refer to the tabular layout of information, Ptolemy uses the word κανονική, such as in the phrase διὰ τῆς κανονικῆς referring to calculation ‘through tabular methods’,⁸⁰ analogous to διὰ τῶν γραμμῶν to refer to an argument demonstrated through trigonometry. The actual drawn table is described as a κανονογραϕία on three occasions,⁸¹ and seven times κανονοποιία is used to refer to ‘the mathematical, or technical, design of the table’.⁸² Finally, after he had composed his theoretical treatise known as the Almagest, titled in Greek σύνταξις μαθηματική, he then extracted the astronomical tables and published them in revised form as a separate treatise titled Πρόχειροι κανόνες, or Handy Tables.⁸³ Hence, while it is not even clear what word a Latin speaker might have used to refer to a tabular matrix, Ptolemy has ‘a well developed terminology for discussing tables’.⁸⁴ This is surely the background that is in view when Eusebius, in his Letter to Carpianus, refers to the ten tables that follow the epistle as κανόνες.⁸⁵ The use of the term κανών and its derivatives to refer to numeric tables probably did not originate with Ptolemy, so this similarity does not prove dependency, but it does clarify the sense of the term as Eusebius uses it here. In this context it does not refer to the church’s canon of scripture or to the ‘canon of truth’. In fact, here it does not relate to any idiosyncratic meaning that the term had in Christian discourse. Instead, it simply means ‘table’, in the sense of numeric grid. At this point we should pause to consider what this background means in terms of how we refer to Eusebius’ paratext. In English scholarship it

⁷⁹ The following is largely based on Ossendrijver et al., ‘Tables in Antiquity’. ⁸⁰ Ptolemy, alm. V.7 (Heiberg, vol. 1.1, p. 383). Cf. Sidoli, ‘Mathematical Tables’, p. 30. ⁸¹ Ptolemy, alm. II.9, III.1, IV.3 (Heiberg, vol. 1.1, pp. 142, 209, 280). ⁸² Ptolemy, alm. III, III.1, III.5, IX.2, IX.3, XI (Heiberg, vol. 1.1, p. 190, 208, 251; vol. 1.2, pp. 211, 218, 359). Cf. Sidoli, ‘Mathematical Tables’, pp. 20, 22. ⁸³ On the relationship of Ptolemy’s two works, see Alexander Jones, ‘Uses and Users of Astronomical Commentaries in Antiquity’, in Commentaries Kommentare, ed. Glenn W. Most (Göttingen: Vandenhoeck & Ruprecht, 1999), pp. 160 1. ⁸⁴ Ossendrijver et al., ‘Tables in Antiquity’. ⁸⁵ As recognized also by Mansfeld and Runia, Aëtiana I, p. 116; Wallraff, Kodex und Kanon, pp. 27 8; McKenzie and Watson, The Garima Gospels, p. 145.

46

The Eusebian Canon Tables

is universally called the ‘Canon Tables’, which is merely a translation of the German compound ‘Kanontafeln’, used by Nordenfalk and others. However, if the above argument about the meaning of κανών is correct, then this appellation is unfortunately tautologous, since for Eusebius κανών simply means ‘table’. It would, therefore, be more accurate to refer to the paratextual device as simply the ‘Tables’ for the gospels, but the alternative designation has so implanted itself in the scholarly discourse that at this point it would probably be futile to try and change it. Hence, the present study continues to use the pleonasm ‘Canon Tables’, though in full awareness of its redundancy. There are at least three formal analogies between Ptolemy’s work and that of Eusebius. The first and most obvious concerns the mis-en-page of the two. Ptolemy, surely as much as Eusebius, gave careful thought to the design of his tables, explaining, for example, that he chose a base of eighteen years for one table because it produces symmetry in the layout.⁸⁶ When Nordenfalk drew attention to Ptolemy’s treatise as a precursor to Eusebius’ paratext, he adduced as an example the copy of the Handy Tables preserved in Vat. gr. 1291, an eighth- or ninth-century manuscript that postdates by several centuries the earliest witnesses to Eusebius’ Canons.⁸⁷ However, as already noted, several fragments of the Handy Tables have survived in the papyri from Oxyrhynchus (P. Oxy. 4167-4171; see figs 16–17), some dating to the third century, and all show the same grid outline, drawn in red, with a header row at the top displaying a title for the columns beneath, and subsequent rows containing two numbers per cell.⁸⁸ Fragments from another set of astronomical tables dated to c.200 CE (P. Lond. 1278) show a similar presentation with lines ruled in red and three or five numbers per cell.⁸⁹ This gridded format is strikingly similar to the numerical grid that characterizes the earliest surviving Greek witnesses to Eusebius’ Canons, such as the ⁸⁶ Ptolemy, alm. III.1 (Heiberg, vol.1, p. 209). Cf. Ossendrijver et al., ‘Tables in Antiquity’: ‘The tables that are preserved in our manuscripts almost certainly still have the format that Ptolemy intended, as is attested by the fact that he prefaces the table itself with a verbal description of how it will be laid out, often including specific details such as how many rows and columns it will contain’. ⁸⁷ For an image of Vat. gr. 1291, fol. 22v, see plate 5 in Nordenfalk, ‘Canon Tables on Papyrus’. On this manuscript, see further I. Spatharakis, ‘Some Observations on the Ptolemy Ms Vat. Gr. 1291: Its Date and the Two Initial Miniatures’, Byzantinische Zeitschrift 71 (1978): pp. 41 9; David H. Wright, ‘The Date of the Vatican Illuminated Handy Tables of Ptolemy and of Its Early Additions’, Byzantinische Zeitschrift 78 (1985): pp. 355 62, and most recently Benjamin Anderson, Cosmos and Community in Early Medieval Art (New Haven, CT: Yale University Press, 2017), pp. 114 26. Anderson dates it to the late eighth or early ninth century, aside from the first four folios, which were made before 830 and bound to the present manuscript some time before the fifteenth century. ⁸⁸ Jones, Astronomical Papyri From Oxyrhynchus, I. pp. 118 19, 160 5 (introduction and commentary), II. pp. 118 49 (edition and translation). ⁸⁹ See O. Neugebauer and T. C. Skeat, ‘The Astronomical Tables P. Lond. 1278. With a Note on the Paleography of the Fragments’, Osiris 13 (1958): pp. 93 113. These fragments are also discussed in Mercier, Πτολεμαίου Πρόχειροι Κανόνες, pp. 182 5.

Eusebius’ Canon Tables as a Paratext

47

Fig. 16. P. Oxy. 4168, side b. Fragments of a codex containing Ptolemy’s Handy Tables, dated to the fourth century. Notice the use of the tabular matrix as a structuring device, including the contrasting red and black inks for ease of use. Header rows are at the top of the table, and each cell below contains two numerals. Courtesy of The Egypt Exploration Society and the University of Oxford Imaging Papyri Project

aforementioned fragments of sixth- or seventh-century papyri discovered at the monastery of Epiphanius in Thebes (see fig. 14). Here too a numerical grid is formed by crisscrossing vertical and horizontal lines drawn in red, with numerals entered in groups of four per cell. This aspect of the design layout of the Canon Tables is not unique to Greek examples but was carried over into most of the languages into which the apparatus was translated, suggesting that it is a feature that goes back to Eusebius himself.⁹⁰ Moreover, unlike the geometric designs that appear in Greek mathematical texts, which were ‘drawn as simple schematics’, Ptolemy’s tables sometimes included ‘ornamental touches’⁹¹ like the figures drawn in the arches at the top of the columns in Vat. gr. 1291. As we will see in chapter seven, Eusebius’ tables were also usually transmitted with ornamentation, which may in some form derive from Eusebius’ original design. ⁹⁰ Cf. Nordenfalk, Die spätantiken Kanontafeln, pp. 74 5. ⁹¹ Ossendrijver et al., ‘Tables in Antiquity’.

48

The Eusebian Canon Tables

Fig. 17. P. Oxy. 4169, side a. Fragments of a codex containing Ptolemy’s Handy Tables, dated to the third century. Courtesy of The Egypt Exploration Society and the University of Oxford Imaging Papyri Project

Hence, the presentation of a grid housing numbers, the use of alternating red and black ink, and the presence of decoration are all features of Eusebius’ design that might have come from Ptolemy or an intermediary. The second formal similarity can be dealt with more briefly. In Ptolemy’s Almagest, ‘a table never appears on its own’, but instead ‘is found in a paratext that relates it to the geometrical model, shows how the parts of the table are related to one another, describes how the table is laid out, and provides algorithms for its use’.⁹² Even Ptolemy’s Handy Tables are prefaced by a lengthy prologue addressed to a certain ‘Syrus’, which provides instructions for using the work.⁹³ Eusebius’ Canon Tables, though admittedly much simpler in terms of mathematical calculation, are likewise headed by the Letter to Carpianus which explains the origins and use of the cross-referencing system. In fact the Letter to Carpianus is a sort of paratext to the paratext of ⁹² Ossendrijver et al., ‘Tables in Antiquity’. ⁹³ Ptolemy, δια. (Heiberg, p. 159 85). A translation may be found at Mercier, Πτολεμαίου Πρόχειροι Κανόνες, pp. 178 81.

Eusebius’ Canon Tables as a Paratext

49

the ten tables, since it exists to ensure the proper reception of the numerical grids that follow. A final formal similarity between Eusebius’ creation and Ptolemy’s Handy Tables is that, of the surviving papyri containing copies of Ptolemy’s work, almost all (four out of five) came from codices rather than rolls.⁹⁴ In fact, in the second and third centuries, when the codex was still far from being the most common format for book production, one frequent use of the codex was for ‘paraliterary’ works such as astronomical tables, as attested by the surviving non-Christian examples of codices from this period (e.g., 25 per cent of the total dated to the second or second–third century are classified as ‘paraliterary’).⁹⁵ There was a practical reason for this, since a complete copy of the Handy Tables would require about 30 metres of papyrus, a length unwieldy as a single roll but one that a single codex could accommodate.⁹⁶ However, another reason may have been that astronomical tables encouraged their user to flip back and forth between tables to achieve a desired calculation, a sort of hypertextual cross-referencing that is more easily achieved in a codex, and one similar to the reading style that Eusebius’ system is designed to facilitate (see chapter three). It is remarkable that the only parallel to the tabular format of Eusebius’ Canons occurs in technical literature such as astronomical tables.⁹⁷ In a book culture in which the aesthetic simplicity and regularity of the scroll was a mark of the high literary quality of the text contained therein and the cultured taste of its users,⁹⁸ it was a striking innovation for Eusebius to incorporate this tool from the genre of manuals and reference works into a codex containing highly valued literary works.⁹⁹ It was a mixing of two worlds ⁹⁴ Cf. Jones, ‘Uses and Users’, p. 161. ⁹⁵ Larry W. Hurtado, The Earliest Christian Artifacts: Manuscripts and Christian Origins (Grand Rapids, MI: Eerdmans, 2006), pp. 50 1. On this point, see also Harry Y. Gamble, Books and Readers in the Early Church: A History of Early Christian Texts (New Haven, CT: Yale University Press, 1997), pp. 65 6, who highlighted the preponderance of ‘professional manuals’ among early codices, interpreting this fact as indicative of the initially ‘utilitarian’ status of this book form. ⁹⁶ Jones, Astronomical Papyri From Oxyrhynchus, p. 160. The normative range for the ancient bookroll was 3 15 metres (William A. Johnson, ‘The Ancient Book’, in The Oxford Handbook of Papyrology, ed. Roger S. Bagnall (Oxford: Oxford University Press, 2009), p. 264). ⁹⁷ Cf. Nordenfalk, Die spätantiken Kanontafeln, p. 117: ‘Wollen wir in der älteren Buchkunst nach Vorgängern der Kanontafeln suchen, so haben wir ims an Schriften mit Tabellen Beilagen zu halten. In der antiken Literatur gehören diese ausschließlich dem wissenschaftlichen Schrift tum an’. Nordenfalk’s conclusion is reaffirmed in Klaus Wessel, ‘Kanontafeln’, in Reallexikon zur byzantinischen Kunst, vol. 3, ed. Klaus Wessel and Marcell Restle (Stuttgart: Anton Hiersemann, 1978), p. 966. ⁹⁸ Cf. Johnson, ‘The Ancient Book’, pp. 262 3: ‘the net effect [of the bookroll] is designed for clarity and beauty but not ease of use, much less mass readership. . . . Strict functionality, clearly, is not a priority in bookroll design. The bookroll seems, rather, an egregiously elite product intended in its stark beauty and difficulty of access to instantiate what it is to be educated’. ⁹⁹ Cf. Johnson, ‘The Ancient Book’, p. 267: ‘Without implying direct cause and effect, we can see that medieval characteristics such as the rise of scriptoria, renewed encyclopedism, and the habit of extensive marginal annotation can be located within the series of changes we associate with the transition to codex form’.

50

The Eusebian Canon Tables

that might have appeared surprising and perhaps even jarring to the elite reader of the fourth century.¹⁰⁰ These formal similarities make Ptolemy’s Handy Tables the closest analogy to Eusebius’ Canon Tables, and, in fact, we can perhaps identify the source that might have transmitted such an idea to the Caesarean bishop. In the seventh book of his Ecclesiastical History Eusebius gave a brief notice of a certain Anatolius of Laodicea who lived in the latter half of the third century. Anatolius was originally from Alexandria and, according to Eusebius, had expertise in a variety of disciplines including philosophy, arithmetic, geometry, astronomy, dialectics, physics, and rhetoric. In fact, the Alexandrians were so impressed by his learning that they appointed him to a chair of Aristotelian philosophy.¹⁰¹ He later became bishop of Laodicea and left behind at least two books that were known to Eusebius.¹⁰² The first was Arithmetical Introductions in Ten Books¹⁰³ and the other was a treatise on the date of Easter bearing the title Κανόνες περὶ τοῦ πάσχα. Eusebius cited several extracts from the latter work, which are sufficient to make clear that it was a technical discussion comparing differing calendrical systems and describing the cycles of the sun, the planets, and the zodiac, in other words, precisely the sort of information one would have drawn from Ptolemy’s Handy Tables.¹⁰⁴ Moreover, the presence of the word κανόνες in the title implies that Anatolius’ composition included not only such theoretical discussion, but also actual tables for calculating the date of Easter, which may have been modelled on Ptolemy’s earlier work. ¹⁰⁰ My suggestion here is analogous to Andrew Riggsby’s proposal that diagrams are sparse in Latin literary works at least partially as a result of the cultural expectations about what an elite book should look like: ‘Diagrams are workmanlike in a world in which elites value speech above manual labor’ (Riggsby, Mosaics of Knowledge, chapter 5). Similarly, Reviel Netz has argued that the presence of diagrams in Greek mathematical works marked them out as a genre sharply distinct from Greek literary texts, so much so that ‘[i]n this respect, at least, Greek mathematics was the anti literature’ (Netz, ‘Authorial Presence’, p. 241). On illustrations in ancient technical literature, see Alfred Stückelberger, Bild und Wort: das illustrierte Fachbuch in der antiken Naturwissenschaft, Medizin und Technik, Kulturgeschichte der antiken Welt 62 (Mainz am Rhein: P. von Zabern, 1994). ¹⁰¹ Eusebius, HE 7.32.6. ¹⁰² On Eusebius’ knowledge of and access to the works of Anatolius, see Andrew Carriker, The Library of Eusebius of Caesarea, VCSup 67 (Leiden: Brill, 2003), pp. 184 7. ¹⁰³ Mentioned in Eusebius, HE 7.32.20. The fragments of the treatise are printed in PG 10, col. 232 6. An epitome titled Περὶ δεκάδος καὶ τῶν ἐντὸς αὐτῆς ἀριθμῶν, which may come from the same introduction, can be found in J. L. Heiberg, Anatolius. Sur les dix premiers nombres (Macon: Protat, 1901). Cf. Aaron P. Johnson, ‘Eusebius’ Praeparatio Evangelica as Literary Experiment’, in Greek Literature in Late Antiquity: Dynamism, Didacticism, Classicism, ed. Scott Fitzgerald Johnson (London: Routledge, 2006), pp. 78 9. ¹⁰⁴ The fragments are cited at Eusebius, HE 7.32.14 19 and printed in PG 10, col. 209 22. It is possible that a portion of the treatise survived in a corrupted Latin translation. Although a work passed down in Latin, attributed to Anatolius, and bearing the title De ratione paschali was long regarded as pseudonymous, recently an argument has been put forward that it does indeed go back to this figure: Daniel P. McCarthy and Aidan Breen, The Ante Nicene Christian Pasch: De Ratione Paschali: The Paschal Tract of Anatolius, Bishop of Laodicea (Dublin: Four Courts Press, 2003). Mercier, Πτολεμαίου Πρόχειροι Κανόνες, p. 2, acknowledged that Christian authors made use of Ptolemy’s tables for calculating the date of Easter.

Eusebius’ Canon Tables as a Paratext

51

Given that Eusebius was well aware of Anatolius’ Tables concerning Easter, the latter treatise may well have been the source from which he derived the idea of using a purely numeric table. If so, then Eusebius’ insight was his realization that such a technique could be applied to a very different data set, namely gospel relationships. If this hypothesis is correct, this in itself is a remarkable achievement because one of the conclusions of Riggsby’s analysis is that those tabular matrices that did exist in the Roman world appear to have been one-off achievements. In other words, once someone realized a particular data set could be usefully displayed in this form, they seem not to have taken the next step of abstraction and applied the same technique to other areas of knowledge. As a result, the transfer of this kind of information technology from a calendrical or astronomical work to the realm of literary analysis is highly unusual against the broader backdrop of the ancient world and represents a significant degree of creative ingenuity.¹⁰⁵ Building upon these formal similarities between Eusebius’ work and those of Ptolemy and Anatolius, I now want to draw attention to a further parallel in the representational nature of these two attempts at information visualization, and the similar, implicit message carried by that representation in each case. Both Ptolemy’s Handy Tables and Eusebius’ Canon Tables established a system of numeric signs arranged in an orderly grid which related to a reality that appeared both larger and more complex than the tables themselves. These tools were never intended as an end in themselves, but rather were designed as a map to guide the reader to an understanding of phenomena whose complexity might otherwise overwhelm the uninitiated. And, like any map, both these sets of tables greatly simplified the empirical reality that they represented. Ptolemy himself was aware of this inescapable fact and indeed exploited it in his Almagest for the purpose of convincing the reader of his overarching theory of celestial motion: As for the evaluation of the paths of the sun and the others, according to the occurrences of each of them, which the composition of the table framework, part by part (ἡ σύνταξις τῆς κατὰ μέρος κανονοποιίας), is disposed to supply as handy, or rather explicit, we hold that it must be set out as a purpose and an aim of the mathematician to show the accomplishment (ἀποτελούμενα) of all phenomena of the heavens by means of regular and circular motions, and that that table framework (κανονοποιία) most appropriate and suited to this purpose is the separation of the regular motion, part by part, from the apparent anomaly, following from the circular models, and then the exhibition of their apparent paths is from the mixing and combination of both of these.¹⁰⁶

¹⁰⁵ I am grateful to Andrew Riggsby for helping me to realize this point. ¹⁰⁶ Ptolemy, alm. III.1 (Heiberg, vol. 1, p. 208). The above translation of this passage is taken from Ossendrijver et al., ‘Tables in Antiquity’. For a further exploration of the function of tables in Ptolemy’s Almagest, see especially Sidoli, ‘Mathematical Tables’, who discusses this passage at p. 20.

52

The Eusebian Canon Tables

To those unfamiliar with ancient astronomy, the above passage might seem challenging. Fortunately the details of it need not concern us. The basic point for my purpose is to observe that Ptolemy says he is deliberately simplifying the table in order to aid the user’s understanding of the material at hand. More specifically, he does so in order to demonstrate ‘how the irregular motions that are apparent are actually composed out of more uniform motions, and ultimately related to regular motion in a circle’.¹⁰⁷ Ptolemy’s tables are, therefore, designed on the basis of his theory of astronomical motion, and their purpose is to convince the user that this theory is correct, despite observational evidence that would, at least initially, suggest otherwise.¹⁰⁸ The same is analogously true for Eusebius’ Canon Tables. They also referred to, and in the process simplified, complex and apparently disorderly phenomena—the texts of the four gospels—and did so probably in order that they might be used in service of a theory about the phenomena at hand— that the four gospels when read properly are harmonious with one another. In other words, just as Ptolemy’s Handy Tables offered its user the ability to discern the patterns and predict the movements of the heavenly bodies by symbolically representing their ordered progression across the heavens in a elegant grid of numbers, so also Eusebius’ gospel Canons, with its symbolic representation of textual relationships in tabular form, offered the reader of the gospels the ability to behold the harmonious order governing the microcosm of the text, hidden yet waiting to be discovered.¹⁰⁹ Someone might be tempted to disparage Eusebius’ paratext precisely for this reason, as though it were a duplicitous attempt to mask the real state of affairs in which gospel contradiction undermines divine truth. Such a critical reading is, however, not warranted. I already observed above that one of the primary functions of a table is to

¹⁰⁷ Ossendrijver et al., ‘Tables in Antiquity’. ¹⁰⁸ Cf. Courtney Roby, ‘Framing Technologies in Hero and Ptolemy’, in The Frame in Classical Art: A Cultural History, ed. Verity J. Platt and Michael Squire (Cambridge: Cambridge University Press, 2017), p. 537. In addition to its obvious utility to someone undertaking astronomical calculation, Byzantine copies of Ptolemy’s astronomical tables could take on political associations as well, as argued by Anderson, Cosmos and Community, p. 126: ‘The Vatican Ptolemy was above all a practical handbook for experts in astronomy and astrology, and numerous marginal annotations demonstrate that it was used as such. . . . If the solar diagram [on fol. 9r of Vaticanus graecus 1291; Anderson’s fig. 57 and 63] evokes the ideal of a well ordered cosmos mirrored by a well ruled empire, the book would then become its guarantee, the manual that ensures imperial access to the immutable laws of heaven’. Anderson drew attention to the fact that the military manual attributed to Emperor Leo VI explicitly mentioned Ptolemy’s Handy Tables as a required text (p. 117). ¹⁰⁹ So also Watson, The Fourfold Gospel, p. 115: ‘It was the achievement of Eusebius of Caesarea to show, through his canon tables, how the apparent chaos of the four different tellings of the same story can be reduced to rational and harmonious order’. Cf. Bruno Reudenbach, ‘Der Codex als heiliger Raum: Überlegungen zur Bildausstattung früher Evangelienbücher’, in Codex und Raum, ed. Stephan Müller, Lieselotte E. Saurma Jeltsch, and Peter Strohschneider, Wolfenbütteler Mitte lalter Studien (Wolfenbüttel: Herzog August Bibliothek, 2009), p. 66.

Eusebius’ Canon Tables as a Paratext

53

simplify a complex data set so that it can be more easily manipulated and explored. When viewed in the light of this principle, it is possible to see Eusebius’ Canon Tables as an information device intended to open up rather than close down discussion of the phenomena in question. In chapter three I will return to this topic and explore more fully what kind of ‘parallels’ Eusebius created, so enough has been said about it at present. In concluding this section I want to bring the discussion back to the idea of function and highlight the analogy between Ptolemy’s work and that of Eusebius with respect to how the end user interacted with the information represented in the tabular form. Though the purposes of the two sets of tables were different, operating in two separate domains of knowledge, the common tabular technology they employ functions in both cases to make specialist knowledge accessible to those without specialist training.¹¹⁰ As I have already mentioned, the Handy Tables consisted of an expanded version of the tables Ptolemy had already included in his Almagest, albeit stripped of all the theoretical discussion contained in the earlier work, replaced by a short preface giving straightforward instructions. Thus, would-be astronomers, or more likely astrologers, could use Ptolemy’s tables to make predictions regardless of whether or not they understood all of the observational and mathematical expertise used by Ptolemy in creating them.¹¹¹ Similarly, even an amateur reader of a gospel codex equipped with Eusebius’ apparatus had the ability to find parallels amongst the gospels, a task that would otherwise require a high degree of literary training, and a great familiarity with all four sources. Both inventions therefore illustrate the principle that information visualization, when successful, can communicate expert knowledge to a broader audience.¹¹²

¹¹⁰ This observation, in its general form, is not new. The eighteenth century French philoso pher Marquis de Condorcet, in his Outlines of an Historical View of the Human Mind, emphasized the epoch making inventions of writing and the printing press, and argued that tables and charts can continue this revolution by making knowledge more accessible, thus furthering the ongoing perfection of humankind. For a discussion, see Roger Chartier, Forms and Meanings: Texts, Performances, and Audiences From Codex to Computer (Philadelphia, PA: University of Pennsylvania Press, 1995), pp. 8 11. ¹¹¹ As remarked by Mercier, the Handy Tables ‘provides in an accessible form all that is needed for the ready determination of the positions of Sun, Moon, and planets . . . the Handy Tables are designed above all to permit efficient calculation’. As a result, ‘we have a work that is much more practical than the Almagest, and designed for the person who may not be primarily interested in the scientific basis of astronomy, but who only wishes to have at hand the means to carry out calculations in the most direct way’ (Mercier, Πτολεμαίου Πρόχειροι Κανόνες, pp. 1 2). The same point was also recognized in Roby, ‘Framing Technologies in Hero and Ptolemy’, 534 5, who pointed out that for the user without expert knowledge the tables remain ‘black boxes’. ¹¹² Burgess and Kulikowski have similarly argued that the chronicle genre in the Middle Ages, which was descended from Jerome’s Latin translation of Eusebius’ Chronicle, represented ‘a democratization of history, allowing everyone, for the first time, to easily access and understand the scope of the Mediterranean, and later the European, past’ (R. W. Burgess and Michael Kulikowski, Mosaics of Time: The Latin Chronicle Traditions from the First Century BC to the Sixth Century AD, Studies in the Early Middle Ages 33 (Turnhout: Brepols, 2013), p. 129).

54

The Eusebian Canon Tables

CONCLUSIO N The above survey of paratextuality and the tabular visualization of information provides us with enough of an orientation to answer the ‘what’ question when it comes to the Eusebian apparatus: the Canon Tables are a paratextual system designed to facilitate the reader’s interaction with the text of the fourfold gospel by performing two primary functions: 1) structuring the reader’s experience of the text by dividing four long narratives into smaller sections; and 2) making apparent the relationships that exist across these four narratives that would otherwise have been implicit.¹¹³ These structuring and linking functions of the paratextual apparatus have the result of organizing or ordering the textual content contained within the fourfold gospel, achieving a twofold classification at the micro-level of discrete passages and at the macro-level of entire gospels. The only thing to add to this summary of what has been said thus far is to point out that Eusebius did not design this new information technology merely for his own use but rather intended for it to circulate widely, which must mean that he published it as a component in a new edition of the fourfold gospel.¹¹⁴ The inclusion of paratexts in a new edition of a literary work was a common way for an editor in antiquity to attempt to exert some hermeneutical control over the reception of the texts at hand, as has been explored most recently by Eric W. Scherbenske’s study of the editions of Paul published by Marcion, Euthalius, and Victor of Capua, and in this respect Eusebius’ efforts were in keeping with standard editorial practice.¹¹⁵ However,

¹¹³ Similarly, the anthropologist Jack Goody has argued that the effect of writing and related information technologies is ‘to make the implicit explicit’ (Jack Goody, The Power of the Written Tradition, Smithsonian Series in Ethnographic Inquiry (Washington, DC: Smithsonian Institu tion Press, 2000), p. 164). ¹¹⁴ I use the term ‘publish’ with an awareness of the differences between how this process functioned in antiquity and today. On this topic, see Raymond J. Starr, ‘The Circulation of Literary Texts in the Roman World’, The Classical Quarterly 37 (1987): pp. 213 23; Gamble, Books and Readers, pp. 82 143; Small, Wax Tablets, pp. 26 40; Johnson, Readers and Reading Culture, pp. 52 6, 85 91, 131 2; Scherbenske, Canonizing Paul, chapter 1. Francis Watson argued similarly that ‘Some four gospel codices were already in circulation during the third century, but Eusebius was responsible for the first four gospel edition’ (Watson, The Fourfold Gospel, p. 123). Cf. Amphoux, ‘La division’, pp. 311 12. ¹¹⁵ Scherbenske, Canonizing Paul. Caroline Humfress has drawn a similar conclusion about Justinian’s legal Digest: ‘by confirming the Digest’s authority in an imperial prologue issued ‘In the Name of Our Lord God Jesus Christ’, Justinian effected the rhetorical Christianization of all the non Christian classical juristic books contained within it. . . . In other words, the Christian authority of the Digest was not achieved by a Christianization of the substantive principles of classical Roman jurisprudence; it was rather created by enveloping the hallowed classical books of the Roman jurists within a new order of texts’ (Caroline Humfress, ‘Judging By the Book: Christian Codices and Late Antique Legal Culture’, in The Early Christian Book, ed. William E. Klingshirn and Linda Safran, CUA Studies in Early Christianity (Washington, DC: The Catholic University of America Press, 2007), p. 143).

Eusebius’ Canon Tables as a Paratext

55

given that the vast array of later manuscripts bearing Canon Tables in such a diverse range of languages all go back, in whatever mediated fashion, to a copy or copies issued from Eusebius’ scriptorium at Caesarea,¹¹⁶ his edition of the gospels surpassed in popularity all of these other editions of biblical texts. In fact, it would not be an exaggeration to say that Eusebius’ edition of the fourfold gospel, equipped with his ingenious paratextual system, turned out to be one of the most successful publishing efforts in all of antiquity.

¹¹⁶ Obviously, this claim relates only to the paratext and not to the textual version of the gospels transmitted in those later traditions.

2 The Origins of Scholarship on the Fourfold Gospel From Alexandria to Caesarea

Works¹ of great ingenuity do not emerge from a vacuum but typically are an improvement upon existing technologies, or the creative realization of previously unforeseen combinations of existing factors. So it was with Eusebius’ Canon Table apparatus. As he himself acknowledged, his scholarly project built upon the work of a prior Alexandrian author by the name of Ammonius. Ammonius’ composition, which I will argue was titled the Diatessaron-Gospel, has left no trace in the manuscript tradition,² with the result that for information about it we are wholly dependent on the short description provided by Eusebius himself in his Letter to Carpianus. Because Eusebius mentions his predecessor’s work in the context of introducing his own Canon Tables, scholars have long recognized some sort of connection between the two figures. However, among those who have previously commented on them there has been a lack of clarity regarding the precise relationship between their respective works, with some using ambiguous terminology that blurs the distinction between the Diatessaron-Gospel and the Canon Tables, and others arguing incorrectly that the two works had nothing in common. Hence, in what follows I intend to highlight and give a more nuanced account of the distinct contributions of these two figures in what was a joint scholarly enterprise stretching over a century or more, which represents the earliest thorough investigation of the patterns of similarity and difference that exist within the fourfold gospel canon. ¹ An earlier version of this chapter appeared as Matthew R. Crawford, ‘Ammonius of Alexandria, Eusebius of Caesarea and the Origins of Gospels Scholarship’, New Testament Studies 61 (2015): pp. 1 29. In the present chapter I have not fundamentally changed any of the arguments of the earlier piece but have made them more precise and provided them with a wider base of evidence. ² Though see Appendix 3 for my assessment of a possible surviving fragment of Ammonius’ work.

The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0003

The Origins of Scholarship on the Fourfold Gospel

57

The first half of this chapter is an attempt to identify Ammonius and contextualize his literary endeavours based on the meagre evidence that can be gleaned from the historical record. Although conclusive proof is lacking, I argue that it is probable that this Ammonius was the teacher of Origen in Alexandria and also a Peripatetic philosopher famed for his philological scholarship.³ If so, then Origen’s well-known Hexapla and Ammonius’ Diatessaron-Gospel were most likely parallel attempts at using cutting-edge scholarly tools to analyse the emerging Christian canon of sacred texts. In the second half of the chapter I pivot to Eusebius and survey the various uses of tabular methods across his corpus in order to show how these novel experiments in information visualization provided him with the insights that enabled him to build upon—or, more precisely, improve upon—Ammonius’ earlier work on the gospels. By using Ammonius’ composition as the raw materials for his own Canon Tables, Eusebius became the channel for the transmission of Alexandrian philological scholarship to the wider Mediterranean world of late antiquity.

THE DIATESSARON-GOSPEL OF AMMON IUS OF ALEXANDRIA

Who was Ammonius? Given its highly specialized focus, Ammonius’ work on the gospels probably found a small audience in the third and fourth centuries, so its circulation was surely limited. This observation helps to explain why the only description of it is found in Eusebius’ Letter to Carpianus, in which the Caesarean historian ³ I use the word ‘scholarship’ to describe the work of Ammonius and Eusebius as a reference to their participation in the wider world of Greek textual learning and investigation, which had its origins in the library and Mouseion at Alexandria in the third century BCE. As defined in the classic study of Rudolf Pfeiffer, ‘Scholarship is the art of understanding, explaining, and restoring the literary tradition. It originated as a separate intellectual discipline in the third century before Christ through the efforts of poets to preserve and to use their literary heritage’ (Rudolf Pfeiffer, History of Classical Scholarship: From the Beginnings to the End of the Hellenistic Age (Oxford: Clarendon Press, 1968), p. 3). Similarly, Eleanor Dickey states that ‘scholarship’ in this sense refers to ‘any type of work concentrating on the words, rather than the ideas, of ancient pagan authors: textual criticism, interpretation, literary criticism of specific passages, grammar, syntax, lexicography, etc.’ (Eleanor Dickey, Ancient Greek Scholarship: A Guide to Finding, Reading, and Understanding Scholia, Commentaries, Lexica, and Grammatical Treatises, From Their Begin nings to the Byzantine Period (Oxford: Oxford University Press, 2006), vii). On literary criticism in antiquity more specifically see René Nünlist, The Ancient Critic At Work: Terms and Concepts of Literary Criticism in Greek Scholia (Cambridge: Cambridge University Press, 2009), and for a broad overview of the philological tradition from antiquity to the modern period, see James Turner, Philology: The Forgotten Origins of the Modern Humanities (Princeton, NJ: Princeton University Press, 2014).

58

The Eusebian Canon Tables

Fig. 18. Portrait of Eusebius above the beginning of his Letter to Carpianus in the Gladzor Gospels (c.1300). The facing page of the manuscript contains a matching portrait of Carpianus above the latter half of the letter. Gladzor Gospels, Library Special Collections, Charles E. Young Research Library, UCLA, Armenian MS 1, p. 4

lays out the origin and function of his system of Canon Tables (see fig. 18). Here Eusebius gives no further biographical details about his predecessor beyond the fact that he was from Alexandria (Ἀμμώνιος ὁ Ἀλεξανδρεὺς).⁴ However, in his Ecclesiastical History, Eusebius also mentions an Alexandrian ⁴ ep. Carp. (NA²⁸, p. 89*).

The Origins of Scholarship on the Fourfold Gospel

59

Ammonius who composed, among other works, a treatise titled On the Harmony of Moses and Jesus (Περὶ τῆς Μωυσέως καὶ Ἰησοῦ συμϕωνίας). This Ammonius, the historian tells us, was ‘highly esteemed among many’ (παρὰ τοῖς πλείστοις εὐδοκιμοῦντος), and his works were still in circulation among the ‘scholarly’ (παρὰ τοῖς ϕιλοκάλοις) in the early fourth century.⁵ The fact that in the Letter to Carpianus Eusebius offers no further information about the Ammonius engaged in study of the gospels may indicate that he knew nothing else about this figure. However, it is more likely that he is brief in his mention of Ammonius because he assumed his readers would already know of his identity, a supposition that coincides well with the reported fame of the Ammonius responsible for On the Harmony of Moses and Jesus. A further argument for the attribution of the two works to the same Ammonius is their common theme. Eusebius does not tell us why Ammonius composed his work on the gospels, but it likely was the same as Eusebius’ own intent behind the Canon Tables, namely, to show the harmony and agreement of the evangelists. Similarly, Ammonius’ other work was focused on presenting the συμϕωνία between Jesus and Moses. Common to both the relationship of Jesus to Moses and the interrelations of the fourfold gospel is the possibility of discord which threatens to undermine divine truth, an Achilles heel exploited by Christians such as Marcion, as well as by pagan critics like Celsus. It is plausible, therefore, that a second- or third-century Christian engaged in this intellectual milieu might deem it necessary to demonstrate both the ‘harmony’ of Moses and Jesus and of the four separate accounts of Jesus’ life. It is best, therefore, to assume these two Ammonii—the one mentioned in the Letter to Carpianus and the one referred to in the Ecclesiastical History— are one and the same, a conclusion already reached by Jerome in the later fourth century who, almost certainly relying on Eusebius, gave a brief notice of a single Ammonius in his De viris.⁶ I will therefore proceed on the assumption ⁵ Eusebius, HE 6.19.10 (SC 41, p. 116). J. Edgar Bruns, ‘The “Agreement of Moses and Jesus” in the “Demonstratio Evangelica” of Eusebius’, Vigiliae Christianae 31 (1977): pp. 117 25, proposed that a passage in Eusebius’ own Demonstratio evangelica draws upon this lost work of Ammonius on Jesus and Moses. For a more recent assessment of whether or not Eusebius himself had firsthand knowledge of this work, see Carriker, The Library of Eusebius of Caesarea, pp. 182 4, who concluded, ‘Eusebius does not clearly indicate his own possession of the Harmony, and because he does not, it is possible that Eusebius knew of it only at second hand’. ⁶ Jerome, vir. 55 (Richardson, pp. 33 4). Jerome attributed to this Ammonius two works: De consonantia moysi et iesu and the Euangelici canones. The latter work is of course the same one that Eusebius referred to as the Diatessaron Gospel, though Jerome referred to it by the title of Eusebius’ own Canon Tables, probably reflecting a confusion already at this stage over the exact relation between the two works. The fact that Jerome named only these two works of Ammonius seems to indicate that he had no independent access to them and was entirely dependent on the reports of Eusebius. Theodor Zahn, ‘Der Exeget Ammonius und andere Ammonii’, Zeitschrift für Kirchengeschichte 38 (1920): pp. 4 5, also argued that the same Ammonius was responsible for both works.

60

The Eusebian Canon Tables

that the report about Ammonius in the Ecclesiastical History sheds light on the person whose work served as the most direct inspiration for Eusebius’ Canon Tables. Saying more about this Ammonius, however, necessarily enters into more contested territory. Indeed there is an ongoing, and perhaps at some level irresolvable debate over the identity of the Ammonius discussed in Eusebius’ Ecclesiastical History. The relevant passage in the Ecclesiastical History occurs in Eusebius’ narrative of Origen’s life and work. The Caesarean historian quotes a section from Porphyry’s treatise against the Christians in which the Neoplatonic philosopher asserted that Origen had been a ‘hearer’ (ἀκροατής) of an Ammonius who was renowned for his philosophical learning. Porphyry then used Ammonius as a (in his view) positive contrast with Origen: whereas Ammonius began life as a Christian and gave up his faith to learn philosophy, Origen received philosophical training, but turned his back on it to live as a Christian. In response to this extract from Porphyry, Eusebius asserts that Origen was in fact a Christian from his youth, and that Ammonius remained a Christian until the end of his life, as evidenced by his many works that were still in circulation, such as On the Harmony of Moses and Jesus.⁷ There are three main options for interpreting this passage. Probably the most common view in scholarship has been to interpret Porphyry’s statement to mean that Origen was a student of the Platonist Ammonius Saccas, who also taught Plotinus, Porphyry’s own instructor, and that Eusebius simply confused a Christian named Ammonius with the non-Christian Platonic philosopher. If this line of thinking is correct, then there is little more we can say about our Ammonius in the way of a more precise date for his flourishing. Theodor Zahn, representing this position, observed that Eusebius spoke of Ammonius as someone who had neither died recently nor been in the distant past, and so placed his literary activity in the years 240–280 CE, making him a younger contemporary of Origen, who died in the mid-250s.⁸ Photius claims that an Ammonius served as bishop of Thmuis and was visited by Origen, and this figure could have been the Christian Ammonius whom

⁷ Eusebius, HE 6.19.1 10. The passage cited here by Eusebius is fr. 39 in von Harnack’s edition of Porphyry’s treatise, fr. 20 in Berchman’s translation, and fr. 6F in Becker’s new edition. ⁸ Zahn, ‘Der Exeget Ammonius’, pp. 4 5. So also Ronald E. Heine, Origen: Scholarship in the Service of the Church (Oxford: Oxford University Press, 2010), pp. 24 5, who accepted that the Christian and pagan Ammonii were distinct individuals, and that Eusebius incorrectly attributed the Ammonian Christian writings to the Platonist Ammonius Saccas. The same position is taken by Richard Goulet, ‘Ammonios dit Saccas’, in Dictionnaire des philosophes antiques, vol. 1, ed. Richard Goulet (Paris: Éditions du Centre national de la recherche scientifique, 1989), pp. 165 8. On this issue see also the remarks of Becker, Contra Christianos, p. 167, and for an overview of scholarship on Ammonius Saccas, see Frederic M. Schroeder, ‘Ammonius Saccas’, Aufstieg und Niedergang der römischen Welt 2.36.1 (1987): pp. 493 526.

The Origins of Scholarship on the Fourfold Gospel

61

Eusebius confused with the philosopher, though this theory is not without problems.⁹ If, however, Eusebius was correct in supposing that the Ammonius who Porphyry says was Origen’s philosophical teacher also composed these two works of a Christian character, then there are at least two further possibilities. It may be, as Elizabeth DePalma Digeser has recently argued, that Ammonius Saccas himself dabbled in Christian topics and so was responsible for the Diatessaron-Gospel, though no other ancient sources make any mention of such literary activities.¹⁰ In her reading, the two named Christian works of Ammonius coincide well with later reports that attribute to Ammonius Saccas the achievement of harmonizing Plato and Aristotle. The final solution has been argued most clearly by Mark Edwards, who pointed out that, in addition to Ammonius Saccas the Platonist, there is also attestation of a further Ammonius who lived around this same time and was regarded as a Peripatetic. The Peripatetic Ammonius, who was praised by Longinus as one of the two ‘most erudite men of their epoch’,¹¹ must have flourished in the last decades of

⁹ The relevant passage from Photius can be found in J. Döllinger, Hippolytus und Kallistus (Regensburg: G. J. Manz, 1853), pp. 264 5. Cf. Adolf von Harnack, Die Chronologie der altchristlichen Litteratur bis Eusebius, Zweiter Band (Leipzig: J. C. Hinrichs, 1904), pp. 81 3; Otto Bardenhewer, Geschichte der altkirchlichen Literatur. Zweiter Band: Vom Ende des zweiten Jahrhunderts bis zum Beginn des vierten Jahrhunderts (Freiburg im Breisgau: Herdersche Ver lagshandlung, 1914), pp. 198 202. ¹⁰ Elizabeth DePalma Digeser, A Threat to Public Piety: Christians, Platonists, and the Great Persecution (Ithaca, NY: Cornell University Press, 2012), pp. 23 48. Digeser asserted that scholars have long been misled into thinking there must have been two Ammonii by the mistaken assumption that Christians and philosophers were two separate groups in antiquity (see esp. her comments on pp. 27 8). However, scholars have not been as blinded as she implies. See, e.g., the short article from over half a century ago by H. Langerbeck, ‘The Philosophy of Ammonius Saccas and the Connection of Aristotelian and Christian Elements Therein’, The Journal of Hellenic Studies 77 (1957): pp. 67 74, which questioned the assumption that Plotinus, as a non Christian, would not have studied Origen’s writings. Moreover, Digeser’s curt dismissal of the alternative view fails to take account of the historical arguments brought forward by Mark Edwards for his position (see n. 12). Ilaria Ramelli similarly implied that Ammonius Saccas wrote On the Harmony of Moses and Jesus, though she did not comment upon the Diatessaron Gospel (Ilaria Ramelli, ‘Origen, Patristic Philosophy, and Christian Platonism: Re Thinking the Christianisation of Hellenism’, Vigiliae Christianae 63 (2009): p. 226). As supporting evidence she pointed out that the Middle Platonist and Neo Pythagorean philosopher Numenius, though not a Christian, engaged in allegorical exegesis of the Old and New Testaments. ¹¹ See the passage from Longinus that Porphyry cites in vit. 20, 49 51 (Henry and Schwyzer, p. 25), which lists two Peripatetics: Ammonius and Ptolemy. Digeser wanted to identify the Platonist and Peripatetic Ammonii, both of whom are mentioned in this single passage from Longinus, in view of the fact that Ammonius Saccas is said by later Neoplatonists to have harmonized the teachings of Plato and Aristotle (Digeser, A Threat to Public Piety, pp. 28 30). This, however, seems unlikely, since the more straightforward reading of the passage from Longinus is that he has in mind two distinct individuals, mentioned as they are in close context but in separate categories, without any explicit indication that the two individuals are one and the same. The Aristotelian Ammonius is also mentioned by Philostratus, vit. 2.27 (LCL 134, p. 290), who praised him in terms similar to Longinus’.

62

The Eusebian Canon Tables

the second century, and so would have been an older contemporary of Origen who could have served as his teacher.¹² It is beyond the scope of this chapter to settle this debate, if it is even possible to do so with a satisfying degree of certainty.¹³ Moreover, on any of the above solutions the main conclusions of this chapter should hold true, since any of the proposed Ammonii would have been a contemporary of Origen, and, as I shall argue below, Origen’s Hexapla provides us with the closest parallel for Ammonius’ Diatessaron-Gospel, illuminating the format and the scholarly context of this work. For my purposes the question then becomes one of priority. If Eusebius confused two distinct individuals, then the Christian Ammonius was perhaps later than Origen, and the DiatessaronGospel may have been modelled on the earlier Hexapla. If, on the other hand, Eusebius was correct that Origen’s instructor in philosophy also composed Christian works, then it is more likely that Ammonius’ scholarship on the gospels provided an impetus for Origen’s text-critical work. My own sympathies lie with Edwards’ position, particularly in the light of his observation that Eusebius had access to a great many more sources, especially about Origen’s life and career, than we ever will, and that we should trust his report unless there are good reasons not to do so.¹⁴ For this reason I incline to the view that Origen’s teacher composed the Diatessaron-Gospel, and will proceed on this basis. Moreover, though I need to delay the discussion

¹² Mark Edwards, ‘Ammonius, Teacher of Origen’, Journal of Ecclesiastical History 44 (1993): pp. 169 81; Mark J. Edwards, Origen Against Plato, Ashgate Studies in Philosophy & Theology in Late Antiquity (Aldershot: Ashgate, 2002), pp. 54 5. As perhaps the sole precursor to his view, Edwards pointed to Heinrich Dörrie, ‘Ammonios, der Lehrer Plotins’, Hermes 83 (1955): pp. 439 77. Edwards observed that in the relevant passage of the Ecclesiastical History neither Eusebius nor Porphyry assert that Origen’s Ammonius was Ammonius Saccas who taught Plotinus. John Dillon regarded the two Ammonii as distinct individuals, one a Peripatetic and one the Platonist teacher of Plotinus, though he held that the Christian Origen studied with the latter (John M. Dillon, The Middle Platonists, 80 B.C. to A.D. 220, Rev. ed. (Ithaca, NY: Cornell University Press, 1996), pp. 380 3). The debate over whether Origen was taught by the Platonist or the Aristotelian Ammonius (if indeed they are separate individuals) is, as one would expect, tied to ongoing attempts to isolate the philosophical sources for various aspects of Origen’s thought. Moreover, it is related to a further debate over whether there was one or two figures named Origen. On the latter view there was a Christian Origen as well as a Platonist Origen, the latter being a younger contemporary of the former. Ilaria Ramelli has argued for a single Origen (Ramelli, ‘Origen, Patristic Philosophy’, pp. 235 44; Ilaria Ramelli, ‘Origen the Christian Middle/ Neoplatonist: New Arguments for a Possible Identification’, Journal of Early Christian History 1 (2011): pp. 98 130), while Mark Edwards holds to the existence of two Origens, as does Dillon, The Middle Platonists, pp. 380 3, referring to this as the ‘customary’ view. On this question, see most recently the clarifying analysis in Mark Edwards, ‘One Origen or Two? The Status Quaestionis’, Symbolae Osloenses 89 (2015): pp. 81 103. ¹³ The Ammonii mentioned thus far are to be distinguished from a later fifth century exegete with the same name who left behind exegetical fragments in the catena tradition. Cf. Joseph Reuss, ‘Der Presbyter Ammonius von Alexandrien und sein Kommentar zum Johannes Evangelium’, Biblica 44 (1963): pp. 159 70. ¹⁴ Edwards, ‘Ammonius, Teacher of Origen’, p. 174.

The Origins of Scholarship on the Fourfold Gospel

63

slightly, we will see that the reported fame of the philological skills of the Peripatetic Ammonius coincides well with what we know about Origen’s Hexapla and the Diatessaron-Gospel, making him a more likely candidate for Eusebius’ source than the Platonist Ammonius Saccas, whose tradition of philosophy, represented preeminently by Plotinus, scorned such philological investigations as unphilosophical. If it seems implausible to some that a Peripatetic philosopher held in such high regard by Longinus also composed works focused on Christian texts, one needs only to recall that Porphyry himself, regardless of which Ammonius he had in mind, asserted that Origen’s philosophical teacher began his life as a Christian but later went on to attain ‘the greatest attainments in philosophy’. In other words, Eusebius’ contention that Origen’s teacher of philosophy composed Christian works cannot be construed as merely an ambit claim from a Christian author attempting to create a past more intellectually respectable than the one that existed, since Porphyry, the most famous ancient critic of Christianity, himself testifies to the one-time Christian faith of a teacher who was held in the highest regard by non-Christian philosophers like himself. Following Porphyry’s lead, we might then suppose that the two works of a Christian nature that we can ascribe to this Ammonius were composed early in his life before he forsook his childhood faith to live as an adherent of traditional Greek religion.¹⁵

Eusebius’ Description in the Letter to Carpianus It is significant that when Eusebius came to describe his system of Canon Tables, he did so by situating his project in the tradition going back to Ammonius. He could have drawn upon his predecessor’s work without acknowledging his intellectual debt to his forebear, as so often happened in antiquity. The fact that he did not do so was probably due to his genuine esteem for Ammonius’ accomplishment. Having found out himself how complicated this issue could be, the first thing Eusebius did when introducing his Canon Tables to the world was to tip his hat to the ‘hard work and study’ (ϕιλοπονίαν καὶ σπουδήν) exerted by Ammonius in his analysis of the gospels. He then provided a one-sentence summary of the work, which is our sole surviving description of Ammonius’ composition: He has left behind for us the Diatessaron Gospel, in which he placed alongside the [Gospel] according to Matthew the concordant sections from the other evangelists. ¹⁵ On this point it is worth observing that Eusebius does not have a very effective reply to Porphyry’s claim that Ammonius forsook his Christian faith. In essence, Eusebius’ argument is that we know he remained a Christian until the end of his life because he wrote Christian books that are still read by the faithful. But this of course is a non sequitur.

64

The Eusebian Canon Tables

τὸ διὰ τεσσάρων ἡμῖν καταλέλοιπεν εὐαγγέλιον, τῷ κατὰ Ματθαῖον τὰς ὁμοϕώνους τῶν λοιπῶν εὐαγγελιστῶν περικοπὰς παραθείς¹⁶

The verb καταλείπω is one of the four standard introductory formulas Eusebius employed in his Ecclesiastical History to refer to the works of previous authors, and when used in conjunction with ἡμῖν, as it is here, it usually indicates that he himself had access to the work in question.¹⁷ This description is frustratingly brief, but is nevertheless sufficient to let the reader discern the main contours of Ammonius’ work, which must have been broadly akin to a modern gospel synopsis with parallel passages from the gospels presented in columns.¹⁸ The arrangement of this text in columns is indicated by the verb παρατίθημι, which was often used in antiquity to refer to a multi-columned work.¹⁹ The one other feature of the work that Eusebius clearly tells us is that the arrangement of the text gave priority to the first gospel—Ammonius dissected the latter three gospels in order to align the parallels he found there with corresponding passages in Matthew. Thus, some Matthean passages would presumably have had corresponding material in three parallel columns (representing Mark, Luke, and John), but many would have included text in fewer columns, probably leaving the columns empty when there was no related material from a given gospel.²⁰ Ammonius’ choice of Matthew as his base text is notable and must have been a deliberate decision, probably related to Matthew’s position at the head of the fourfold gospel canon. Eusebius does not tell us what Ammonius did with passages from Mark, Luke, and John that had no correlates in Matthew, so any conclusions drawn

¹⁶ ep. Carp. (NA²⁸, p. 89*). For a full translation of the letter, see Appendix 1. ¹⁷ Carriker, The Library of Eusebius of Caesarea, pp. 56 8. ¹⁸ So also Theodor Zahn, Tatian’s Diatessaron, Forschungen zur Geschichte des neutesta mentlichen Kanons und der altkirchlichen Literatur, Tl. 1 (Erlangen: Deichert, 1881), p. 33, and, again, Zahn, ‘Der Exeget Ammonius’, pp. 6 7, who rightly pointed out that Ammonius made, not a gospel harmony, but a gospel synopsis. Confusion over this issue began as early as the sixth century with Victor of Capua, who supposed that the works of Tatian and Ammonius were similar (see his praef. (Ranke, p. 1)). Zahn perceptively noted that whereas Eusebius uses the verb συντίθημι (‘combine, compose’) in HE 4.29.6 (SC 31, p. 214) to describe Tatian’s composition, he here uses παρατίθημι (‘place alongside’) for Ammonius’ undertaking. ¹⁹ See, e.g., Ptolemy, geo. 2.1.3 (Grasshoff and Stückelberger, vol. 1, p. 138). Other examples are adduced in Mansfeld and Runia, Aëtiana I, pp. 111 16. Similarly, the verb was also used to refer to the position of notes in the margin of a book (Nigel G. Wilson, ‘The Relation of Text and Commentary in Greek Books’, in Il Libro e il testo: atti del convegno internazionale: Urbino, 20 3 settembre 1982, ed. Cesare Questa and Renato Raffaelli (Urbino: Università degli studi di Urbino, 1984), p. 108). When describing Origen’s Hexapla Eusebius used the related verb ἀντιπαρατίθημι, which even more clearly expresses the mutli columned nature of that text (HE 6.16.4 (SC 41, p. 110)). On another occasion he used ἀντιπαρατίθημι to refer to his own Χρονικοὶ Κανόνες. See ecl. 1.1 (Gaisford, pp. 1 2). ²⁰ Similarly, Anthony Grafton and Megan Hale Williams, Christianity and the Transform ation of the Book: Origen, Eusebius, and the Library of Caesarea (Cambridge, MA: Belknap Press of Harvard University Press, 2006), p. 88, suggested that Origen, in his Hexapla, probably left his columns empty when he had no text to serve as a parallel.

The Origins of Scholarship on the Fourfold Gospel

65

on this question are speculative. Adolf von Harnack suggested that the fact that Eusebius calls the work τὸ εὐαγγέλιον implies that the remaining material from these latter three gospels was probably also included.²¹ Moreover, Eusebius does not criticize Ammonius for leaving out this bulk of material, even though, as we shall see, he was critical of other aspects of his predecessor’s work. These two observations may suggest that this non-Matthean material was included, but, if so, it is difficult to say how he presented it. All that is certain is that, if this other textual content was included, it must not have interrupted the continuous flow of Matthew’s text, since Eusebius claims only that the narrative sequences of the latter three were disrupted. We should linger for a moment over the title Eusebius gives for Ammonius’ composition. In his brief account in De viris Jerome calls it the Euangelici canones, but his account is clearly derivative from that of Eusebius so it seems unlikely that he had actually seen Ammonius’ composition.²² Instead, he was clearly borrowing the title of Eusebius’ own Canon Tables and applying it retrospectively to Ammonius’ earlier work. In contrast, it is quite likely that in this passage from the Letter to Carpianus Eusebius gives us the actual title that originated with Ammonius: τὸ διὰ τεσσάρων εὐαγγέλιον. Of course, this is the same title that Eusebius also gives for the more famous composition of Tatian, the so-called Diatessaron. I will say more about Tatian’s work shortly. For now we should consider how this title might relate to the work of Ammonius. Interpreting the title of Ammonius’ composition largely centers on how one should understand the preposition διά. Here there are at least three possibilities. First, it has long been supposed that the phrase διὰ τεσσάρων alludes to classical musical theory, specifically the interval of a fourth, one of the various possible συμϕωνίαι.²³ Though somewhat late, Boethius is a good representative of this tradition, referring to the symphonia diatessaron, quae

²¹ Adolf von Harnack, Geschichte der altchristlichen Litteratur bis Eusebius, Erster Theil (Leipzig: J. C. Hinrichs, 1893), pp. 406 7. Harnack was disagreeing with Zahn, Tatian’s Diates saron, p. 33, who had asserted that Ammonius did not include material from the latter three gospels that lacked a Matthean parallel. However, Zahn later changed his position in Zahn, ‘Der Exeget Ammonius’, 7, where he suggested that Ammonius left large gaps in his column of text from Matthew to allow the material from the other gospels without Matthean parallels to be displayed appropriately. Mansfeld and Runia, Aëtiana I, p. 115, followed Nordenfalk, Die spätantiken Kanontafeln, pp. 46 7, in supposing that the non Mathean material was omitted. Watson, The Fourfold Gospel, pp. 116 23, argued that it was included. ²² Jerome, vir. 55 (Richardson, pp. 33 4). ²³ For a survey of this material, see Paul A. Underwood, ‘The Fountain of Life in Manuscripts of the Gospels’, Dumbarton Oaks Papers 5 (1950): pp. 119 22. For further discussion on this idea, including an overview of older scholarship, see William L. Petersen, Tatian’s Diatessaron: Its Creation, Dissemination, Significance, and History in Scholarship, VCSup 25 (Leiden: E. J. Brill, 1994), pp. 49 51. As pointed out by Petersen, the fact that Victor of Capua called the gospel harmony in Codex Fuldensis a Diapente rather than a Diatessaron has also given rise to speculation about whether Victor had in mind such musical connotations.

66

The Eusebian Canon Tables

princeps est.²⁴ If we recall that Ammonius, according to Eusebius’ history, also composed a work aimed at demonstrating the συμϕωνία between Moses and Jesus, the possibility of a musical background for the phrase διὰ τεσσάρων is strengthened. There are, however, at least two other possibilities that must be considered. A second explanation comes from the fact that some sources attest the use of διά to indicate the material out of which something is made.²⁵ For example, Diodorus Siculus speaks of ‘images made from ivory and gold’ (εἴδωλα δι᾽ ἐλέϕαντος καὶ χρυσοῦ), and Plutarch mentions sacrifices ‘made with flour, drink-offerings, and the least costly gifts’ (δι’ ἀλϕίτου καὶ σπονδῆς καὶ τῶν εὐτελεστάτων πεποιημέναι).²⁶ In keeping with these parallels, the title τὸ διὰ τεσσάρων εὐαγγέλιον could imply that the four gospels were Ammonius’ source material, and the result of his editorial labour was a εὐαγγέλιον constructed from these four parts. A third possibility is that διά here might be referring, not to Ammonius’ source materials, but rather to some formal characteristic of his work. For example, Athenaeus in the third century noted that Timachidas of Rhodes wrote a treatise on banquets ‘in epic verse (δι’ ἐπῶν) in eleven or possibly more, books’.²⁷ Here διά indicates not the source of Timachidas’ work, but rather its format or style of composition. Similarly, Eleanor Dickey pointed out that discussions of spelling in antiquity ‘normally use the formula διά + genitive’, such as in the phrase διὰ τοῦ α γράϕεται which means ‘it is written with an α’.²⁸ Here the comparison with Origen’s Hexapla becomes relevant, because one of the best surviving descriptions of the work uses διά in this sense. As is well known, the Hexapla consisted of between six and eight texts arranged in parallel columns, including the Hebrew text of the Old Testament, a Greek transliteration of the Hebrew text, and as many Greek translations as Origen had available for any given book, using the Septuagint, Aquila, Symmachus, and Theodotion as a core. Two surviving fragments of the Hexapla for the psalms have confirmed this layout of the work, and they show that Origen allowed only one word per line, maximizing the potential for comparative analysis.²⁹ Origen’s Hexapla is justly famous and known primarily because of its importance for textual criticism of the Old Testament, and also

²⁴ Boethius, arith. 2.48 (CCSL 94A, p. 200). Cf. Ammonius’ likely contemporary, the Pyrrho nian Skeptic Sextus Empiricus, who referred in passing to ‘the harmony of the fourth in music’ (ἐν μὲν μουσικῇ τῆς διὰ τεσσάρων συμϕωνίας) (math. 1.77 (Mau and Mutschmann, p. 616)). ²⁵ LSJ, s.v. διά, A.III.2. ²⁶ Diodorus Siculus, bibl. hist. 17.115.1 (Fischer, p. 310); Plutarch, Num. 8.8 (LCL 46, p. 334). Cf. Athenaeus, deip. 14.56 (βρώματα διὰ μέλιτος καὶ γάλακτος γινόμενα; Kaibel, vol. 3, p. 429). ²⁷ Athenaeus, deip. 1.8 (Kaibel, vol. 1, p. 10). ²⁸ Dickey, Ancient Greek Scholarship, p. 118. ²⁹ See the images at Grafton and Williams, Christianity and the Transformation, pp. 97, 99.

The Origins of Scholarship on the Fourfold Gospel

67

for its enormous scholarly achievement, given that the complete work may have filled nearly forty codices of 400 folios each.³⁰ By calling his work the Ἑξαπλᾶ, Origen was highlighting its most distinctive feature, namely its format, consisting of six parallel columns. As Rufinus stated, ‘on account of this manner of composition, he [i.e., Origen] called the exemplar itself Hexapla, which means “written in sixfold order”’ (propter huiuscemodi compositionem exemplaria ipsa nominauit Ἑξαπλᾶ, id est sextiplici ordine scripta).³¹ A similar passage that is even more important for my argument is found in the description of the Hexapla given by Epiphanius in his Panarion. After listing the Greek versions used by Origen, Epiphanius noted that the Alexandrian had included the Hebrew text in Hebrew characters. Then, ‘using a second, parallel column opposite [the first]’ (ἐκ παραλλήλου δὲ ἄντικρυς, δευτέρᾳ σελίδι χρώμενος), he placed the Hebrew words, though ‘in Greek letters’ (δι’ Ἑλληνικῶν δὲ [τῶν] γραμμάτων). The result was that these [books] were in fact, and were called, Hexapla, since in addition to the [four] Greek translations there were two additional juxtaposed [columns]: Heb rew in the natural manner with Hebrew letters and Hebrew with Greek letters, such that the entire Old Testament was in a sixfold form, being so called due to the two [columns] of Hebrew words ὡς εἶναι μὲν ταῦτα καὶ καλεῖσθαι Ἑξαπλᾶ, ἐπὶ τὰς Ἑλληνικὰς ἑρμηνείας δύο ὁμοῦ παραθέσεις, Ἑβραϊκῆς ϕύσει δι’ στοιχείων καὶ Ἑβραϊκῆς δι’ Ἑλληνικῶν στοιχείων, ὥστε εἶναι τὴν πᾶσαν παλαιὰν διαθήκην δι’ ἑξαπλῶν καλουμένων καὶ διὰ τῶν δύο τῶν Ἑβραϊκῶν ῥημάτων.³²

Note first the use of the term παραθέσις (‘juxtaposition’), cognate to the verb παρατίθημι used by Eusebius to describe how Ammonius placed passages from the gospels alongside one another. In addition, the usage of διά here provides the clearest parallel for the function of the preposition in Ammonius’ title. Epiphanius uses it three times with reference to the characters in which the text is written, either ‘with Hebrew letters’ or ‘with Greek letters’. Here the ³⁰ Grafton and Williams, Christianity and the Transformation, 105, pp. 322 3 n. 35, who argue that the figure of 6,500 total pages (offered by Henry Barclay Swete, An Introduction to the Old Testament in Greek (Cambridge: The University Press, 1900), p. 74) is an underestimation, since it does not take into account Origen’s decision to write one Hebrew word per line. Cf. Heine, Origen, pp. 73 6. ³¹ Rufinus, HE 6.16.4 (GCS 9.2, p. 555). As noted by Grafton and Williams, Christianity and the Transformation, pp. 94 5, although Rufinus’ work was a translation of Eusebius’ Ecclesias tical History, here he diverged from his source by giving greater detail. ³² Epiphanius, pan. 64.3.5 7 (GCS 31, pp. 407 8). The Greek text is also cited, with transla tion at Grafton and Williams, Christianity and the Transformation, 92 3, p. 318, n. 12. I have followed the translation of Grafton and Williams, with some modifications. See also the slightly fuller description given by Epiphanius at mens. 510 35, where he again explains the name Hexapla as resulting from the six juxtaposed σελίδες in which the text was presented. The Greek text of the latter passage is cited, with English translation, at Grafton and Williams, Christianity and the Transformation, pp. 318 20, n. 13.

68

The Eusebian Canon Tables

sense of διά alludes to the format or presentation of the text in these columns, a usage in keeping with the third option mentioned above. Then, drawing his summary to a close, he refers to the resulting six-column format of Origen’s work with the phrase δι’ ἑξαπλῶν, a striking parallel to Ammonius’ διὰ τεσσάρων. Additionally, this passage also calls to mind the remainder of Ammonius’ title: τὸ . . . εὐαγγέλιον. Just as Ammonius put τὸ εὐαγγέλιον in the form of διὰ τεσσάρων, so also Origen put ἡ παλαιὰ διαθήκη in the form of δι’ Ἑξαπλῶν. In both cases, the διά clause refers to the format of the work, while the rest of the title refers to its content. The most significant difference between the title of Origen and that of Ammonius is that Origen uses the compound form ἑξαπλοῦς (‘sixfold’) from ἕξ + ἁπλόος and cognate to ἐξαπλόω (‘to multiply by six’).³³ In contrast, Ammonius’ title uses the simple cardinal form τέσσαρες. Despite this minor difference, the usage of διά plus a number to describe the format of Origen’s work is the most suitable parallel for the sense of the preposition in Ammonius’ title. On this reading, the διά of Ammonius’ title refers not to his four source texts, as one might assume, but rather to the four-column format in which he presented these texts, just as Origen’s title drew attention to the unusual multi-column format of the Hexapla. If this is correct, it makes it difficult to settle on a title in English that adequately captures the sense of the Greek. The closest equivalent might be ‘The Four-Columned Gospel’ or ‘The Gospel in Four Columns’, but perhaps the best way to refer to Ammonius’ composition is simply by transliterating it as we do with the Hexapla, calling it the Diatessaron-Gospel. There is one further passage, highlighted nearly a century ago by Theodor Zahn, that must also be considered to round out this investigation. In book five of his Commentary on John, Origen, while refuting the Marcionite error, argued that as he is one whom the many preach, so the gospel recorded by the many is one in its meaning, and there is truly one gospel through the four. ὡς εἷς ἐστιν ὃν εὐαγγελίζονται πλείονες, οὕτως ἕν ἐστι τῇ δυνάμει τὸ ὑπὸ τῶν πολλῶν εὐαγγέλιον ἀναγεγραμμένον καὶ τὸ ἀληθῶς διὰ τεσσάρων ἕν ἐστιν εὐαγγέλιον.³⁴

³³ This compound form may derive from ancient library traditions. The twelfth century Byzantine scholar Joannes Tzetzes reported that the library of Alexandria consisted of 400,000 ‘mixed books’ (βίβλων συμμιγῶν) and 90,000 ‘simple and unmixed books’ (ἁπλῶν . . . καὶ ἀμιγῶν βίβλων) (Arist. 2 (Koster, p. 32)), and Plutarch claimed that the libraries of Pergamum contained two hundred thousand ‘simple books’ (βιβλίων ἁπλῶν) (Ant. 58 (LCL 101, p. 270)). In these references ἁπλόος seems to mean scrolls containing only a single author, or perhaps only a single work. The compound form ἑξαπλοῦς, then, would imply a work comprising six components. On these two passages from Plutarch and Tzetzes, see Eva Nyström, Containing Multitudes: Codex Upsaliensis Graecus 8 in Perspective, Acta Universitatis Upsaliensis, Studia Byzantina Upsaliensia 11 (Uppsala: Uppsala Universitet, 2009), p. 39. ³⁴ Origen, Jo. 5.7 (SC 120, pp. 386 8).

The Origins of Scholarship on the Fourfold Gospel

69

Zahn was right to argue that the unusualness of the phrase τὸ διὰ τεσσάρων εὐαγγέλιον makes it unlikely that Origen’s statement here has no relation to the other usages of the phrase in antiquity.³⁵ Nevertheless, the fact that Origen used the phrase in passing, without giving it any sustained attention or attributing it to any other source, makes it difficult to interpret his usage. Zahn supposed that Origen had in mind the so-called ‘Diatessaron’ of Tatian, and that the adverb ἀληθῶς was intended as a polemical contrast with that earlier work. However, given that Origen nowhere else gives any indication of knowing Tatian’s gospel, this seems unlikely. Another explanation is suggested by the book culture of Origen’s day. When codices containing all four gospels began to be produced, which occurred by the mid-third century at the latest and so within Origen’s lifetime,³⁶ the phrase τὸ διὰ τεσσάρων εὐαγγέλιον would certainly have been a suitable description for such a book, referring to the one gospel that proceeds ‘through’ the four separate, consecutive versions. In this case, the phrase could represent an extension of Ammonius’ usage, and would still be a reference to a distinctive format of a book, although now referring to four successive versions, rather than four simultaneously parallel texts. This, of course, is assuming that Ammonius’ work was prior to or at least contemporary with that of Origen, either of which would be compatible with any of the Ammonii proposed above. At this point someone will probably object that Tatian’s usage of the same title for his work undermines the preceding argument, since his edition of the gospel contained neither parallel columns nor multiple sequential texts, but simply a single, continuous narrative. However, this objection only applies if Eusebius was correct in calling Tatian’s work the ‘Diatessaron’.³⁷ In fact, I have argued elsewhere, based on the earliest Syriac evidence, that Tatian most likely called his text simply ‘The Gospel’, and that Eusebius is not to be trusted in this instance, especially since he himself implies that he had never seen a copy of Tatian’s edition.³⁸ If so, then the title τὸ διὰ τεσσάρων εὐαγγέλιον quite likely ³⁵ Zahn, ‘Der Exeget Ammonius’, pp. 5 6. Cf. Zahn, Tatian’s Diatessaron, p. 34. ³⁶ The earliest undisputed papyrological evidence for a four gospel codex is P⁴⁵, usually dated c.250. T. C. Skeat argued that P⁶⁴, P⁶⁷, and P⁴ together once formed a four gospel codex (T. C. Skeat, ‘The Oldest Manuscript of the Four Gospels?’, New Testament Studies 43 (1997): pp. 1 34). However, see the response to Skeat’s proposal in Peter M. Head, ‘Is P⁴, P⁶⁴ and P⁶⁷ the Oldest Manuscript of the Four Gospels? A Response to T. C. Skeat’, New Testament Studies 51 (2005): pp. 450 7, and a survey of the matter in Hurtado, The Earliest Christian Artifacts, pp. 36 7. The manuscript designated as P⁷⁵ comes from roughly the same time period and contains portions of Luke and John. ³⁷ At HE 4.29.6 (SC 31, p. 214) Eusebius stated that Tatian called his work τὸ διὰ τεσσάρων [εὐαγγέλιον]. The εὐαγγέλιον in brackets is an emendation to the text suggested by Petersen, Tatian’s Diatessaron, 37, in view of the fact that the word ‘gospel’ appears in both the Latin and the Syriac translations of Eusebius’ history. ³⁸ Matthew R. Crawford, ‘Diatessaron, a Misnomer? The Evidence From Ephrem’s Com mentary’, Early Christianity 4 (2013): pp. 362 85. Eusebius’ lack of first hand knowledge of Tatian’s gospel was also recognized by Carriker, The Library of Eusebius of Caesarea, pp. 259 60.

70

The Eusebian Canon Tables

originated with Ammonius and later came to be attached erroneously to Tatian’s very different editorial work, with the influence of Eusebius ensuring that this confusion would become dominant in the later tradition. In fact, it is probably because the title ‘Diatessaron’ has been traditionally associated primarily with Tatian’s work that scholars have not previously considered the possibility that the parallel with the Hexapla might shed light on the meaning of the phrase as it relates to Ammonius’ gospel synopsis.

Alexandrian Scholarly Traditions and Ammonius’ Work Before leaving Ammonius’ work we should pause to consider what sort of intellectual milieu is implied by the projects of Origen’s Hexapla and Ammonius’ Diatessaron-Gospel. One of the recent advances in the study of Origen’s corpus is the recognition of the importance of classical philology to his intellectual endeavours. The work of Bernhard Neuschäfer was pioneering in this respect and has been followed by many since.³⁹ Neuschäfer drew attention to the fact that it was Alexandrian philology, which had been developed for the study of Homer, that provided the tools necessary for Origen’s creation of the Hexapla. Following the lead of literary critics like Zenodotus and Aristarchus, Origen called his text-critical work an exercise in διόρθωσις, since he, like his predecessors, engaged in the comparative analysis of rival versions in an attempt to establish an authoritative text.⁴⁰ More recently Anthony Grafton and Megan Williams have emphasized that Origen was not simply appropriating the tools of Hellenistic philology for the church. Rather, he was on the cutting edge of scholarship, since no other classical text existed in as many versions as the Hebrew scriptures, and no other philological undertaking was executed on such a grand scale as his Hexapla. Thus, they conclude that the Hexapla was ‘one of the greatest single monuments of Roman scholarship, and the first serious product of the application to Christian culture of the tools of Greek philology and criticism’.⁴¹ Grafton and Williams were no doubt correct to emphasize the unprecedented scale of the project Origen embarked upon, but they failed to note that he may have had at least one significant precursor who carried out a similar project also based in Alexandrian philological scholarship. In a fragment of a letter ³⁹ Bernhard Neuschäfer, Origenes als Philologe, Schweizerische Beiträge zur Altertumswis senschaft 18/1 2 (Basel: Friedrich Reinhardt Verlag, 1987). Cf. Grafton and Williams, Chris tianity and the Transformation, pp. 86 132; Peter W. Martens, Origen and Scripture: The Contours of the Exegetical Life, OECS (Oxford: Oxford University Press, 2012), pp. 25 66. ⁴⁰ Neuschäfer, Origenes als Philologe, pp. 85 138. ⁴¹ Grafton and Williams, Christianity and the Transformation, p. 131. Cf. Hollerich, ‘Eusebius’, p. 632: ‘The creation of the Hexapla demonstrated that Christian scholarship was thoroughly conversant with the editing methods developed by pagan scholars at Alexandria and elsewhere’.

The Origins of Scholarship on the Fourfold Gospel

71

quoted by Eusebius, Origen justified his own interest in philosophy by pointing to the prior example of Heraclas, bishop of Alexandria, whom he found ‘with the teacher of philosophical studies’ (παρὰ τῷ διδασκάλῳ τῶν ϕιλοσόϕων μαθημάτων). Because Eusebius picks up in mid-letter, Origen does not state who this philosopher was, but his citation of this passage occurs in the context of his discussion of Porphyry’s statement about Origen studying with Ammonius. For this reason, Mark Edwards is surely right that Eusebius intends the reader to assume that the unnamed master with whom bishop Heraclas studied was the same Ammonius mentioned by Porphyry as also the teacher of Origen. In other words, he was the Ammonius who probably produced the Diatessaron-Gospel.⁴² The importance of this passage for my argument is that Origen goes on to say that, by studying with this philosopher, Heraclas ‘was constantly engaged in the philological criticism of the books of the Greeks, so far as he was able’ (βιβλία τε Ἑλλήνων κατὰ δύναμιν οὐ παύεται ϕιλολογῶν).⁴³ Similarly, if we accept Edwards’ argument that Origen’s teacher Ammonius was the Peripatetic Ammonius, it is striking that Longinus singled out this Ammonius precisely for his philological learning, calling him, along with a certain Ptolemy, ‘the most erudite (ϕιλολογώτατος) men of their epoch, Ammonius in particular, whose broad learning (πολυμαθίαν) was without parallel’.⁴⁴ The sense of the term ϕιλόλογος in this context can be illuminated by considering Plotinus’ disparaging assessment of Longinus himself as someone who was ‘a philologist but certainly no philosopher’ (ϕιλόλογος . . . ϕιλόσοϕος δὲ οὐδαμῶς).⁴⁵ The rationale behind Plotinus’ judgement becomes clear in the light of the list of Longinus’ works in the Suda, including ‘Difficulties in Homer, Whether Homer is a Philosopher, Homeric Problems and Solutions in Two Books, Things Contrary to History that the Grammarians Explain as Historical, On Words in Homer with Multiple Senses in Four Books, two publications on Attic diction arranged alphabetically, [and] the Lexicon of Antimachus and Heracleon’.⁴⁶ These are clearly works of a more philological ⁴² Edwards, ‘Ammonius, Teacher of Origen’, p. 171. So also Digeser, A Threat to Public Piety, pp. 43 4. ⁴³ Eusebius, HE 6.19.12 14 (SC 41, pp. 116 17). ⁴⁴ Porphyry, vit. 20, 49 52 (Henry and Schwyzer, p. 25), on which see Edwards, ‘Ammonius, Teacher of Origen’, pp. 179 80. For a full analysis of the lengthy fragment from Longinus in which this statement occurs, see Irmgard Männlein Robert, Longin, Philologe und Philosoph. Eine Interpretation der erhaltenen Zeugnisse, Beiträge zur Altertumskunde 143 (Berlin: De Gruyter, 2011), pp. 167 220, and for her discussion of this description of Ammonius, see pp. 192 4. She points out that, in contrast to Plotinus, Longinus thought it a compliment to call a philosopher a ϕιλόλογος, in keeping with his own intellectual interests. ⁴⁵ Porphyry, vit. 14, 19 20 (Henry and Schwyzer, p. 18). ⁴⁶ Suda, s.v. Λογγῖνος (Adler, entry 645). Ἀπορήματα Ὁμηρικά, Εἰ ϕιλόσοϕος Ὅμηρος, Προβλήματα Ὁμήρου καὶ λύσεις ἐν βιβλίοις βʹ, Τίνα παρὰ τὰς ἱστορίας οἱ γραμματικοὶ ὡς ἱστορικὰ ἐξηγοῦνται, Περὶ τῶν παρ’ Ὁμήρῳ πολλὰ σημαινουσῶν λέξεων δʹ, Ἀττικῶν λέξεων ἐκδόσεις βʹ, εἰσὶ δὲ κατὰ στοιχεῖον, Λέξεις Ἀντιμάχου καὶ Ἡρακλέωνος· καὶ ἄλλα πολλά. See also the description of

72

The Eusebian Canon Tables

bent which draw on the Aristarchan tradition of Homeric criticism emnating from Alexandria, and as such they explain what Plotinus had in mind when he called Longinus a ϕιλόλογος and, by extension, what Longinus himself meant when he referred to Ammonius as ϕιλολογώτατος. If this Ammonius was Origen’s tutor, as seems likely, then Origen’s application of Alexandrian literary scholarship to Christian sources was not completely unprecedented. Rather, Ammonius had already pioneered this approach, which his more famous pupil later deployed on a grander scale.⁴⁷ Scholarly concerns about the internal unity and coherence of a text had emerged in the third and second centuries BCE in the work of Zenodotus and Aristarchus on Homer, and Alexandrian Jewish scholars adopted their methods for the purpose of dealing with apparent contradictions in the Septuagint.⁴⁸ The projects of Ammonius and Origen were no doubt motivated at some level by a similar concern to avoid textual dissonance, though the scale of the problems they attempted to address was unparalleled. Just as the Hebrew scriptures existed in more editions than any other ancient text, and so required the development of new methods to handle this textual plurality, so also the fourfold gospel, consisting of four irreducibly distinct yet similar texts in a single corpus, was a situation without exact parallel in classical or Jewish sources, and it is therefore not surprising that it too elicited a cuttingedge response from the scholars of Alexandria.⁴⁹ Longinus in Eunapius, vit. 4.1.2 4.1.6, who similarly emphasizes his achievements in literary criticism. John Dillon echoed Plotinus’ sentiment, stating that Longinus ‘may fairly rank as the last “regular” Middle Platonist a most civilized and learned man, but not an original philosopher of any significance’ (Dillon, The Middle Platonists, p. 382). On the entry on Longinus in the Suda, see further Männlein Robert, Longin, Philologe und Philosoph, pp. 97 104. A rather speculative argu ment has been made that, after the crushing of Zenobia’s revolt and Longinus’ attendant execution, his library eventually made its way into Eusebius’ hands: Paul Kalligas, ‘Traces of Longinus’ Library in Eusebius’ Praeparatio Evangelica’, Classical Quarterly 51 (2001): pp. 584 98. ⁴⁷ Zahn, Tatian’s Diatessaron, 34; Zahn, ‘Der Exeget Ammonius’, pp. 7 8, also pointed out the similar format of Ammonius’ work and the Hexapla, though he thought that the Hexapla inspired Ammonius, rather than the other way around, as I am suggesting here. Similarly, Mansfeld and Runia, Aëtiana I, p. 115 said, ‘Ammonius may have been inspired by the example of his fellow Alexandrian Origen’. On the relation of these works to one another, and a consideration of the broader Greco Roman context from which they emerge, see the insightful study by Andrew M. Riggsby, ‘Learning the Language of God’, in Modes of Knowing and Ordering Knowledge in Early Christianity, Volume 1, ed. Lewis Ayres, Michael Champion, and Matthew R. Crawford (forthcoming). ⁴⁸ On concerns about contradictions and coherence in Homer, see Nünlist, The Ancient Critic, pp. 27 34, 157 64, 174 84. On the appropriation of these scholarly methods for Jewish biblical exegesis, see Maren Niehoff, Jewish Exegesis and Homeric Scholarship in Alexandria (Cambridge: Cambridge University Press, 2011), especially pp. 39 46, 118 29. Digeser, A Threat to Public Piety, pp. 35 7, stressed the concern for ‘harmony’ that characterizes Ammonius’ work, but she neglected Alexandrian philological scholarship as an important context in which such topics had long been debated. ⁴⁹ Hence I concur with the assessment of Watson, The Fourfold Gospel, p. 105, who argued, ‘More than any other part of the Old or New Testaments, the four gospels demand to be fitted out with cross references’.

The Origins of Scholarship on the Fourfold Gospel

73

The two most significant points of similarity in the projects of Ammonius and Origen are: first, the fact that both engaged in what was at root a comparative analysis of texts—for Ammonius comparison of variations on a single life story and for Origen comparison of multiple translations of the same text—and, secondly, that both recognized that parallel columns could greatly facilitate such comparative analysis. Recently Eleanor Dickey has drawn attention to a genre of ancient school texts that used parallel columns to show the translation of terms between Latin and Greek, and Andrew Riggsby has argued persuasively that this genre of literature must have informed Origen’s Hexapla.⁵⁰ Yet the use of parallel columns in this manner is otherwise very rarely attested in this period, and it can hardly be a coincidence that two Christian authors, both in Alexandria around the same time, realized that this pedagogical tool could be used on a much vaster scale as a powerful methodology for a scholarly research project. Ammonius and Origen, in other words, must have come from the same intellectual culture in which the methods of Alexandrian philological criticism were finding fresh vitality through their application to new corpora of texts.⁵¹ Hence, although Grafton and Williams may be correct that the Hexapla ‘spawned a range of imitations and adaptations intended for a variety of uses’ and that later Christian authors ‘attributed the whole tradition [of multicolumn Bibles] to Origen as its intellectual father’⁵², Origen in fact probably drew the inspiration for his work from the earlier literary scholarship of Ammonius who had already pioneered this format as a convenient way to analyse the relationships among the four gospels. It is therefore unfortunate that, while Origen is regularly and rightly lauded for his monumental contribution to Old Testament textual criticism with his Hexapla, Ammonius is not typically accorded the same respect when it comes to scholarship on the gospels. In fact an appreciation of the complexity of the similarities and differences amongst the gospels did not first emerge in the eighteenthcentury. Rather, educated Christians became aware of this issue perhaps not

⁵⁰ Eleanor Dickey, ‘Columnar Translation: An Ancient Interpretive Tool That the Romans Gave the Greeks’, Classical Quarterly 65 (2015): pp. 807 21; Riggsby, ‘Learning the Language of God’. On the purpose of the two columns in the Hexapla containing Hebrew, see Martin J. Martin, ‘Origen’s Theory of Language and the First Two Columns of the Hexapla’, Harvard Theological Review 97 (2004): pp. 99 106. ⁵¹ Edwards also highlighted Ammonius’ scholarship on the gospels as evidence for ‘the supremacy of the philological method in Alexandria’ (Edwards, Origen Against Plato, p. 18). ⁵² Grafton and Williams, Christianity and the Transformation, p. 114. On some of the less well known medieval polyglots that used multiple columns, perhaps under the influence of the Hexapla, see Ronny Vollandt, ‘The Conundrum of Scriptural Plurality: The Arabic Bible, Polyglots, and Medieval Predecessors of Biblical Criticism’, in The Text of the Hebrew Bible and Its Editions: Studies in Celebration of the Fifth Centennial of the Complutensian Polyglot, ed. Andrés Piquer Otero and Pablo Torijano Morales, Supplements to the Textual History of the Bible 1 (Leiden: Brill, 2016).

74

The Eusebian Canon Tables

more than a century after the last evangelist put down his pen, and the attempt to use state-of-the-art scholarly tools to better understand the problem is to be traced back to the Christian appropriation of philological scholarship in late-antique Alexandria.

E U S E B I U S’ CONTINUATION OF AMMONIUS ’ SCHOLARLY P ROJECT

Tabular Methods in Eusebius’ Broader Corpus Let us now consider how this early tradition of scholarship on the fourfold gospel, which began in the second or third century in Alexandria, was carried forward a century later by Eusebius in Caesarea. Despite the rarity of tables in the ancient world, the Canon Tables were not Eusebius’ only experimentation with this technology. Examining his other uses of tabular methods will situate his adaptation of Ammonius’ Diatessaron-Gospel within his broader corpus. This is an aspect of his work that has been explored most recently by Grafton and Williams, who aptly describe the Caesarean bishop as ‘a Christian impresario of the codex’,⁵³ thereby highlighting the fact that Eusebius’ innovations in information technology exploited the increasingly dominant late antique book form which was well on its way to replacing the traditional bookroll by the early fourth century. Scholars have long recognized that Christians in general adopted the codex earlier and more consistently than their non-Christian peers, particularly for sacared, liturgical texts,⁵⁴ so by Eusebius’ time there was a long-standing Christian preference for the codex over the bookroll. Eusebius, however, more than any other figure in late antiquity, seems to have recognized and exploited the possibilities afforded by the new book form.⁵⁵ In the previous chapter I highlighted the fact that most of the surviving papyrus fragments of Ptolemy’s Handy Tables came from codices, and suggested that this format was most likely employed because it better facilitated the back-and-forth kind of reading required to ⁵³ The title of chapter 4 of Grafton and Williams, Christianity and the Transformation. ⁵⁴ Though there is agreement on the fact of Christian preference for the codex, debate over its causes continues. For surveys of possible answers, see Hurtado, The Earliest Christian Artifacts, pp. 43 93; Wallraff, Kodex und Kanon, pp. 8 25, and for a recent influential proposal, see Roger S. Bagnall, Early Christian Books in Egypt (Princeton, NJ: Princeton University Press, 2009), pp. 70 90. For a recent study of the material aspects of the transmission of the gospels in the second and third centuries, see Scott D. Charlesworth, Early Christian Gospels: Their Production and Transmission, Papyrologica Florentina 47 (Firenze: Edizioni Gonnelli, 2016). ⁵⁵ Cf. Grafton and Williams, Christianity and the Transformation, p. 200: ‘No early creator of codices understood more vividly than Eusebius the possibilities that the new form of the book created for effective display of texts and information’.

The Origins of Scholarship on the Fourfold Gospel

75

use the book for calculations, a mode of interaction similar to that envisioned by the Canon Tables. As Jason König and Tim Whitmarsh have observed, ‘It is surely no coincidence that the earliest codices contained Christian and technical material, two genres of discourse that privilege, indeed insist upon, crossreferencing and non-linear reading’.⁵⁶It is worth emphasizing this point again here because the other Eusebian works to be discussed below encourage a similar kind of usage and so also are particularly suited to the codex format. At least two of Eusebius’ other works contained lists or tables that he described as κανόνες, in keeping with his usage of this term in the Letter to Carpianus. The most wide-ranging of these was his Chronicle, an ambitious attempt to condense Babylonian, Egyptian, Jewish, Greek, and Roman chronography into a single, coordinated timeline of world history. The first half of this two-part work consisted of a discussion of the sources and problems attendant on such an enterprise, and the second half, bearing the title Χρονικοὶ Κανόνες, boldly combined these sources into a unified tabular format allowing cross-referencing between various national histories.⁵⁷ His Chronicle was not entirely original, but rather borrowed ‘the content, structure, style, and historical approach of the Hellenistic chronicle’, a genre of literature that typically reported important events for a given city or kingdom which were arranged chronologically and dated according to Olympiads.⁵⁸ Eusebius, however, transformed this existing genre in two ways.⁵⁹ The first was the scope of his ⁵⁶ Jason König and Tim Whitmarsh, ‘Ordering Knowledge’, in Ordering Knowledge in the Roman Empire, ed. Jason König and Tim Whitmarsh (Cambridge: Cambridge University Press, 2007), p. 34. Cf. Chartier, Forms and Meanings, 19: ‘the codex undeniably facilitates organization and handling of the text. It permits pagination, the creation of indexes and concordances, and the comparison of one passage with another; better yet, it permits a reader to traverse an entire book by paging through’. On non linear reading, see also James J. O’Donnell, Avatars of the Word: From Papyrus to Cyberspace (Cambridge, MA: Harvard University Press, 1998), pp. 54 7. ⁵⁷ See Eusebius’ mention of the title of the work at HE 1.1.6; PE 10.9.11. For a discussion of Eusebius’ achievement with his Chronicle, see Barnes, Constantine and Eusebius, pp. 111 20; Grafton and Williams, Christianity and the Transformation, pp. 133 77; Burgess and Kulikowski, Mosaics of Time, pp. 119 26; Wallraff, Kodex und Kanon, pp. 30 1; Johnson, Eusebius, pp. 86 9; Riggsby, Mosaics of Knowledge, ‘Conclusion’. The textual tradition of Eusebius’ Chronicle is complex. While only fragments of the original Greek survive, the second half of the work was translated into Latin and supplemented by Jerome, and the full work survives in Armenian, though with some lacunae and corruption. For a study of these issues, see Richard W. Burgess and Witold Witakowski, Studies in Eusebian and Post Eusebian Chrono graphy, Historia, Einzelschriften 135 (Stuttgart: Franz Steiner, 1999). On the Armenian recep tion of the Chronicle, see also Armenuhi Drost Abgaryan, ‘The Reception of Eusebius of Caesarea (c.264 339) in Armenia’, in Greek Texts and Armenian Traditions: An Interdisciplinary Approach, ed. Francesca Gazzano, Lara Pagani, and Giusto Traina, Trends in Classics (Berlin: De Gruyter, 2016), pp. 215 29. ⁵⁸ Burgess and Kulikowski, Mosaics of Time, pp. 121 2. See, e.g., Plutarch’s reference to χρονικοῖς κανόσιν at Sol. 27.1 (LCL 46, p. 478). On the indebtedness of Eusebius’ Chronicle to the prior chronicle tradition, see also Brian Croke, ‘The Origins of the Christian World Chronicle’, in History and Historians in Late Antiquity, ed. Brian Croke and Alanna M. Emmett (Sydney: Pergamon Press, 1983), pp. 116 31. ⁵⁹ On these two points, see Burgess and Kulikowski, Mosaics of Time, p. 122.

76

The Eusebian Canon Tables

ambition. He recorded a total of fifteen distinct national chronologies,⁶⁰ which would have presented a bewildering array of competing dating systems, yet he managed to synchronize them all by situating the chronologies with reference to the birth of Abraham as a universal starting point and marking intervals of time subsequent to this point by the tabular rows that followed. The second innovation was his creative use of parallel columns to display this potentially overwhelming amount of information. Each vertical column in the Χρονικοὶ Κανόνες represents a single kingdom, so the number of columns varies depending on how many kingdoms were in existence at any given time. At most this meant that nine parallel columns had to be presented at once across a two-page spread, a complex format that represents an impressive achievement in terms of information visualization.⁶¹ This arrangement of information resulted in the creation of a table in the strict sense of the term (as discussed in the previous chapter), one that allowed the user to navigate either vertically or horizontally in open-ended exploration of the information contained therein. By reading vertically one could follow the individual history of a given kingdom diachronically, or by reading horizontally one could see what events were occurring simultaneously at any given point in time. This presentation of information across two axes was apparently without precedent in the prior tradition of Olympiad chronicles,⁶² and Eusebius himself highlighted the utility of such a format: And, lest the long series of numbers should lead to some sort of confusion, we have divided the whole accumulation of years into decades, which we gathered together from the individual histories of the nations and placed in turn opposite one another. The purpose of this was to provide a simple method of finding out (facilis praebeatur inuentio) at which time whether Greek or barbarian the prophets, kings, and priests of the Hebrews existed; likewise the time at which the gods of the various nations were falsely believed in, along with heroes; the time at which a city was founded; with respect to illustrious persons,

⁶⁰ In Eusebius’ words: ‘I gathered together all sorts of chronological records and the kingdoms of the Chaldaeans, Assyrians, Medes, Persians, Lydians, Hebrews, Egyptians, Athenians, Argives, Sicyonians, Spartans, Corinthians, Thessalians, Macedonians, and Latins (who were later called the Romans) fifteen altogether’. From Eusebius’ preface to the Χρονικοὶ Κανόνες, quoted in George Syncellus, chron. p. 73 (Mosshammer). Translation from appendix 2 of R. W. Burgess, ‘The Dates and Editions of Eusebius’ Chronici Canones and Historia Ecclesiastica’, Journal of Theological Studies 48 (1997): pp. 503 4. ⁶¹ Cf. Marchese, ‘Tables and Early Information Visualization’, pp. 40 1. ⁶² Eusebius’ innovation in this regard was defended by Rudolf Helm, Eusebius’ Chronik und ihre Tabellenform, Abhandlungen der preussischen Akademie der Wissenschaften, Philosophisch historische Klasse 1923, Nr. 4 (Berlin: 1924), and seems to have been subsequently uncontested: Alden A. Mosshammer, The Chronicle of Eusebius and Greek Chronographic Tradition (Lewisburg, PA: Bucknell University Press, 1979), p. 62; Barnes, Constantine and Eusebius, p. 120; Brian Croke, ‘The Originality of Eusebius’ Chronicle’, American Journal of Philology 103 (1982): pp. 195 200; Mansfeld and Runia, Aëtiana I, p. 113; Burgess and Witakowski, Studies in Eusebian and Post Eusebian Chronography, p. 83; Grafton and Williams, Christianity and the Transformation, pp. 172 3; Burgess and Kulikowski, Mosaics of Time, p. 123.

The Origins of Scholarship on the Fourfold Gospel

77

when philosophers, poets, princes, and writers of various works arose; and any other ancient occurrences that were thought to be worth remembering. All of this we will put in its proper place with the greatest brevity.⁶³

This passage is one of the clearest statements we have demonstrating Eusebius’ awareness of the impact of formal design on the use of a text. The goal he envisions is to be able to find what one is looking for with ease, and he recognizes that the best way to accomplish that goal is with a presentation that is as concise and clear as possible—the juxtaposition of diverse material in discrete, parallel columns synchronized by rows that mark off regular intervals. Hence, although the discipline of chronography was centuries old by the time Eusebius tried his hand at it,⁶⁴ his approach was innovative in the way in which it arranged the complex material he drew from his sources for the purpose of comparative analysis. This, of course, was precisely the aim of the mis-en-page of Origen’s Hexapla, and, given this similarity and Eusebius’ familiarity with the Hexapla, it seems undeniable that Eusebius’ innovation with the Chronicle was inspired by that earlier endeavor.⁶⁵ Nevertheless, the application of this technique to historical materials ‘represented a dramatic formal innovation’, which resulted in ‘a stunningly original work of scholarship’⁶⁶. Moreover, the transfer of this technology from the realm of textual scholarship, as it was used in the Hexapla, to the domain of chronological historiography is another remarkable sign of Eusebius’ ingenuity, given the paucity of such crossovers in the Greco-Roman world. Before leaving the Chronicle I want to examine one further passage in which Eusebius comments on the relation between it and his better-known Ecclesiastical History. At the outset of the Ecclesiastical History he explains how he has gone about writing the present work, what its aim is, and its significance as the first work of its kind. After expounding on such matters, he then comments, Previously I have already made a summary of these things in the chronological tables that I formed, but, however, in the present work I have undertaken to present the most complete narrative of these same matters.⁶⁷

⁶³ This passage survives as a preface to Jerome’s Latin translation of book two of the Chronicle. See Jerome, chron. praef. (GCS 47, pp. 18 19). In the preface to book one of the work, which only survives in Armenian, Eusebius states his intention to convert the material in book one into tables when he comes to book two, and says that the purpose for doing so is to facilitate ease of access, to make using it ‘easier and faster’ (դիւրաւ եւ պատրաստագոյնս) (Aucher, p. 9). ⁶⁴ This wider context is the focus of Burgess and Kulikowski, Mosaics of Time. ⁶⁵ So Barnes, Constantine and Eusebius, p. 120; Mansfeld and Runia, Aëtiana I, pp. 111 12 (with some qualifications); Grafton and Williams, Christianity and the Transformation, pp. 142 3; Burgess and Kulikowski, Mosaics of Time, p. 122. ⁶⁶ Grafton and Williams, Christianity and the Transformation, p. 175. ⁶⁷ Eusebius, HE 1.1.6 (SC 31, p. 5).

78

The Eusebian Canon Tables

ἤδη μὲν οὖν τούτων καὶ πρότερον ἐν οἷς διετυπωσάμην χρονικοῖς κανόσιν ἐπιτομὴν κατεστησάμην, πληρεστάτην δ’ οὖν ὅμως αὐτῶν ἐπὶ τοῦ παρόντος ὡρμήθην τὴν ἀϕήγησιν ποιήσασθαι.

Clearly this means that Eusebius wrote this passage of the Ecclesiastical History after he had finished book two of the Chronicle, containing his chronological canons. More importantly, the latter work served as a ‘summary’ of the narrative account extended over multiple books in the Ecclesiastical History. Indeed, the information contained in the latter portion of the chronological tables, especially the succession lists of bishops and the notices of the key events pertaining to important figures in the history of the church, would have served as an ideal skeletal framework upon which to construct a proper historical narrative.⁶⁸ The reason for drawing attention to this passage here is that it serves as a fitting inverse analogy for the way in which the Canon Tables relate to the text of the gospels. What this passage establishes is that in Eusebius’ mind a tabular matrix was a condensed summary of information that could be expanded upon in a different genre of writing. If so, then just as the chronological canons were an ἐπιτομή of the extended ἀϕήγησις in the Ecclesiastical History, so also by implication the Canon Tables were an ἐπιτομή of the four ἀϕηγήσεις of the gospels. The second Eusebian work that shows a similar experimentation with information visualization was much more restricted in scope and simpler in function, and was also less successful, surviving in only a single manuscript (Oxford, Bodleian Library, Auct. D.4.1, fol. 24v‒25r). In a recent publication Martin Wallraff has called this work the ‘Canon Tables of the Psalms’, though the title in the tenth-century Greek manuscript is rather different: Πίναξ ἐκτεθεὶς ὑπὸ Εὐσεβείου τοῦ Παμϕίλου.⁶⁹ Πίναξ does not have precisely the ⁶⁸ Cf. Burgess, ‘The Dates and Editions’, p. 498: ‘[Eusebius] simply took the bare outline that he had produced in tabular form and expanded it into a narrative, concentrating on individuals he had already named, those who had made the church great and those who had caused it to suffer, both from within and from without. Such a project required further detailed research, of course, and this explains many of the differences between the Canones and the HE’. The relationship between these two works is complex, as has been demonstrated by Robert M. Grant, Eusebius as Church Historian (Oxford: Clarendon Press, 1980), pp. 6 9, who high lighted some material included in the Chronicle but not brought over to the Ecclesiastical History, as well as divergences in the details about some topics treated in both. On the relation of the two works, see also T. D. Barnes, ‘Some Inconsistencies in Eusebius’, Journal of Theological Studies 35 (1984): p. 472. ⁶⁹ Martin Wallraff, ‘The Canon Tables of the Psalms: An Unknown Work of Eusebius of Caesarea’, Dumbarton Oaks Papers 67 (2014): pp. 1 14. Wallraff includes colour images of the sole surviving manuscript, as well as a transcription and translation on pp. 4 7. Attention was first drawn to this Eusebian work in Giovanni Mercati, Osservazioni a proemi del Salterio di Origene, Ippolito, Eusebio, Cirillo Alessandrino e altri, con frammenti inediti, ST 142 (Vatican City: Biblioteca apostolica vaticana, 1948), pp. 95 104, and it is also mentioned in M. J. Rondeau, Les commentaires patristiques du Psautier (IIIe Ve siècles) (Rome: Pont. Institu tum Studiorum Orientalium, 1982 1985), I.71 2; Grafton and Williams, Christianity and the Transformation, 198 9; Wallraff, Kodex und Kanon, pp. 31 2.

The Origins of Scholarship on the Fourfold Gospel

79

same semantic range as κανών and seems to refer to a ‘list’ in contrast to κανών in the sense of a tabular matrix, as used by Ptolemy. For example, in his Ecclesiastical History, Eusebius at one point mentions the πίνακες drawn up by Pamphilus for the library in Caesarea, referring to the catalogues of authors and works contained therein.⁷⁰ This sense of the word goes back at least to Callimachus (c.305–c.240 BCE), who titled his 120-book catalogue of Greek authors and works in the Alexandrian library the Πίνακες.⁷¹ Since the Eusebian work in question is a list of texts arranged according to authorship, the word Pinax is an apt title, in keeping with the Callimachan tradition and the work of Pamphilus and Eusebius in cataloguing the Caesarean library. The usage of this term may also be related to the fact that this Pinax is not a true tabular matrix with vertical columns and horizontal rows, but a more simple collection of vertical lists. However, in the sole surviving copy each of these lists is called a κανών, so perhaps the sense of the two terms overlapped in Eusebius’ understanding. As suggested by Wallraff ’s title, this work was focused on the psalter and was an attempt to classify the one hundred and fifty psalms contained in it according to authorship. Eusebius accomplished this by producing seven κανόνες, each of which listed the psalms attributed to a given author: psalms of David, psalms of Solomon, unlabelled psalms, psalms of the sons of Korah, psalms of Asaph, anonymous psalms, and Hallelujah psalms, with two further categories containing only a single psalm each, attributed to Ethan and Moses. Because each of these categories of material is in no way cross-referenced with any other, there is no need to read horizontally across the κανόνες. Rather, one simply uses the list to identify all of the psalms that belong to a given category, which can then be explored in whatever order one would like. Though less complex than the Canon Tables for the gospels, this work presents at least two parallels to Eusebius’ more successful paratext. The first is that the Pinax for the Psalms is an attempt at providing a reader with an improved navigation of a text through the classification of portions of that text into discrete categories, precisely the same goal and method as the Canon Tables apply to the gospels. The second is that, in the sole surviving copy of the Pinax, the κανόνες are housed within architectural frames, which, if original, lend support to the hypothesis that the architectural frames usually found in copies of the Canon Tables for the gospels likewise go back to Eusebius.⁷² We do not know when Eusebius created his Canon Tables for the gospels, so it is impossible to situate them in a developmental chronology with these

⁷⁰ Eusebius, HE 6.32.3 (SC 41, p. 135). ⁷¹ For a discussion of Galen’s use of πίναξ in a similar sense, see Houston, Inside Roman Libraries, p. 259. ⁷² On this point cf. Wallraff, ‘The Canon Tables of the Psalms’, p. 14.

80

The Eusebian Canon Tables

other two works.⁷³ Nevertheless, all three speak to a common attempt to present complex information in a visual form for the purpose of rendering it more accessible. We might then say that Eusebius was the Steve Jobs of the codex, someone who realized the power of design for shaping a user’s interaction with content. More specifically, the Canon Tables for the gospels overlap with each of these other projects in different ways.⁷⁴ Like the Pinax for the Psalms, the Canon Tables provided a classificatory scheme for a body of textual material in order to structure the user’s experience of the text, making it easier to navigate. And like the Chronological Tables, the Canon Tables required the coordination of multiple data sets, each of which had their own internal logic, which could not simply be discarded but rather had to be synchronized according to some all-encompassing, external system of demarcation. For the Chronological Tables this was managed by placing each national chronography in relation to the number of years from Abraham, while for the Canon Tables it was accomplished by the numeric matrices, which acted as an external reference system that left the text of each of the gospels intact.⁷⁵ Hence, if the Chronicle and the Pinax for the Psalms predate the Canon Tables, then they provided Eusebius with the experimental expertise required for producing the latter work. However, even if not, these two works demonstrate those aspects of Eusebius’ ingenuity that he would bring to bear in order to improve upon Ammonius’ prior scholarly investigations.

What did Eusebius Take from Ammonius? With this background in place, it is now time to return to the Letter to Carpianus to continue analysing Eusebius’ comments on Ammonius’ DiatessaronGospel. Though, as noted above, he praises Ammonius’ industriousness and acknowledges his own debt to him, Eusebius also highlights the fundamental ⁷³ Nordenfalk gave a terminus ante quem of 331 on the basis of the assumption that Eusebius included Canon Tables in the Bibles that he produced for Constantine’s new eastern capital, and a terminus a quo of 314 when Eusebius became bishop of Caesarea (Nordenfalk, Die spätantiken Kanontafeln, pp. 50 1). Barnes tentatively suggested the 290s, since Eusebius ‘boldy omitted the spurious last twelve verses of Mark’, whereas he later came to accept the longer Markan ending (Barnes, Constantine and Eusebius, p. 122). The situation is actually more ambiguous than Nordenfalk and Barnes realized. There is no reason Eusebius could not have devised the idea of the Canon Tables before becoming bishop in 314, and Eusebius’ view on the longer ending of Mark remained complicated even in his later years. See Clayton Coombs, A Dual Reception: Eusebius and the Gospel of Mark, Emerging Scholars (Minneapolis, MN: Fortress Press, 2016), pp. 234 7. ⁷⁴ Similarly, Wallraff, ‘The Canon Tables of the Psalms’, p. 14, referred to the ‘canon tables of the psalms’ as the ‘missing link’ between the Chronicle and the Canon Tables for the gospels. ⁷⁵ Hollerich, ‘Eusebius’, p. 638, similarly pointed out that the ‘tabular arrangement’ of the Canon Tables was ‘suggested’ by Eusebius’ prior work in the Chronicle.

The Origins of Scholarship on the Fourfold Gospel

81

insufficiency of his predecessor’s work. Ammonius’ method of arranging the parallel material from the latter three gospels alongside Matthew had the unavoidable result that the continuous thread of the other three was des troyed, as far as a text for reading is concerned ὡς ἐξ ἀνάγκης συμβῆναι τὸν τῆς ἀκολουθίας εἱρμὸν τῶν τριῶν διαϕθαρῆναι ὅσον ἐπὶ τῷ ὕϕει τῆς ἀναγνώσεως.⁷⁶

Though Ammonius’ Ditaessaron-Gospel usefully places similar passages alongside one another so that they can be compared, it makes it impossible to read anything other than Matthew in its proper sequence. This serious limitation meant that the Diatessaron-Gospel could never rise beyond the category of an innovative scholarly reference tool, epiphenomenal to the gospelbook itself, as indeed is the case today with modern gospel synopses.⁷⁷ Recognizing this problem, Eusebius presents a twofold purpose for his composition. He intends to achieve the same fundamental goal as Ammonius— showing parallel material between the gospels—but to do so ‘while preserving the body and sequence of the other [gospels] throughout’ (σωζομένου καὶ τοῦ τῶν λοιπῶν δι’ ὅλου σώματός τε καὶ εἱρμοῦ). His criticism of Ammonius is, therefore, carefully measured. He does not wholly reject his predecessor’s work, and his earlier praise for the Alexandrian’s labour should be taken as sincere. Nevertheless, he recognizes an inevitable limitation in Ammonius’ procedure, and hopes to improve on this earlier composition by using a ‘different method’ (καθ’ ἑτέραν μέθοδον). About the format of Ammonius’ work there is widespread agreement. However, the exact relation between Eusebius’ Canon Tables and Ammonius’ Diatessaron-Gospel is not uncontested territory. Although the sections and numbers in Eusebius’ system have often been called by the adjective ‘Ammonian’,⁷⁸ there exists a significant contrary trend that maintains that these are the sole creation of Eusebius himself. This position was stated emphatically in 1871 by John W. Burgon, who, as a part of his rather heated defense of the longer ending of Mark, described it as a ‘vulgar error’ to ‘designate the Eusebian Sections as the “Sections of Ammonius” ’.⁷⁹ Burgon’s arguments to this end were, 1) that ⁷⁶ ep. Carp. (NA²⁸, p. 89*). For a similar use of ὕϕος in Eusebius’ corpus, see qu. Steph. suppl. 3 (PG 22, col. 961). A similar construction is also found in the Euthalian apparatus in relation to Paul’s letters, though with the word dependency reversed: ἐπιπορευόμενοι τῇ τῆς ὑϕῆς ἀναγνώσει (Euth. prol. Paul. (Blomkvist, p. 107)). ⁷⁷ So also Zahn, Tatian’s Diatessaron, p. 33, who pointed out that Ammonius’ work was intended to serve ‘nicht gottesdienstlichen, sondern gelehrten Zwecken’. ⁷⁸ E.g., Parker, An Introduction, p. 316, who asserted, ‘the paragraphs are properly called the Ammonian Sections, and the numbers themselves the Ammonian Section numbers, they and the table numbers themselves being the Eusebian Numbers’. Similarly, Amphoux, ‘La division’, p. 309, claimed that Ammonius had numbered the sections. ⁷⁹ John W. Burgon, The Last Twelve Verses of the Gospel According to S. Mark Vindicated Against Recent Critical Objectors and Established (London: James Parker and Co., 1871), p. 304.

82

The Eusebian Canon Tables

Eusebius’ Canon Tables were designed to show non-Matthean parallels among Mark, John, and Luke, and to show material distinct to the latter three, whereas such was impossible on Ammonius’ system; and 2) the Canons and the sections ‘mutually imply one another’ such that one without the other would be useless.⁸⁰ Hence Eusebius must have created them both. In 1881 Theodor Zahn made the same point, though for different reasons, and nearly forty years later Zahn was still trying to convince the scholarly guild of its error in this respect.⁸¹ In the early twentieth century Eberhard Nestle, expressly following Burgon, similarly said one should never speak of the ‘Ammonian sections’, since the section division was entirely the work of Eusebius.⁸² More recently Timothy Barnes has concluded that the term ‘Ammonian Sections’ ‘does Eusebius a grave injustice, for the division of the Gospels into numbered sections is his idea’.⁸³ These arguments are rightly aimed at giving Eusebius due credit for his invention. Nevertheless, I want to suggest that these scholars have gone too far in claiming that the two works had nothing whatsoever in common. Rightly articulating the relationship between the contributions of these two figures centres around the interpretation of a key word in Eusebius’ letter. As he transitions to describing his own creation, Eusebius notes that ‘he took (his) starting points from the labour of the aforementioned [Ammonius]’ (ἐκ τοῦ πονήματος τοῦ προειρημένου ἀνδρὸς εἰληϕὼς ἀϕορμάς).⁸⁴ What exactly did Eusebius take from Ammonius? Barnes translates ἀϕορμάς as ‘point of departure’, Burgon as ‘hint’ or ‘suggestion’, and Zahn as ‘Anregung’.⁸⁵ For the term LSJ lists ‘occasion’ or ‘pretext’ as well as ‘means with which one begins’ or ‘resources’. In keeping with the latter sense, the word can take on the economic meaning ‘capital’ or the rhetorical meaning ‘food for argument, material, subject’.⁸⁶ So the semantic range is broad enough to encompass the more generic causal sense of ‘occasion’ or ‘impetus’, as well as a more specific sense indicating a greater degree of material continuity between the ἀϕορμή and the resulting piece of work.

Burgon goes on forcefully: ‘to reason about the lost work of Ammonius from the Sections of Eusebius (as Tischendorf and the rest habitually do) is an offence against historical Truth which no one who values his critical reputation will probably hereafter venture to commit’. ⁸⁰ Burgon, The Last Twelve Verses, pp. 295 8. ⁸¹ Zahn, Tatian’s Diatessaron, 31 2; Zahn, ‘Der Exeget Ammonius’, p. 8. ⁸² Nestle, ‘Die Eusebianische Evangeliensynopse’, p. 41, disagreeing with the description of the sections in the edition of the Vulgate by Wordsworth and White. ⁸³ Barnes, Constantine and Eusebius, p. 122. ⁸⁴ ep. Carp. (NA²⁸, p. 89*). ⁸⁵ Burgon, The Last Twelve Verses, p. 127; Zahn, Tatian’s Diatessaron, p. 32; Barnes, Con stantine and Eusebius, p. 121. Zahn specifies that the only thing Eusebius took from Ammonius was the idea of presenting parallel passages alongside one another. Barnes’ translation is also followed by David Laird Dungan, A History of the Synoptic Problem: The Canon, the Text, the Composition, and the Interpretation of the Gospels, The Anchor Bible Reference Library (New York: Doubleday, 1999), pp. 109 10. ⁸⁶ LSJ, s.v. ἀϕορμή, I.2; I.3; I.4; I.5.

The Origins of Scholarship on the Fourfold Gospel

83

The latter sense, I want to argue, is what Eusebius had in mind, especially in view of the fact that he uses the term in the plural. According to Thesaurus Linguae Graecae, the word occurs some thirty-three times in his corpus, in both the singular and plural forms. When it is used in the singular, it usually means something more general like ‘occasion’ or ‘pretext’. For example, Eusebius quotes Melito of Sardis stating that certain persons were persecuting the Christians by taking their ‘occasion’ from the imperial edicts. In another passage the church historian reports that he is unable to give more precise information about a number of figures because he does not have the ‘occasion’ to do so. In another instance Eusebius passes on a report that Satan entered into the schismatic Novatus, becoming the ‘occasion’ for his unbelief.⁸⁷ However, when it is used in the plural, the term often implies that the ἀϕορμαί bear some more material relation to that which results from them. For example, Eusebius quotes the report of Irenaeus that the heretic Cerdon took his ‘material’ (ἀϕορμάς) from those who followed Simon Magus. Later on the historian asserts that by studying the scriptures from his childhood Origen had stored up ‘no small amount of resources (ἀϕορμάς) of the words of the faith’.⁸⁸ In addition to these two occurrences in the plural in the Ecclesiastical History, there are also over a dozen plural usages in Eusebius’ Praeparatio Evangelica. For example, he cites Philo of Byblos stating that Pherecydes, ‘taking his starting points from the Phoenicians, theologized about the god he called Ophion’ (παρὰ Φοινίκων δὲ καὶ Φερεκύδης λαβὼν τὰς ἀϕορμὰς ἐθεολόγησεν περὶ τοῦ παρ’ αὐτῷ λεγομένου Ὀϕίονος θεοῦ). Just a little further on, in the same extract, Philo of Byblos wrote that ‘everyone, taking their starting points from Tauthus, discoursed on the natural principles’ (πάντες δὲ τὰς ἀϕορμὰς παρὰ τοῦ Τααύτου λαβόντες ἐϕυσιολόγησαν). Aristobulus, quoted by Eusebius in the eighth book of the treatise, explained that certain ‘philosophers and many others, even poets, took great material from [Moses]’ (ϕιλόσοϕοι καὶ πλείονες ἕτεροι καὶ ποιηταὶ παρ’ αὐτοῦ μεγάλας ἀϕορμὰς εἰληϕότες). Later, Eusebius quotes Clement of Alexandria, who claimed that Genesis 1:1 ‘presented to the [Greeks] the starting points of a material substance’ (ἡ λέξις ἡ προϕητικὴ ἐκείνη· ‘Ἡ δὲ γῆ ἦν ἀόρατος καὶ ἀκατασκεύαστος’ ἀϕορμὰς αὐτοῖς ὑλικῆς οὐσίας παρέσχηται), and that the Orphic word μητρο πάτωρ ‘gave to those who introduce emissions the means to think that God has a consort’ (ἐνδέδωκε δὲ ἀϕορμὰς τοῖς τὰς προβολὰς εἰσάγουσι τάχα καὶ σύζυγον νοῆσαι τοῦ θεοῦ). Finally, in book fourteen Eusebius himself asserts that Epicurus ‘took the material from [the Cyrenaic sect] for the exposition of [humanity’s] end’ (ἀϕ’ ἧς τὰς ἀϕορμὰς Ἐπίκουρος πρὸς τὴν τοῦ τέλους ἔκθεσιν εἴληϕεν) and that a certain saying of Metrodorus ‘gave evil material

⁸⁷ HE 4.26.5; 5.27.1; 6.43.14 (SC 31, p. 209; SC 41, pp. 74, 157). ⁸⁸ HE 4.11.2; 6.2.7 (SC 31, p. 174; SC 41, p. 84).

84

The Eusebian Canon Tables

to Pyrrho who came afterwards’ (κακὰς ἔδωκεν ἀϕορμὰς τῷ μετὰ ταῦτα γενομένῳ Πύρρωνι).⁸⁹ This series of examples is sufficient to indicate that when Eusebius says that one individual ‘took’ or ‘gave’ ἀϕορμάς to or from someone else, he intends to establish an intellectual genealogy between the two, implying a significant degree of continuity between them. This is not a uniquely Eusebian usage. A particularly clear passage on this point is found in the summary of the views of the grammarians provided by the sceptic Sextus Empiricus: ‘That poetry provides numerous starting points (ἀϕορμάς) toward happiness is clear from the fact that the most truly powerful and character-building philosophy was rooted at the beginning in the gnomic utterances of the poets, and for this reason if philosophers say something by way of exhortation they seal it, so to speak, with poetic sayings’.⁹⁰ Note that the justification for seeing philosophical ‘starting points’ in the poets is that philosophy from the beginning ‘was rooted’ in poetry. Like Eusebius, Sextus Empiricus assumes that ἀϕορμαί are what one takes from someone else as a kind of raw material, to which one then applies some kind of process for the purpose of formulating one’s own contribution to the issue at hand. The final product will not be identical to that with which one began, but it is materially indebted to it. In the light of this background, when in the Letter to Carpianus Eusebius stated that he had ‘taken’ ἀϕορμάς from Ammonius, he was claiming precisely such an intellectual pedigree. It was not merely the deficiencies of Ammonius’ creation or the general idea of comparing gospel passages that inspired his labour. Rather, the Diatessaron-Gospel provided for him the ‘starting points’ or perhaps ‘raw material’ that he reworked according to a different method for his own composition. Therefore, in the light of Eusebius’ usage of ἀϕορμή, those scholars who have argued there is no real relation between the Diatessaron-Gospel and the Canon Tables are mistaken, since Eusebius himself indicates that there is a significant continuity between the two works.

Eusebius’ Modus Operandi We can state more precisely what this continuity was by speculating for a moment about how Eusebius might have gone about his work.⁹¹ ⁸⁹ PE 1.10.50; 1.10.53; 8.10.4; 13.13.4; 13.13.53; 14.18.31; 14.19.9 (GCS 8.1, p. 53, 54, 452; GCS 8.2, p. 199, 222, 313, 315). ⁹⁰ Sextus Empiricus, math. 1.271 (Mau and Mutschmann, p. 271; trans. Blank, p. 54). For a similar passage see Philodemus, On the Good King according to Homer, col. 43, discussed in Elizabeth Asmis, ‘Philodemus’s Poetic Theory and “On the Good King According to Homer” ’, Classical Antiquity 10 (1991): pp. 20 1. ⁹¹ Though he misunderstood the nature of Eusebius’ Canon Tables, David L. Dungan’s description of the difficulty of producing a gospel synopsis overlaps with the steps Eusebius

The Origins of Scholarship on the Fourfold Gospel

85

Ammonius had essentially already demarcated the Gospel of Matthew into sections, simply based on where he ended one parallel and began another. Similarly, his method had the same effect for the other three gospels, at least for those passages with Matthean parallels, since he had to decide which chunks of these gospels to use as parallels for Matthew in a cut-and-paste method. In other words, what Ammonius had already accomplished was establishing a set of parallels with Matthew, consisting of sets of discrete passages of text, and these parallels could easily have been taken over by Eusebius into his new system. In fact, given that he clearly had access to Ammonius’ work and surely realized how much easier it would have made his own task, the burden of proof should rest on those who want to maintain that he did not incorporate his predecessors’ parallels into his new apparatus. It follows, then, that the parallel passages noted in these Canons likely, in some form, go back to Ammonius, though we cannot exclude the possibility that Eusebius tweaked them here and there to his own liking.⁹² As a result, it is best to avoid talk of the ‘Ammonian sections’, though when we speak of the ‘Eusebian sections’ we should always bear in mind his indebtedness to his Alexandrian forebear. The data constituted by these parallels provided the ‘starting points’ for Eusebius’ labour. To rework them according to his own method, he would only have had to follow five steps.⁹³ First, as he looked through the DiatessaronGospel he could easily have discovered that there were eight possible combinations of passages that appeared: (1) Mt-Mk-Lk-Jn, (2) Mt-Mk-Lk, (3) Mt-LkJn, (4) Mt-Mk-Jn, (5) Mt-Lk, (6) Mt-Mk, (7) Mt-Jn, and (8) material distinctive to Matthew without any parallel. Eusebius probably took over these eight combinations, making them Canons I–VII and Canon XMt in his system.

must have undertaken to produce his Canon Tables. Dungan listed three ‘basic problems’ attendant to creating a synopsis: ‘a) how to divide the material into pericopes’; ‘b) how to decide what are genuine parallels’; ‘c) how to arrange the parallels throughout the synopsis’ (David L. Dungan, ‘Theory of Synopsis Construction’, Biblica 61 (1980): p. 321). Eusebius had to deal with the first of these problems but not the third, since his system does not require a sequential ordering of parallels, relying instead on the numerical randomness of his ten tables. As I argue in the following chapter, this is in fact one of the virtues of his cross referencing system. ⁹² Though he did not recognize that only eight of Eusebius’ tables could have come from Ammonius, David Laird Dungan was, therefore, basically correct in asserting that Eusebius ‘took over Ammonius’ paragraph divisions (sectiones)’ (Dungan, A History of the Synoptic Problem, p. 109). It is unclear to me why Dungan refers to these divisions as sectiones when both Ammonius and Eusebius were writing in Greek. ⁹³ For a similar description of this process, see Watson, The Fourfold Gospel, pp. 116 23. The primary difference between Watson’s presentation and my own is that he argued that Eusebius obtained all of his categories from Ammonius’ Diatessaron Gospel, on the basis of his assump tion that it also included the material from the latter three gospels that did not have Matthean parallels, presented in synopsis like fashion. I prefer to remain agnostic regarding the presence or absence of the non Matthean content, and so argue here that he obtained eight of his categories from Ammonius.

86

The Eusebian Canon Tables

This specific point was already recognized in the ninth century by the Irish scholar Sedulius Scottus, whom we will meet again in chapter six.⁹⁴ In other words, Eusebius first used Ammonius to establish the categories of relationship among the gospels simply by analysing the parallels included in the Diatessaron-Gospel. Eusebius’ second step would have been to make marginal notations in an unmarked copy of Matthew to demarcate the text into sections based on the established parallels and then to enumerate these sections. Eusebius never tells us that Ammonius numbered the sections in his synopsis, and the functioning of his composition did not require enumeration, so we should infer that the numbers are Eusebius’ contribution, and we should accordingly speak of the ‘Eusebian numbers’. Thirdly, Eusebius had to demarcate Mark, Luke, and John into sections. Here again Ammonius had already contributed some of the work, since Eusebius could have worked through Ammonius’ parallels with Matthew and use them to mark the parallels in the margin of the text of the latter three gospels. This, however, would have been a more difficult task than sectioning Matthew, since these parallels would have been included in Ammonius’ scheme according to Matthew’s narrative order, not according to the sequence of the other three, and so would have required much turning of pages to find the correct passages in the latter three gospels. Moreover, if Ammonius had not included non-Matthean parallels in his work, Eusebius would then have had to work through the remaining text from these three gospels that was not yet sectioned in order to establish his own parallels among this remaining material. At this point he could have used any of the three as a plumbline to check for parallels. His apparent choice was to use Luke, most likely because Luke, being the longest of the remaining three, offered the potential for the most parallels with the other two.⁹⁵ Working back through Luke’s gospel, looking for material similar to Mark and John, Eusebius must have further subdivided these three gospels, noting down the parallels for his Canons VIII (Lk–Mk) and IX (Lk–Jn). Eusebius’ fourth step would have been to enumerate the sections he had created in Mark, Luke, and John, and his fifth and final step was to collate into tables the section numbers according to the relational categories he had established. This entire process was intricate and complex, so the fact that the resulting system is almost entirely free of errors is a remarkable scholarly achievement.⁹⁶ One of the small errors that does show up in the system is ⁹⁴ See his ex. Eus. pp. 2 3. ⁹⁵ As also noted by Nordenfalk, Die spätantiken Kanontafeln, p. 48. ⁹⁶ Occasional slips have been identified in Harvey K. McArthur, ‘Eusebian Sections and Canons’, Catholic Biblical Quarterly 27 (1965): pp. 255 6; Walter Thiele, ‘Beobachtungen zu den

The Origins of Scholarship on the Fourfold Gospel

87

that on a handful of occasions a new section is started where it is not needed.⁹⁷ Recall that according to the logic of the Canons, a new section should only begin when the text transitions to a passage with a different relationship to other gospel passage(s). In theory this should mean that there will never be consecutive sections which are attributed to Canon X, since Canon X passages have no parallels in the other gospels, and in fact consecutive Canon X passages almost never occur, though there are a few exceptions. Luke 6:24–25 and 6:26 (Lk §50, 51) are both ascribed to Canon X, as is Luke 7:11–16 and 7:17 (Lk §67, 68), Luke 9:61–2 and 10:1 (LK §106, 107), Luke 13:1–5 and 13:6–13 (Lk §163, 164), and John 7:33 and 7:34–49 (Jn §80, 81). In each of these instances, the two consecutive passages should have been treated as one section and enumerated accordingly. Whence then did these errors arise? Ammonius’ work had already included all of Matthew, noting parallel and unique passages from beginning to end. However, if Ammonius had not included the content from the latter three gospels that does not have Matthean parallels, Eusebius would have had to undertake original research with Mark, Luke, and John. Among these three Mark, as the shortest, had the least amount of remaining text to be placed in an appropriate Canon. Hence, Eusebius probably had to do the largest amount of original work with the remaining portions of Luke and John that were omitted from Ammonius’ edition, so it is sensible that these are the places where errors would most likely occur. Probably what has happened is that Eusebius divided each of these passages into two sections because he intended to place them in different Canons, but later either forgot to do so or realized that the parallels he had in mind were not close enough to warrant juxtaposition. Hence these passages were simply relegated to Canon X as distinctive material to each of these gospels. These are errors that could have been removed by recombining the unnecessary doublets and renumbering the subsequent sections, but for whatever reason, Eusebius chose not to do this. At any rate, they are at least errors that do not hinder the goal of the overall system to provide crossreferences between gospels. In conclusion, then, we can say that the majority of the parallels, at least those for Canons I–VII and one quarter of Canon X, go back in some form to the work of Ammonius. However, the enumerating of the sections and the collating of their numbers were the work of Eusebius. Therefore, the resulting composition was truly the product of the labour of both scholars, with Eusebius appropriating and improving upon his predecessor’s work. eusebianischen Sektionen und Kanones der Evangelien’, Zeitschrift für neutestamentliche Wissen schaft 72 (1981): pp. 106 9; Carl Nordenfalk, ‘The Eusebian Canon Tables: Some Textual Prob lems’, Journal of Theological Studies 35 (1984): pp. 96 104; Watson, The Fourfold Gospel, p. 149. ⁹⁷ These errors were also identified in McArthur, ‘Eusebian Sections and Canons’, p. 256.

88

The Eusebian Canon Tables

Accordingly, the opening page of the sixth-century Rabbula Gospels, which presents a combined portrait of Ammonius and Eusebius, is indeed a fitting tribute to the work of the two men (see figs 19–20).⁹⁸

Fig. 19. Portraits of Ammonius (right) and Eusebius (left) in the Rabbula Gospels (sixth c.). Florence, Biblioteca Medicea Laurenziana, MS. Plut. 1.56, fol. 2r

⁹⁸ On the dating of the Rabbula Gospels, see n. 10 in the Introduction.

The Origins of Scholarship on the Fourfold Gospel

89

Fig. 20. Image of Eusebius (top left), Carpianus (top right), and Ammonius (bottom) in Parma, Biblioteca Palatina, MS. gr. 5, fol. 12v (latter half of eleventh c.). By permission of the Ministry of Cultural Heritage and Activities and Tourism

90

The Eusebian Canon Tables

I argued in the last chapter that the design of Eusebius’ Canon Tables is indebted to astronomical tables such as those produced by Ptolemy, or perhaps by Christian paschal tables that had drawn upon that earlier astronomical genre. The present argument for his indebtedness to Ammonius is not meant to undermine that earlier claim. It remains the case that as a purely numerical matrix, Ptolemy’s Handy Tables presents by far the closest ancient analogy to the Canon Tables, and this formal similarity seems hardly coincidental. However, it is also possible that the mis-en-page of Eusebius’ Canon Tables was partially inspired by Ammonius’ composition. The columnar format of the Canon Tables could have been simply carried over directly from the design layout of the Diatessaron-Gospel, with Eusebius retaining a discrete column devoted to each gospel and the usage of rows to juxtapose parallel passages.⁹⁹ In other words, all Eusebius had to contribute to the mis-en-page of Ammonius was the replacement of sections of text with numbers. This, in fact, is the fundamental conceptual breakthrough that enabled him to advance beyond the work of his predecessor—the realization that passages of text could be symbolically represented by numbers. As reviewed in the last chapter, by Eusebius’ time authors regularly divided lengthier works up into numbered books, as Eusebius himself did with his other works. Moreover, as pointed out by Martin Wallraff, this insight is also evident in the Pinax for the Psalms, in which the various psalms are referred to by their numbers.¹⁰⁰ Hence, Eusebius’ usage of this method in the Canon Tables was not wholly original, but it is nevertheless the case that no one, either Christian or non-Christian, had ever exploited this insight to such a degree.¹⁰¹ It was, therefore, from the confluence of all of these factors that the Canon Tables emerged: in the Chronicle Eusebius had devised an all-encompassing reference scheme that left intact the distinct internal logic of the data sets he was combining; in the Pinax for the Psalms he had experimented with referring to texts by numbers and categorizing constituent parts of a literary ⁹⁹ Werner H. Kelber, ‘The History of the Closure of Biblical Texts’, Oral Tradition 25 (2010): p. 126, and William R. S. Lamb, The Catena in Marcum: A Byzantine Anthology of Early Commentary on Mark, Texts and Editions for New Testament Study 6 (Leiden: Brill, 2012), p. 146, both suggested that Eusebius’ format was inspired by Origen’s Hexapla, but seem not to have recognized that it was most likely also indebted to Ammonius’ composition. ¹⁰⁰ Wallraff, ‘The Canon Tables of the Psalms’, p. 14. ¹⁰¹ Cf. Riggsby, Mosaics of Knowledge, chapter 1: ‘By the time we have longer surviving literary texts in Latin, they are divided by their authors into “books” that is to say, units that will notionally fit on a single papyrus roll which are then numbered. . . . Given this nearly universal convention, it is then striking that smaller units of Latin texts are seldom if ever numbered. In principle, several such units were available. Today we refer to bits of Latin poetry by line number and (in the case of collections of short poems) the number of a given poem within its book. There seems to be no evidence for either practice in antiquity, whether directly in manuscripts or in external reference (say, in commentaries). This is despite the fact that lines were often counted in individual manuscript copies, apparently as part of the pricing process’. Further on Riggsby comments, ‘there are apparently no indices in any classical Roman texts’.

The Origins of Scholarship on the Fourfold Gospel

91

corpus; and in Ammonius he had a predecessor who had inaugurated the study of gospel parallels by using a columnar format. Perhaps all it took was Eusebius seeing something like a Ptolemaic table for him to realize that numbers in a matrix could be used as an external reference scheme for the fourfold gospel, one that, unlike Ammonius’ Diatessaron-Gospel, preserved the structure of the gospel narratives. The result of this innovation was that a codex with Canon Tables began with a symbolic summary of the fourfold gospel before one even arrived at the start of the actual text. In an age accustomed to scholarly tools such as footnotes, indexes, and cross-references, it is difficult for us to imagine how startlingly innovative the Canon Tables must have appeared to a late-antique reader. For probably the first time in intellectual history, a problem of textual complexity was rendered in numerical form. Yet despite its innovative quality, Eusebius’ creation proved to be incredibly popular, owing in part to one significant advantage it had over Ammonius’ earlier work. While the Diatessaron-Gospel could never have been more than a secondary reference tool, Eusebius found a way to accomplish the same goal while leaving the text of each gospel intact, thereby allowing for his new system to be included in liturgical codices and widely disseminated. The implication of this advance should not be missed. Eusebius contributed significantly to the developing Christian traditions of late antiquity and the early medieval period by making the elite philological scholarship of Alexandria available on a scale few could ever have imagined. The present argument therefore stands in agreement with the recent assessment of Michael J. Hollerich that Eusebius was ‘a true founder of Christian biblical scholarship’.¹⁰²

From Formlessness to Polymorphic Diversity Yet Eusebius’ creation was not merely a scholarly tool in the modern sense of the genre. It was also laden with a symbolic, even mystical significance that may not be immediately obvious to the modern reader. Eusebius’ tendency to use a visual medium to make a theological statement has been emphasized recently by Grafton and Williams in their study of his experimentations with the opportunities afforded by the new technology of the codex. They drew attention to the way in which the Chronicle began with tables of parallel monarchies, but, as the reader progressed through history, all others fell away to leave only the list of Roman rulers. As a kind of ‘dynamic hieroglyph’,

¹⁰² Hollerich, ‘Eusebius’, p. 629. Hollerich made this comment on the basis of a wider study of Eusebius’ corpus as it relates to biblical scholarship, but he did note that ‘Eusebius’ most original contributions to biblical studies were technical instruments like the Onomasticon and the Canon Tables’ (p. 636).

92

The Eusebian Canon Tables

this display of textual material communicated Eusebius’ conviction that ‘world history culminated in the contemporary Roman Empire’.¹⁰³ Scholars have long speculated about a similar impulse evident in the Canon Tables. Nordenfalk pointed out that the ten Canons do not exhaust all of the possible combinations presented by the fourfold gospel, since the parallels Mk–Lk–Jn as well as Mk–Jn are absent.¹⁰⁴ Eusebius’ omission of these categories is, however, more likely due to the small amount of content for these Canons, since there is little material shared by Mark, Luke, and John that is not also shared by Matthew, and even less non-Matthean content that is contained jointly in Mark and John but not in Luke. A more compelling observation related to the number ten is that Eusebius did not assign the distinctive material for each gospel to a separate Canon, as his prior method with Canons I–IX would imply, but instead grouped all four distinct categories together into a single table, Canon X. This clear departure from his pattern with the first nine categories suggests a desire to preserve the number ten, as though the number of the Canons was intrinsic to the overall message communicated by the tables. The number ten seems also to have been a factor in the page-layout of the tables. Nordenfalk attempted to reconstruct the Eusebian archetype from the surviving late-antique and early medieval models, and, after examining a number of later examples, he concluded that the tables were originally distributed over five folia, comprising ten pages.¹⁰⁵ Nordenfalk plausibly suggested that this highlighting of the number ten is due to a Pythagorean recognition of the relation of the numbers four and ten, specifically the fact that ten is the sum of the numbers from one to four.¹⁰⁶ An example of this tradition probably contemporary with Eusebius is the ps-Iamblichan treatise The Theology of Arithmetic, which offers a commentary on each of the first ten numbers. About the number ten the text notes ‘a natural equilibrium and commensurability and wholeness existed above all in the decad’.¹⁰⁷ Earlier in the treatise, the author opens the discussion of the ¹⁰³ Grafton and Williams, Christianity and the Transformation, p. 141. The same observation was made in Burgess and Witakowski, Studies in Eusebian and Post Eusebian Chronography, p. 81. Both Grafton and Williams as well as Burgess and Witakowski follow Glenn Chesnut, The First Christian Histories, 2nd edn (Macon, GA: Mercer University Press, 1986), pp. 76, 137. ¹⁰⁴ McArthur, ‘The Earliest Divisions of the Gospels’, p. 251, suggested that these combinations were omitted because Eusebius found no parallel sections in the gospels in question. Nordenfalk, ‘Canon Tables on Papyrus’, 30, n.6, objected that there were, in fact, some parallels that would have been suitable for these hypothetical Canons, but he was able to offer only a handful of examples. ¹⁰⁵ Nordenfalk, Die spätantiken Kanontafeln, pp. 102 3. The Armenian tradition was espe cially ardent in preserving a total of ten pages for the prefatory Eusebian material. See the discussion in chapter seven. The Syriac and Latin traditions expanded the total number of pages well beyond the original sequence. ¹⁰⁶ Nordenfalk, ‘Canon Tables on Papyrus’, pp. 29 30. The Pythagorean symbolism is also accepted by Marchese, ‘Tables and Early Information Visualization’, pp. 44 5. ¹⁰⁷ Ps Iamblichus, arith. 10 (de Falco, p. 79; trans. Waterfield, p. 109). ϕυσικὴ δέ τις συσταθμία καὶ μετριότης καὶ ὅλωσις ἐν τῇδε μάλιστα ὑπῆρχε.

The Origins of Scholarship on the Fourfold Gospel

93

tetrad by commenting, ‘Everything in the universe turns out to be completed in the natural progression up to the tetrad, in general and in particular, as does everything numerical—in short, everything whatever its nature. The fact that the decad . . . is consummated by the tetrad along with the numbers which precede it, is special and particularly important for the harmony which completion brings’.¹⁰⁸ As noted by Nordenfalk, Eusebius himself highlighted this Pythagorean theme in his Oration in Praise of Constantine, defining ten as the sum of the numbers from one to four, and calling ten ‘a full and perfect number’, since it contains ‘every kind and measure of all numbers, proportions, concords, and harmonies’.¹⁰⁹ The fact that the Canon Tables proceed in descending order from parallels between four, then three, then two gospels, and finally passages unique to each gospel reveals a pattern of 4-3-2-1, highlighting the mathematical relationship made explicit in the passage from the Oration. Thus, both the number of Canon Tables and the progression within them were probably intended to represent the sense of ‘equilibrium, commensurability, and wholeness’ inherent in the number ten. Further evidence on this point can be found elsewhere in Eusebius’ corpus, as well as among the later users of his Canon Tables. Eusebius’ Ecclesiastical History may have originally been published with fewer books, but the final version consisted of ten, and in the preface to the tenth one he justified the addition of the final book by asserting that it was ‘reasonable’ to end with ‘a perfect number’ (εἰκότως . . . ἐν ἀριθμῷ τελείῳ).¹¹⁰ If he thought it necessary to explain the number of books of his Ecclestiastical History in this manner, it is no stretch to suppose that he thought the same with respect to his Canon Tables, though he never said as much explicitly. Certainly later interpreters thought that this number was deliberately chosen, or, perhaps more accurately, divinely necessary. Victor of Capua in the sixth century asserted that the number of Canons was determined ‘according to reason and a natural rule’ (ratione et regula naturali), since no other number, either less or more than ten, was possible.¹¹¹ Similarly in the twelfth century, the Syriac author Dionysius bar Salibi argued that Eusebius established ten tables because ‘the number ten is born from the number four. How? Consider: four, three, two, and one. Behold these add up to ten’.¹¹² Hence, both in the creation and in the reception of Eusebius’ Canon Tables, the fact that there were exactly ten tables was taken to be deeply symbolic. One might object that this is merely a post hoc justification for what is in fact an arbitrary ¹⁰⁸ Ps Iamblichus, arith. 4 (de Falco, p. 20; trans. Waterfield, p. 55). Ὅτι ἐν τῇ μέχρι τῆς τετράδος ϕυσικῇ ἐπαυξήσει πάντα συντελούμενα ϕαίνεται τὰ ἐν τῷ κόσμῳ, καθόλου καὶ κατὰ μέρος, καὶ τὰ ἐν ἀριθμῷ, ἐν πάσαις ἁπλῶς ϕύσεσιν· ἐξαίρετον δὲ καὶ πρὸς τὴν ἐϕάρμοσιν τοῦ ἀποτελέσματος μάλιστα συντεῖνον τὸ τὴν δεκάδα ὑπ’ αὐτῆς ἅμα τοῖς ὑπόπροσθεν συγκορυϕοῦσθαι. ¹⁰⁹ Eusebius, l.C. 6.5; 6.14 (GCS 7, pp. 207, 210). ¹¹⁰ Eusebius, HE 10.1.3 (SC 55, p. 77). ¹¹¹ Victor of Capua, praef. (Ranke, pp. 2 3). ¹¹² Dionysius bar Salibi, comm. ev. pref. 44 (CSCO 15, p. 26).

94

The Eusebian Canon Tables

number, and it is certainly true that the Eusebian paratext would function just as well as a cross-referencing system if its creator had numbered them consecutively up to thirteen rather than artificially stopping at ten. Nevertheless, the decision to have only ten tables does reflect an aspect of Eusebius’ own understanding of his project, one that otherwise would be easy for those who live in a world saturated with numeric matrices to miss. For him as well as for the users of his paratext, the ten canons were thought to illustrate the deep concord that existed between the divine revelation in the fourfold gospel and the divinely created cosmos. Moreover, the fact that the number ten was understood to be not simply a ‘perfect number’ but also that which gave ‘order’ to the other numbers makes it a particularly apt analogy for the function that the Eusebian paratext performs in relation to the text of the gospels. Just as Ps-Iamblichus had stated that ‘things from heaven to earth, existing harmoniously according to the principles in [the decad], are found, both in general and in particular, to have been ordered by [the decad]’,¹¹³ so too the reader of the four gospels could find the divine revelation contained therein ‘ordered’ by Eusebius’ ten tables. There is further, unnoticed, material from the Oration in Praise of Constantine that highlights the resonances that I am arguing are implicit in the visual presentation of Eusebius’ Canons. The passage cited by Nordenfalk in which Eusebius mentions the formula 1 + 2 + 3 + 4 = 10 occurs in a long discussion of the divine ordering of the cosmos, and Eusebius’ discourse about these cosmological themes at times sounds almost like a description of his Canon Tables. He begins by pointing out that ‘eternity’ (αἰών) is ‘indestructible’ and ‘endless’ and ‘will not submit to mortal comprehension’, yet the divine ‘Sovereign and Master’ is able to ‘ride it from above’, using his wisdom like reins to control it. The manner of the deity’s control is specified when Eusebius says that he ‘punctuates’ (κατεβάλετο) the monadic, undifferentiated unity of eternity with the ‘complete harmony’ (σὺν ἁρμονίᾳ τῇ πάσῃ) of months, dates, seasons, and years, thereby ‘circumscribing it with manifold boundaries and measurements’. In other words, although eternity is ‘alike in all its parts (or rather has no parts and is indivisible)’, [God] has, by marking it off into intermediate sections and dividing it like an extended straight line cut with compass points, established a great multitude in it. Thus, though [eternity] is one and comparable to a monad, he has delimited it with all kinds of numbers, causing its formlessness to subsist as a polymorphic diversity. ὁ δὲ μέσοις αὐτὸν διαλαβὼν τμήμασι καὶ ὥσπερ εὐθεῖαν γραμμὴν εἰς μῆκος τεταμένην διελὼν κέντροις πολὺ πλῆθος ἐν αὐτῷ κατεβάλετο, ἕνα τε ὄντα καὶ μονάδι παρεικασμένον παντοίοις κατεδήσατο ἀριθμοῖς, πολύμορϕον ἐξ ἀμόρϕου τὴν ἐν αὐτῷ ποικιλίαν ὑποστησάμενος.¹¹⁴ ¹¹³ Ps Iamblichus, arith. 10 (de Falco, pp. 79 80; trans. Waterfield, p. 109 [modified]). διόπερ τοῖς κατ’ αὐτὴν λόγοις συμϕώνως ἔχοντα τὰ ἀπ’ οὐρανοῦ μέχρι γῆς ὁλοσχερέστερόν τε καὶ κατὰ μέρος εὑρίσκεται [καὶ] διακεκοσμημένα κατ’ αὐτήν. ¹¹⁴ Eusebius, l.C. 6.3 4 (GCS 7, pp. 206 7).

The Origins of Scholarship on the Fourfold Gospel

95

In the next section Eusebius further elaborates that God first created formless matter; second ‘by the power of the dual’ created ‘form out of the formless’; third, he ‘tripled dimension’ into breadth, length, and depth to produce a body uniting matter and form; fourth, he ‘doubled the dual’ to create the four elements (earth, water, air, and fire). It is at this point that he points out that the ‘quaternion produces the decade’ in the passage cited by Nordenfalk, and after this statement he moves on to a discussion of the months, seasons, and years, which ‘give form to shapeless and formless eternity for the refreshment and delight of those who are riding out their lives on its track’.¹¹⁵ As the above passage illustrates, Eusebius, like nearly all of his philosophically inclined contemporaries, invested arithmetical computation with cosmological significance, assuming that the ordered multiplicity of human experience was an outflow of the unitary timelessness of divinity.¹¹⁶ Hence, the changing of the seasons, the passage of the heavenly bodies, and the methods of arithmetic were all visible expressions of the deep structural concord at the heart of reality which stemmed from the Creator himself. What is remarkable about the above passage is that this view of multiplicity and unity could almost serve as a description of Eusebius’ own editorial work on the gospels. Like the divine artisan, he too ‘marked off ’ or ‘punctuated’ the undifferentiated, continuous text of the gospels into ‘intermediate sections’, ‘delimiting it with all kinds of numbers’, with the result that the previously ‘formless’ text came to exist instead as a paratextually structured new edition that displayed to the reader the harmonious relation of all its constituent parts. Perhaps it is just a coincidence that Eusebius’ description of the creation of the universe is reminiscent of his own paratextual invention. However, given the fact that Christians had, since at least the time of Irenaeus,¹¹⁷ connected the ordered microcosm of the sacred text with the ordered cosmos at large, it seems to me not a stretch to suppose that Eusebius believed himself to be following in his Creator’s footsteps, producing a new technology that revealed the divinely ordered ‘polymorphic diversity’ evident in the church’s four canonical texts.¹¹⁸

¹¹⁵ Eusebius, l.C. 6.5 (GCS 7, p. 207). ¹¹⁶ Cf. Joel Kalvesmaki, The Theology of Arithmetic: Number Symbolism in Platonism and Early Christianity, Hellenic Studies 59 (Washington, DC: Center for Hellenic Studies, 2013). ¹¹⁷ Cf. Irenaeus, haer. 3.11.8, on which see Graham N. Stanton, ‘The Fourfold Gospel’, New Testament Studies 43 (1997): pp. 319 22. ¹¹⁸ Cf. König and Whitmarsh, ‘Ordering Knowledge’, p. 34: ‘The Bible was, indeed, perhaps the first book conceived of as a textual embodiment of the cosmos. In the beginning was the word: God’s language was the sacred transcription of the mysteries of the universe, of human society and mortality, of bodily suffering and spiritual redemption’. The recognition that paratextual features can serve ends beyond the strictly utilitarian is also evident in Riggsby, ‘Guides to the Wor(l)d’, p. 102, who contended that the table of contents for Pliny’s Natural History is just as much about displaying Pliny’s mastery of knowledge as it is about helping the reader find the information that he is seeking: ‘Reduction of the whole world to thirty seven books asserts control of that world; reduction of those thirty seven books to one does the same thing’.

3 Reading the Gospels with the Eusebian Canon Tables Having examined the ‘what’ and ‘whence’ of Eusebius’ paratextual apparatus, the goal of the present chapter is to clarify the ‘whither’. In other words, the question at hand is: to what end did Eusebius expect people to use his Canon Tables? I have of course already given a preliminary answer to this question by arguing in chapter one that the tables clarify the relationships amongst the gospels and so order the textual material in the fourfold gospel, and by arguing in chapter two that Eusebius wanted to enable the kind of comparative analysis fostered by Ammonius’ Diatessaron-Gospel while preserving intact the text of the four gospels. More, however, can be said to expand on these two points, both by theoretical reflection on the pressure the paratextual apparatus exerts upon the reader and by consideration of the actual parallels created by Eusebius. Specifically, I want to argue in the present chapter that Eusebius’ paratext has a threefold effect on the reading of the fourfold gospel: 1) it binds together four originally separate texts into a single corpus of literature, and by implication excludes all other texts from this category; 2) it encourages a kind of hypertextual reading, in which passages from one gospel are read alongside passages from one or more other gospels; and 3) within this hypertextual mode the Eusebian apparatus is decidedly underdetermined, expressing a Bakhtinian openness with respect to the resolution of tensions internal to the corpus. Discussion of the first two of these three points will be more theoretical in nature, drawing on manuscript studies and literary theory, while that of the third will elucidate the intent of the Canon Tables by comparing them once again with Eusebius’ Chronicle. Finally, in order to illustrate concretely all of these points, I will conclude this chapter by examining a sampling of the actual cross-referencing links forged by Eusebius’ apparatus.

THE CANONIZIN G EFFECT O F THE CANON TABLES The decision to include all four of the gospels in a single codex represented a significant departure from the book culture of the scroll, since ‘Greek and The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0004

Reading the Gospels with the Eusebian Tables

97

Roman book rolls never, so far as we know, included a miscellany of works by more than one author when they were first made’¹. A four-gospel codex is, therefore, what has come to be labelled in manuscript studies a ‘Multiple-Text Manuscript’, containing as it does works composed by various authors in discrete acts of production. Research into various manuscript cultures has identified that so-called MTMs were often constructed on the basis of notional judgements about the status of their contents.² In other words, the physical form of the book tends to reflect what its creators believed about the texts contained therein. In its simplest form, this might mean that the works included in a codex constitute a canon of literature. It would, of course, be an error to insist on an unqualified correlation between form and function. A canon of literature can exist even if its constituent parts are housed in separate manuscripts, and various texts can be included together in a single manuscript without always being thought to be a single canon³. Nevertheless, in some cases a multiple-text manuscript clearly is performing this function for its users, as for example is the case with the four-gospel codex that became a standardized feature of the various late-antique and medieval Christendoms throughout the Mediterranean world. A codex with the four gospels was far more common than a pandect containing all of scripture, and its special status is evidenced by the unique role it performed in the liturgy, in church councils, and in Christian art.⁴ Therefore, the inclusion of just these four texts and no ¹ Houston, Inside Roman Libraries, 78. Houston was following the earlier work of Enzo Puglia, ‘Il catalogo di un fondo librario di Ossirinco del III d. C. (PSILaur. inv. 19662)’, Zeitschrift für Papyrologie und Epigraphik 113 (1996): pp. 51 65, who was building upon A. Petrucci, ‘Dal libro unitario al libro miscellaneo’, in Tradizione dei classici trasformazioni della cultura, ed. Andrea Giardina, Società romana e impero tardoantico (Rome: Laterza, 1986). Cf. Edoardo Crisci, ‘I più antichi codici miscellanei greci: materiali per una riflessione’, in Il codice miscella neo: Tipologie e funzioni. Atti del Convegno internazionale Cassino 14 17 maggio 2003, ed. Edoardo Crisci and Oronzo Pecere, Segno e testo 2 (Cassino: Università degli studi di Cassino, 2004), pp. 109 10. ² On this point see Alessandro Bausi, ‘A Case for Multiple Text Manuscripts Being “Corpus Organizers” ’, Manuscript Cultures Newsletter 3 (2010): pp. 34 6. ³ Cf. Edmon L. Gallagher and John D. Meade, The Biblical Canon Lists from Early Christian ity: Texts and Analysis (Oxford: Oxford University Press, 2017), p. xvi xvii. ⁴ See, e.g., Germanus, hist. 24 (Meyendorff, p. 73): ‘The entrance of the Gospel signifies the coming of the Son of God and His entrance into this world’. At the council of Chalcedon a copy of the gospels was placed in the centre of the assembled bishops to denote the presence and authority of Christ (Richard Price and Michael Gaddis, The Acts of the Council of Chalcedon, 3 vols, Translated Texts for Historians 45 (Liverpool: Liverpool University Press, 2005), vol. 1, pp. 43, 129; vol. 2, p. 125). As for art, among the many examples that might be highlighted, one of the most famous is surely the Christ Pantocrator icon from Saint Catherine’s Monastery, in which Christ holds a copy of the gospels under his left arm. A similar conclusion was reached by Martin Wallraff: ‘diesem neuen „kanonischen‟ Buch des Christentums eignet gerade als Gesamtkunstwerk etwas Sakral Normatives. Die Kanontafeln des Euseb, die Inhaltsverzeichnisse und Pinakes, die Kapiteleinteilungen und last, but not least eben auch Einband und Illustrationen haben durchaus auch eine stabilisierende und sichernde Funktion. Sie tragen zu einer Aura der Unver änderbarkeit und Unberührbarkeit des Textes bei’ (Wallraff, Kodex und Kanon, p. 48). On the book of the gospels as representing the presence of Christ, see also John Lowden, ‘The Word

98

The Eusebian Canon Tables

other surely was understood to indicate that they stood in a category alone, to the exclusion of all other texts, occupying a unique place even within the wider canon of Christian scripture. The first argument I want to advance here is that the Canon Tables intensify this canonizing effect of the codex, and that they do so in two ways.⁵ First, they perform this function as a result of the tabular form in which they represent the text of the fourfold gospel. Scholars who work on ancient mathematical tables have identified ‘completeness’ as one of the ‘tabular effects’ produced by rendering information in a matrix as opposed to some other form.⁶ Similarly, Andrew Riggsby has drawn attention to the ‘all-or-nothing quality’ of a table: ‘an item is included at a certain place or it is not’ unlike more open-ended information devices such as lists, which can be added to indefinitely.⁷ With respect to the Canon Tables specifically, the effect of the tabular form is to communicate that every component part of the fourfold gospel is accounted for somewhere within the ten Canons, so, by implication, what is not included is therefore not a part of the fourfold gospel. In other words, as has been emphasized recently by Jeremiah Coogan, Eusebius’ apparatus is a ‘totalizing’ form of paratextuality, in that it ‘clearly delimits canonical space’, excluding all other texts, both other non-canonical gospels and even other portions of the New Testament canon that present parallels to the fourfold gospel.⁸ In addition, the Canon Tables may be seen as reinforcing the canonical status of the fourfold gospel by the way in which they create a series of intertextual references across the corpus, each of which acts as a thread binding together two or more texts. All of these tiny threads cumulatively forge a strong, interconnected web of meaning amongst the gospels,⁹ with the

Made Visible: The Exterior of the Early Christian Book as Visual Argument’, in The Early Christian Book, ed. William E. Klingshirn and Linda Safran, CUA Studies in Early Christianity (Washington, DC: The Catholic University of America Press, 2007), p. 28; Claudia Rapp, ‘Holy Texts, Holy Men, and Holy Scribes: Aspects of Scriptural Holiness in Late Antiquity’, in The Early Christian Book, pp. 196 200; Reudenbach, ‘Der Codex als heiliger Raum’, pp. 71 3. ⁵ The conceptual relationship implied by the two uses of ‘canon’ in this sentence is, of course, a mere felicitous coincidence. As argued in chapter one, when Eusebius uses κανών to refer to the Canon Tables he simply means the word in the sense of ‘table’. Conversely, when Eusebius talks about the church’s accepted and rejected books in HE 3.25, he does not use the word κανών at all. ⁶ Ossendrijver et al., ‘Tables in Antiquity’. ⁷ Riggsby, Mosaics of Knowledge, chapter 2. Riggsby here included lists and tables as both having this ‘all or nothing quality’ but it seems to me that within these two types there is a further gradation, with tables being more exclusive than lists. On this feature of lists see also Umberto Eco, The Infinity of Lists: From Homer to Joyce, trans. Alistair McEwen (London: MacLehose Press, 2012). ⁸ Jeremiah Coogan, ‘Mapping the Fourfold Gospel: Textual Geography in the Eusebian Apparatus’, Journal of Early Christian Studies 25 (2017): pp. 352 4. ⁹ Wallraff, Kodex und Kanon, pp. 34 5 pointed out that for this reason, the term ‘text’ is particularly appropriate for the Canon Tables.

Reading the Gospels with the Eusebian Tables

99

result that, from a purely literary, non-theological perspective, it becomes apparent that the four gospels stand in a unique relationship to one another and so constitute a special body of literature.¹⁰ One way of testing this argument would be to imagine whether or not it would have been possible to create a similar cross-referencing paratext if one was using one of the noncanonical gospels in place of one of the four canonical gospels. Could one, for example, create a system of Canon Tables for a codex that comprised the Gospels of Matthew, Mark, Luke, and Mary? It is hard to see how one might do so given how very different the Gospel of Mary is in style and content from the first three, a more radical divergence than the already pronounced difference between John and the Synoptics. The problem would be even more acute if, within this hypothetical codex, one replaced the Gospel of Mary with the Gospel of Truth, the Gospel of Philip, or the Gospel of Judas. The situation would become somewhat easier if one were to use the Gospel of Thomas, since it overlaps in places with the Synoptics, but this gospel consisting entirely of dominical sayings is still largely unique material, and would present nothing parallel to any of the narrative portions of Matthew, Mark, and Luke. The only two non-canonical gospels that might have worked in an alternative system of Canon Tables are the Gospel of Peter and the Egerton Gospel, both of which present some parallels, and on occasion even verbatim agreements, with the canonical gospels, though the fragmentary nature of these two texts precludes the possibility of making judgements about their overall nature. Eusebius’ Canon Tables, therefore, demarcate the fourfold gospel canon by drawing attention to the unusual intertextual proximity that exists amongst these four texts, which is apparent from a straightforward reading of them, irrespective of how one accounts for the similarities that exist.¹¹ ¹⁰ My argument here is analogous to, and in agreement with, the more extensive and theoretically rigorous proposal of Will Kynes that genre be defined according to intertextual relations: Will Kynes, An Obituary for ‘Wisdom Literature’: The Birth, Death, and Intertextual Reintegration of a Biblical Corpus (Oxford: Oxford University Press, 2018), chapter 4, ‘The Universe of Texts: The Intertextual Network of Genres from Multiple Perspectives’. Kynes’ argument that genre may be defined intertextually seems to me easily adapted into an argument that a canon similarly can be defined intertextually. ¹¹ A parallel to my argument here is the way in which Mark Goodacre has described the attempt to include the Gospel of Thomas alongside Matthew, Mark, and Luke as a new Synoptic Problem: ‘the Synoptic Problem does have a unique profile. Its fascination lies in the fact that these three Gospels are so very close to one another, both in terms of high verbatim agreement and order . . . There is simply no doubt that there is a literary link of some kind between Matthew, Mark and Luke, and it is this agreed premise that provides the platform for the investigation of the inter relationship between them. Even though I am persuaded that Thomas knows the Synoptics, its agreement with them is not on the same scale as their agreement among them selves, and the same is true of the Gospel of Peter the verbatim agreement is nothing like as impressive in scope or volume’ (Mark Goodacre, ‘Did Thomas Know the Synoptic Gospels? A Response to Denzey Lewis, Kloppenborg and Patterson’, Journal for the Study of the New Testament 36 (2014): p. 291). Goodacre, of course, is discussing only the Synoptics and excluding John, but I think the analogy still holds.

100

The Eusebian Canon Tables

HYPERTEXTUAL READING WITHIN A BO UNDED CORPUS Building upon the previous point, I want to argue that within this bounded, canonical space, the Canon Tables encourage a hypertextual mode of reading. Before coming to what I mean by this term, however, it will be useful to observe something that the Canon Tables are not doing. Eusebius’ paratext does not provide any single fixed route through the overlapping narratives of the fourfold gospel, and this is in fact a part of the genius of its design. A persistent temptation in the reception history of the canonical gospels is to create from the fourfold diversity a singular gospel, represented most famously by Tatian’s socalled Diatessaron. Though we know frustratingly little about what motivated Ammonius’ Diatessaron-Gospel, the dismembering effect that it had upon the latter three gospels was similarly destructive in terms of the internal integrity of each component of the fourfold witness. The fundamental difference between these earlier attempts at gospel unity and Eusebius’ paratext is that the Canon Tables do not dissolve the diversity of the four texts into some primal or reconstructed unity, but rather leave it intact. Also important in this regard is the fact that the parallel passages juxtaposed in each of the tables do not proceed according to any narrative sequence. Rather, the sequence of parallels within each table simply follows the increasing numbers for the sections in the Gospel of Matthew for Canons I–VII and those for Luke for Canons VIII and IX. But even if one started at the beginning of a single table and read all of the passages it contained in sequence, the farraginous succession of fragmentary episodes and sayings would hardly suggest itself as a compelling narrative sequence that might rival the distinct ‘itineraries’ of the four.¹² Instead the apparent randomness of the numbers, and therefore of the passages that they represent, ensures that the Canon Tables perform their paratextual function remarkably well by constantly directing attention back to the four gospels to which they refer, ever subservient to those texts and never threatening to rival their centrality.¹³

¹² I use the language of ‘itineraries’ under the influence of Coogan, ‘Mapping the Fourfold Gospel’, who insightfully applied Michel de Certeau’s concept of textual geography to the Eusebian apparatus. ¹³ Cf. Coogan, ‘Mapping the Fourfold Gospel’, p. 352: ‘With its manifold intersections between gospels, Eusebius’s canonical map also defies the possibility of a single chronological itinerary. Indeed, Eusebius displays a remarkable unconcern about time, and does not attempt to place the gospel material into any overarching sequence. His map correlates four textual itineraries. As a result, Eusebius’s method respects the narrative structures of all four gospels as literary compositions, different ways for the reader to walk through the life of Jesus’. For this reason D. L. Dungan was incorrect to lump Eusebius’ Canon Tables into the category of ‘harmonies’, along with the works of Ammonius, Tatian, and Augustine, all of which have as ‘their intent . . . to arrange the Gospel material so that a harmonized chronological narrative results, giving an orderly account of “the life and ministry of Christ” ’ (Dungan, ‘Theory of Synopsis Construction’, 306). This, of course, is precisely what Eusebius’ paratext does not do.

Reading the Gospels with the Eusebian Tables

101

Yet, even though the Canon Tables leave the original ‘itineraries’ of the four intact, they are also, in the words of Jeremiah Coogan, ‘productive of new itineraries through the fourfold canonical gospel’,¹⁴ by virtue of their creation of new textual links across the canon. It is the reader’s experience of travelling along these intertextual itineraries that I want to describe as ‘hypertextual’. Although this term has never before been used specifically with reference to the Eusebian apparatus, it has previously been adapted from its usage in the modern period in order to describe the shift in book culture that occurred in late antiquity: Scrolls were designed for information storage in libraries; they slotted neatly into horizontal slots, with an identification tag visible at the end. The codex, however, allowed for quick scanning back and forth across several pages an early form of what we now call “hypertextuality” . . . .The Christian Bible, in particular, was a text that many exegetes wanted to read hypertextually, as they grappled with its multiple authors and often conflicting demands.¹⁵

This movement back and forth across the codex to find relevant texts, which König and Whitmarsh term ‘hypertextuality’, is apt as a description of the style of reading encouraged by Eusebius’ apparatus. Moreover, the analogy between online media and the Eusebian Canons has been observed before. James J. O’Donnell, Anthony Grafton, and Meredith Williams have referred to the Canon Tables as ‘the world’s first hot links’, since they function as ‘extraordinarily original and effective information retrieval devices’.¹⁶ Given my usage of Gérard Genette’s poetics in the first chapter of this book, I should point out here that Genette also included hypertextuality as another of his five types of transtextuality: ‘By hypertextuality I mean any relationship uniting a text B (which I shall call the hypertext) to an earlier text A (I shall, of course, call it the hypotext), upon which it is grafted in a manner that is not that of commentary’. Genette’s sense of this term has in view specifically the production of a literary work. Hypertextuality is a way of studying ‘a text in the second degree’ in which text B is ‘not speaking of text A at all but [is] unable to exist, as such, without A, from which it originates through a process I shall provisionally call transformation, and which it consequently evokes more or less perceptibly without necessarily speaking of it or citing it’. The examples he provides to explain this phenomenon are the Aeneid and Ulysses, which do not relate to the hypotext of the Odyssey in the

¹⁴ Coogan, ‘Mapping the Fourfold Gospel’, p. 346. ¹⁵ König and Whitmarsh, ‘Ordering Knowledge’, p. 34. ¹⁶ These phrases are taken from Grafton and Williams, Christianity and the Transformation, p. 199, who acknowledged their debt to O’Donnell and referenced a website that, rather ironically, now seems to be a dead link. Cf. Wallraff, Kodex und Kanon, p. 33.

102

The Eusebian Canon Tables

form of a commentary (i.e., ‘metatextuality’) but instead arise from that earlier work ‘by a transformative process’.¹⁷ Viewed in the light of Genette’s comments about the Odyssey and its two later hypertexts, the four gospels—or at least the three Synoptics—may seem at first to be an excellent example of hypertextuality, given how nearly all scholars today agree that the striking similarities of these texts to one another require the hypothesis of some kind of literary dependency among them. Although a variety of solutions to the Synoptic Problem continue to be debated, all of them presuppose that later evangelists used earlier gospels as hypotexts upon which to effect literary transformations resulting in new hypertexts, texts, that is, whose existence would be unimaginable in the absence of the text(s) from which they originated. At this point, however, we encounter a significant difference between Genette’s notion of hypertextuality and the Eusebian apparatus—Eusebius, like virtually everyone else prior to Griesbach, did not view the relationships among the gospels in terms of literary dependence, and, as a result, he did not envision the comparative analysis enabled by his paratext as an exercise in source criticism.¹⁸ Nevertheless, I would suggest that the term ‘hypertextuality’ can still be fruitfully applied to the Eusebian apparatus if, along with König and Whitmarsh, we think of it as denoting a style of reading rather than a theory of textual production. Seen in this light, hypertextual reading is a mode of engagement in which the reader goes beyond (ὑπέρ) a focal text by considering it in the light of other texts. The Eusebian apparatus is, therefore, a paratext that creates intertextual links and then encourages hypertextual reading

¹⁷ Genette, Palimpsests, p. 5. For an attempt to appropriate Genette’s notion of hypertextuality and apply it to Eusebius’ corpus, see Jeremy M. Schott, ‘Textuality and Territorialization: Eusebius’ Exegeses of Isaiah and Empire’, in Eusebius of Caesarea: Tradition and Innovations, ed. Aaron Johnson and Jeremy Schott, Hellenic Studies 60 (Washington, DC: Center for Hellenic Studies, 2013), pp. 174 5. ¹⁸ Against Satoshi Toda, ‘The Eusebian Canons: Their Implications and Potential’, in Early Readers, Scholars, and Editors of the New Testament: Papers From the Eighth Birmingham Colloquium on the Textual Criticism of the New Testament, ed. H. A. G. Houghton (Piscataway, NJ: Gorgias Press, 2014), pp. 31 2, who assumed Eusebius did envision one evangelist using another as a source and that the Canon Tables were intended to represent such relationships of dependency. Eusebius’ Canon Tables could, of course, be used with this end in view. The point here is simply that there is no reason to suppose that he thought of the goal in these terms, since no one prior to Griesbach seems to have done so. Augustine is sometimes presented as an early exception to this claim, on the grounds that in his De consensu evangelistarum he did eventually come to view the gospels as being literarily dependent upon one another. See, however, my analysis of this argument in the next chapter. Eusebius’ lack of concern for source criticism is also noted by Watson, The Fourfold Gospel, p. 115, who commented, ‘The almost complete absence of genetic hypotheses in the ancient church represents not so much a mistake about the true nature of the gospels, needing to be corrected by modern scholarship, but rather a decision about how the canonical gospels are to be read as independent yet parallel texts whose interrelations justify both their continued existence as separate entities and their incorporation into a single definitive embodiment of the gospel message’.

Reading the Gospels with the Eusebian Tables

103

in which the user attends closely to intra-canonical relationships. It is important at this point, however, to emphasize a fundamental discontinuity between the hypertextuality of the Internet and the hypertextuality of Eusebius’ apparatus. Unlike the fourfold gospel, the Internet is not a bounded corpus of literature centred upon a single character, but a seemingly endless web of information.¹⁹ Moreover, hyperlinks do not always refer the reader to another web page with similar content, but often instead function more like footnotes that reference fuller discussions of tangentially related material. As a result, a reader who clicks on a hyperlink ‘is liable to become restless, to become a surfer looking for satisfaction down an endless chain of links, rather than finding it in the window opened up by the text itself ’.²⁰ In contrast, the Canon Tables imply that the reader should view a certain closed set of passages alongside one another, as parallel statements requiring a single act of interpretation.²¹ In this way, the Eusebian apparatus exerts a hermeneutical pressure upon the reader. The key theoretical insight here is the awareness that the meaning of a text emerges through its relationship to other texts. As Christian Vandendorpe has written, The text produces its meaning and its effects through the connection of blocks of text at various levels. . . . In principle the more levels of connection a text has, the greater the potential production of meaning. The traditional superiority of written language over other forms of expression arises from the fact that text was from the outset conceived in terms of reading, that is, it was designed for reception by an addressee. Also, the very nature of its medium grants readers complete mastery over all its components; every element of the text can be isolated, analyzed, and placed in relation to other elements of the same text, which permits a thorough analysis of the intended or possible meanings. Both on the material level and on the level of significations, the book has a layered structure, which gives readers a grasp over the content that film or other forms of spectacle cannot offer.²² ¹⁹ For an argument that the much vaunted properties of reading hypertext ‘provide an inhospitable context for literary reading’ since they ‘enforce repetition, discontinuity, and reductiveness’ see David S. Miall, ‘Confounding the Literary: Temporal Problems in Hypertext’, in From Codex to Hypertext: Reading at the Turn of the Twenty First Century, ed. Anouk Lang, Studies in Print Culture and the History of the Book (Amherst, MA: University of Massachusetts Press, 2012), pp. 203 16, at p. 213. Miall’s criticisms concerning reading online hypertext appear to me to be largely inapplicable to Eusebius’ apparatus, primarily because of the bounded nature of the hypertextuality the system encourages and the unobtrusiveness of its marginal annotation. ²⁰ Miall, ‘Confounding the Literary’, p. 205. ²¹ It should be noted that Genette explicitly disavowed any sort of ‘hypertextual hermeneut ics’, asserting ‘I view the relationship between the text and its reader as one that is more socialized, more openly contractual, and pertaining to a conscious and organized pragmatics’ (Genette, Palimpsests, p. 9). ²² Vandendorpe, From Papyrus to Hypertext, pp. 56 7. Cf. the bold proclamation of Harold Bloom, ‘there are no texts, but only relationships between texts’ (Harold Bloom, A Map of Misreading (New York: Oxford University Press, 1975), p. 3, emphasis original). I am grateful to Will Kynes for drawing my attention to this passage.

104

The Eusebian Canon Tables

The hermeneutical pressure of the Eusebian apparatus derives from the fact that it makes apparent the layered structure of meaning within the fourfold canon by disclosing the otherwise implicit connections between discrete blocks of text. In short, what the Eusebian Canons are saying to the reader is that each of the gospels in a fourfold gospel codex no longer has the same meaning it would have if it were housed in a single codex coterminous with a canon of only one member. Rather, this juxtaposition of four texts between two covers, understood as constituting a closed canon of sacred literature, now has a new meaning as a result of this editorial process. By reading hypertextually, attending to the intertextual relationships between similar passages, one sees that the meaning of each passage is qualified by the presence of those to which it is connected, leading to a multiplicity of perspectives on the common subject matter about which those passages speak.²³ In other words, through their juxtaposition of formerly separate passages the Canon Tables are in fact creating new meaning, or—perhaps better put— they are creating the conditions for a reader to construct new meaning from the textual juxtaposition.²⁴ It is of course true that there is nothing strictly new that the Canon Tables provide. A reader of the gospels with a sufficiently thorough awareness of all four texts could jump back and forth between texts as he or she reads, accomplishing the same textual juxtaposition. What Eusebius’ paratext does, then, is to provide the reader with a significant portion of this task already accomplished, effectively offloading much of the mental burden by inscribing within the codex the cognitive process that would otherwise be required of the reader.²⁵ As a result, the reader has more mental space free to attend to the more important and creative task of interpreting the passages at hand. Another effect of this offloading is that the apparatus acts as a persistent reminder of just how complex and rich are the interrelations between the four gospels. The marginal annotation throughout the fourgospel codex serves as a constant invitation to the reader to go beyond the text of an individual gospel and confront the question of how to relate these four narratives to one another.

²³ Similarly Guy Stroumsa asserted that as a result of the two part nature of the Christian scriptures (i.e., Old and New Testaments), ‘intertextuality is built into the system’ (Guy G. Stroumsa, Barbarian Philosophy: The Religious Revolution of Early Christianity, WUNT 112 (Tübingen: Mohr Siebeck, 1999), p. 35). For an exploration of the theological implications of the multilayered nature of the Christian scriptures, see especially Joseph Ratzinger, God’s Word: Scripture Tradition Office, ed. Peter Hünermann and Thomas Söding, trans. Henry Taylor (San Francisco, CA: Ignatius Press, 2008), pp. 41 67. ²⁴ On juxtaposition as a way of producing meaning, see also Francis Watson, ‘Towards a Redaction Critical Reading of the Diatessaron’, Early Christianity 7 (2016): p. 111, who focused on the way in which gospel writers juxtaposed material from different sources. ²⁵ I use the term ‘offloading’ under the influence of Riggsby, Mosaics of Knowledge, ‘A Brief Orientation’, who adopted it from cognitive science to describe the shift from ‘purely symbolic reasoning to bodily action, especially interaction with the physical world’.

Reading the Gospels with the Eusebian Tables

105

The concern that such a claim will arouse in some is that this kind of reading results in the distinct message of a given gospel being effectively effaced by a more dominant neighbour. Concretely, the worry is, for example, that Mark would always be read through the lens of Matthew or Luke, rather than on its own terms. Jeremy M. Schott has provocatively expressed this point, using postcolonial theory to argue that Eusebius’ paratext ‘asks readers to colonize the text of Mark with the text of Matthew, populate Luke with John, and so forth’ all as a process of ‘territorialization’.²⁶ This danger is real, but unless one is willing to remove the presupposition of the fourfold canon, it can only be negotiated, not eliminated. The latter, of course, was not an option for Eusebius, who had received the fourfold gospel as the patrimony of two and a half centuries of gospel writing, redaction, and collection.²⁷ For Eusebius, and for those who like him view the fourfold gospel as canonical, the formal diversity of the church’s gospel canon demands some kind of hypertextual reading, and it is the great merit of Eusebius’ paratext that it facilitates this process so extremely well. Moreover, this mode of reading need not be construed as such a reductive undertaking. A more positive appraisal is possible if one concedes that the juxtaposition of passages from different texts creates the potential for new meaning out of the old, dependent upon and yet not reducible to its component parts, and if one considers that a range of practices is possible under the broad rubric of hypertextuality.

THE INDETERMINATE QUALITY OF EUSEBIUS ’ PARALLELS In the next section of this chapter I want to set aside one possible implication of the argument that has been pursued thus far. In the first two chapters I suggested that the Eusebian apparatus was intended to make apparent the divinely ordered harmony of the inspired text. Combined with the argument of this chapter that the Canon Tables have a ‘canonizing’ effect, such a claim might seem to imply that Eusebius’ project was primarily reductive and defensive, designed merely to lead to predetermined exegetical outcomes, rather than creating new or unexpected ones. Such a reading would align well with the notion, which has occasionally been suggested in prior scholarship, that his ²⁶ Schott, ‘Textuality and Territorialization’, p. 186. ²⁷ For a survey of the traditions about gospel origins that Eusebius preserved and passed on in his Ecclesiastical History, see Philip Sellew, ‘Eusebius and the Gospels’, in Eusebius, Christianity, and Judaism, ed. Harold W. Attridge and Gohei Hata, Studia Post Biblica 42 (Leiden: E. J. Brill, 1992); Francis Watson, Gospel Writing: A Canonical Perspective (Grand Rapids, MI: Eerdmans, 2013), pp. 436 52.

106

The Eusebian Canon Tables

creation of the Canon Tables was inspired by the severe criticism levied against the gospels by the Neoplatonist Porphyry in his now fragmentary treatise, Against the Christians. For example, in his book, A History of the Synoptic Problem, David Laird Dungan presented the Canon Tables as one component of Eusebius’ four-part response to Porphyry, along with his Preparation for the Gospel, Demonstration of the Gospel, and Ecclesiastical History.²⁸ Though seemingly without being aware of Dungan, Thomas O’Loughlin has reached a similar conclusion, asserting that ‘the main motive behind its [i.e., the Canon Tables] creation’ was the ‘apologetic’ attempt to ‘refute’ critics of the gospels by ‘showing that the accounts can be read consistently with one another’,²⁹ and in two later studies he specifically highlighted Porphyry’s attacks on the gospels as the likely occasion for Eusebius’ paratext.³⁰ In response to these scholarly assertions, it is important first to point out that it is not at all clear that Porphyry’s Against the Christians focused extensively on contradictions across multiple gospels of the kind that the Canon Tables make apparent.³¹ He seems instead primarily to have highlighted ²⁸ Dungan, A History of the Synoptic Problem, pp. 98 111. For a critique of the common view that the main purpose of the Demonstratio evangelica and Praeparatio evangelica was to oppose Porphyry, see Sébastien Morlet, ‘Eusebius’ Polemic Against Porphyry: A Reassessment’, in Reconsidering Eusebius: Collected Papers on Literary, Historical, and Theological Issues, ed. Sabrina Inowlocki and Claudio Zamagni, VCSup 107 (Leiden: Brill, 2011). ²⁹ O’Loughlin, ‘The Eusebian Apparatus in Some Vulgate Gospel Books’, p. 3. ³⁰ O’Loughlin, ‘Harmonizing the Truth’, pp. 10 11; O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, p. 107. ³¹ I realize that this assertion runs counter to what is commonly assumed and stated about Porphyry’s project, and that I do not have space to defend the claim adequately here. Charac terizing the nature of Porphyry’s criticism naturally depends upon how one reconstructs his treatise from the scattered fragments and testimonia that survive in Christian sources, a matter about which there is ongoing scholarly dispute. For recent surveys of this problem see Sébastien Morlet, ‘Comment le problème du Contra Christianos peut il se poser aujourd’hui?’, in Le traité de Porphyre contre les chrétiens. Un siècle de recherches, nouvelles questions. Actes du colloque international organisé les 8 et 9 septembre 2009 à l’Université de Paris IV Sorbonne, ed. Sébastien Morlet, Collection des Études Augustiniennes. Série Antiquité 190 (Paris: Institut d’Études Augustiniennes, 2011), pp. 11 49, and Ariane Magny, Porphyry in Fragments: Reception of an Anti Christian Text in Late Antiquity (Burlington, VT: Ashgate, 2014). Relying on von Har nack’s edition of Against the Christians, Helmut Merkel, Die Widersprüche zwischen den Evangelien. Ihre polemische und apologetische Behandlung in der Alten Kirche bis zu Augustin, WUNT 13 (Tübingen: Mohr Siebeck, 1971), pp. 13 18, asserted that Porphyry undertook a ‘minuziösen Einzeluntersuchung des Neuen Testaments’, including criticisms concerning contradictions between gospels, but most of the fragments he focused upon come from the treatise of Macarius Magnes and are formally anonymous. Similarly, Robert M. Berchman claimed that Porphyry ‘compared the different gospel testimonies with each other’ by relying on ‘nuanced literary and historical criticisms drawn from a gospel parallels [sic]’ (Porphyry Against the Christians, p. 115). However, when one examines the collection of supposedly Porphyrian material Berchman assembled, it is apparent that in none of those passages in which authors actually name Porphyry as the source of a criticism is it ever the case that the critique in view is based on a contradiction among multiple gospels. The only passages in his collection that might support his claim are those that he includes from Augustine’s De consensu evangelistarum, though it is not at all obvious that Porphyry, or even any other specific critic,

Reading the Gospels with the Eusebian Tables

107

claims made by the gospels that would strike one schooled in Neoplatonism as absurd, or errors and inconsistencies within a single gospel. Yet, even if Porphyry or some other critic had emphasized gospel contradictions and even if the Canon Tables had begun as an apologetic project responding to such criticism, they exceed this narrow beginning and in fact are profoundly underdetermined in terms of what they imply about how one should interpret the passages juxtaposed through this system of cross-reference. My argument here is analogous to one that has been made by R. W. Burgess about Eusebius’ Chronicle. Responding to the claim of T. D. Barnes that the Chronicle was ‘primarily a work of pure scholarship’,³² Burgess argued convincingly that the work had an apologetic aim, as can be seen, for example, in Eusebius’ allusion to Porphyry’s Against the Christians in the preface to book two of the work (i.e., the Χρονικοὶ Κανόνες).³³ Nevertheless, Burgess conceded that the Chronicle was ‘a work that surpassed the narrow confines of a straight rebuttal to Porphyry’ and that it demonstrated that Eusebius was not only ‘a mere apologist’ but also ‘a scholar and historian unique to his age’.³⁴ Further complicating a view of the Chronicle as simply an apologetic work is Eusebius’ remarkable awareness of the ineliminable uncertainty intrinsic to historical investigation, a point that has been emphasized most clearly by Anthony Grafton and Meredith Williams. Whereas the second book of the Chronicle, consisting of the Chronological Tables, implies an ambitious view that diverse national histories can be coordinated with one another, Eusebius’ comments in book one are much more measured, owing primarily to two realities. First, he recognized the impossibility of reconciling the accounts of primeval history

served as the foil for books two to four of this treatise. Cf. the critical remarks about Berchman’s edition at Morlet, ‘Comment le problème’, pp. 33 6, and on De consensu, cf. my discussion in the following chapter. The new edition of Contra Christianos produced by Matthias Becker pru dently sets aside the anonymous material from Macarius Magnes that others had assumed derived from Porphyry. For a survey of Porphyry’s treatment of various New Testament texts, see John Granger Cook, The Interpretation of the New Testament in Greco Roman Paganism, STAC 3 (Tübingen: Mohr Siebeck, 2000), pp. 103 67, who similarly stated that ‘Finding “contraries” was one of Porphyry’s best loved methods of attacking the New Testament’ (p. 135), though again in his survey there are no passages of indisputably Porphyrian origin which highlight contradictions between gospels (see the possible exceptions at pp. 137, 147). I am grateful to Francis Watson for drawing my attention to this point. ³² Barnes, Constantine and Eusebius, p. 113. ³³ Burgess, ‘The Dates and Editions’, especially pp. 488 95. The relevant passage from Eusebius’ Χρονικοὶ Κανόνες is preserved in George Syncellus, chron. p. 73 (Mosshammer). See the translation of the opening of the preface in appendix 2 of Burgess, ‘The Dates and Editions’, pp. 503 4. Whereas Eusebius simply referred to Porphyry as ‘that man’, Jerome, in his Latin translation of the Chronicle, made the reference explicit by naming him. See Jerome, chron., praef. (GCS 47, pp. 7 9). Eusebius also discussed the chronological problem, naming Porphyry explicitly, in PE 10.9.11 25. Recently Burgess has reaffirmed his anti Porphyrian argument: Burgess and Kulikowski, Mosaics of Time, p. 120. ³⁴ Burgess, ‘The Dates and Editions’, pp. 495, 497.

108

The Eusebian Canon Tables

that he found among the various national chronologies that he investigated.³⁵ Secondly, as a result of having Origen’s Hexapla at his disposal, he was better informed than most regarding the divergent patriarchal lifespans in the various recensions of the Hebrew scriptures, which similarly defied a simple resolution.³⁶ It was for these two reasons that the tables in the second half of the work began with Abraham and counted forward, rather than trying to count back to the creation of the world, as other Christian chronographers, both before and after Eusebius, have done. This strategy conveniently sidestepped the problem, but the recognition of chronological uncertainty lingered, casting a shadow across the overall tenor of Eusebius’ Chronicle. Eusebius was surprisingly frank about these difficulties, and provided justification for his more restrained approach by citing Jesus’ assertion in Acts 1:7 that no one knows the hour of his return. This passage is obviously a reference to the future, but Eusebius extended its applicability to refer to ‘all time’ (վասն ամենայն ժամանակաց), both past and present, with the implication that a ‘chronology of the whole world’ (զամենայն աշխարհի զժամանակագրութիւն), whether based on Greek, barbarian, or even Hebrew historical sources, is simply impossible.³⁷ Even starker are the two general conclusions he wished his readers to draw from his work, which he stated shortly after the citation of Acts 1:7, at the beginning of book one of the Chronicle: first, that no one should imagine, as those others have, that he can achieve absolute precision in chronological calculation and so deceive himself; but, second, that one may realize that, after undertaking an investigation, it is possible to know the answer to a given question, and not have any doubt.³⁸ Մի զի մի այլոցն ոք նմանեալ կարծիցէ ճըշդիւ գտակաւ հասանել համարոյ ժամանակացն եւ խաբիցի: Բայց զայն եւ եթ ՚ի հանդիսի անցուցեալ գիտիցէ, թէ որպէս զիարդ մարթ իցէ գիտել զխնդիրս որ առաջի կայ, եւ չլինել երկմիտ

The ‘others’ whom Eusebius has in mind are previous Christians (especially Julius Africanus) who had applied themselves to the work of chronology and arrived at precise predictions about the end times, so his comment in this passage is partly motivated by the fact that his eschatology differs from that of his predecessors. Nevertheless, Grafton and Williams were surely right that this admission of uncertainty—of the ever unfinished nature of the task of historical investigation—was also born from Eusebius’ close attention to the

³⁵ Cf. Grafton and Williams, Christianity and the Transformation, pp. 160 6. ³⁶ Cf. Grafton and Williams, Christianity and the Transformation, pp. 154 8, 168 70. ³⁷ Eusebius, chron. (Aucher, pp. 4 5). See the discussion of this passage in Grafton and Williams, Christianity and the Transformation, pp. 152 3. ³⁸ Eusebius, chron. (Aucher, p. 5).

Reading the Gospels with the Eusebian Tables

109

insoluble problems presented by the sources themselves. Commenting on this passage, Grafton and Williams asserted, Nothing was more distinctive in Eusebius’s approach than the Bakhtinian open ness that he showed here, his willingness to turn his early books into so odd a conversation among priests of several nations and to accept that their Pinteresque dialogue necessarily ended in uncertainty. Like Origen, he produced not only a synthesis, but also a polyglot collection of research material from which other scholars could draw what conclusions they liked.³⁹

In other words, Eusebius did not attempt to resolve all the problems in his sources because he knew he could not do so, but instead presented them as a literary dialogue of his own making, inviting his readers to join him in an attempt to attain some further degree of historical understanding by critically evaluating the competing voices. I suggest that the same basic approach can be seen in the Canon Tables for the fourfold gospel, that they too exhibit this same ‘Bakhtinian openness’. Far from resolving the dialogic tension inherent to the fourfold gospel, the Eusebian apparatus both formalizes it and makes it impossible to ignore, ensuring that the reader is ever attentive to the multiple voices present within the canon.⁴⁰ To be sure, Eusebius himself may have believed that it was possible to harmonize all the apparent contradictions in the gospels, and in other works he engaged in this kind of project, albeit on a limited scale.⁴¹ ³⁹ Grafton and Williams, Christianity and the Transformation, p. 166. Note that in their citation of the relevant passage from book one of the Chronicle, Grafton and Williams do not include the final clauses about finding an answer and being free from doubt (pp. 167 8), perhaps giving the impression that the Chronicle was rather more ‘open’ than Eusebius intended. Though he wished to avoid excessively precise claims and did not resolve all of the historical problems presented by his sources, the aforementioned passage certainly implies that he thought his readers could find answers and achieve some certainty by consulting his work. In fact, the two stated aims of his treatise stand in tension with each other, since Eusebius asserted that he wanted his readers to be free from doubt, but also criticized the degree of certitude assumed by Africanus. ⁴⁰ The adjective ‘Bakhtinian’ of course alludes to the literary theorist Mikhail Bakhtin, who proposed that ‘any concrete discourse (utterance) finds the object at which it was directed already as it were overlain with qualifications, open to dispute, charged with value, already enveloped in an obscuring mist or, on the contrary, by the “light” of alien words that have already been spoken about it. It is entangled, shot through with shared thoughts, points of view, alien value judgements and accents. The word, directed towards its object, enters a dialogically agitated and tension filled environment of alien words, value judgments, and accents, weaves in and out of complex interrelationships, merges with some, recoils from others, intersects with yet a third group’ (M. M. Bakhtin, ‘Discourse in the Novel’, in Dialogic Imagination: Four Essays, ed. Michael Holquist (Austin, TX: University of Texas Press, 1981), p. 276). For an introduction to Bakhtin’s notion of a polyphonic text and dialogic truth, along with its application to a biblical text, see Carol A. Newsom, The Book of Job: A Contest of Moral Imaginations (Oxford: Oxford University Press, 2009), pp. 21 31. I am grateful to Devin White for directing my attention to Newsom’s work. ⁴¹ I have in mind here particularly his Gospel Problems and Solutions, which are focused solely on questions arising from the nativity accounts and resurrection appearances. On this work, see the following studies: Merkel, Die Widersprüche, pp. 130 46; Claudio Zamagni, ‘Eusebius’

110

The Eusebian Canon Tables

Moreover, one will look in vain for any explicit statement from him about the gospels comparable to the aforementioned one about history (e.g., that ‘absolute precision’ in harmonizing all the details of the gospels is impossible). Nevertheless, Eusebius’ Canon Tables do not, as is often supposed, harmonize the gospels at all but instead simply juxtapose passages and allow the reader to make of them what she will.⁴² No doubt a Christian reader, like Eusebius himself, will come to these passages with a presupposition of canonical coherence rather than dissonance, and will seek some way of reconciling apparently discordant passages (and there are a variety of ways of attempting to do so, some more sophisticated than others). Nevertheless, the paratext itself can also be used as a tool to attend to the diversity and even contradiction within the canon. Hence, although O’Loughlin has written, ‘once the Apparatus had been produced, anyone seeking to refute an opponent like Porphyry would have the texts laid out before him in an apparently compelling way, and have had his research made much easier’,⁴³ it is also the case that Eusebius’ apparatus would equally have made the critic’s own task much easier. If one wanted to find all the apparent contradictions in the gospels, some sort of cross-referencing system like the Canon Tables would be required, and the remarkable thing is that Eusebius’ apparatus is just as open to a deconstructive project attempting to undermine the credibility of the gospels as it is to the expected apologetic response from Christian readers.⁴⁴ As Grafton and Williams said about the Chronicle, so too the Canon Tables are a ‘collection of research material from which other scholars could draw what conclusions they liked’.⁴⁵

Exegesis Between Alexandria and Antioch: Being a Scholar in Caesarea (a Test Case From Questions to Stephanos I)’, in Reconsidering Eusebius: Collected Papers on Literary, Historical, and Theological Issues, ed. Sabrina Inowlocki and Claudio Zamagni, VCSup 107 (Leiden: Brill, 2011), pp. 151 76. See also Eusebius’ resolution of the divergent chronologies of the Synoptics and John at HE 3.24.7 14. On the rhetorical background to the criticisms of the gospels to which Eusebius responded, see Allan E. Johnson, ‘Rhetorical Criticism in Eusebius’ Gospel Questions’, Studia Patristica 18 (1983): pp. 33 9. ⁴² Despite his argument for an anti Porphyrian origin of the Canon Tables, Dungan ended his chapter with a similar statement, conceding that Eusebius avoided ‘literal harmonization’ by leaving the gospels ‘in their original literal diversity’ (Dungan, A History of the Synoptic Problem, p. 111). ⁴³ O’Loughlin, ‘Harmonizing the Truth’, p. 11. ⁴⁴ Similarly Chesnut, The First Christian Histories, p. 119, argued that the ‘purely utilitarian aim’ of the Canons, which drew upon Eusebius’ library expertise, was not occluded by his apologetic or theological goals. ⁴⁵ The argument here does not stand in disagreement with the contention of Watson, The Fourfold Gospel, p. 123 that ‘Eusebius’s canons bring to light new possibilities for a canonical reading of the fourfold gospel, one that highlights not only its diversity but also its coherence’. My point is simply that Eusebius’ Canons are open to a variety of uses.

Reading the Gospels with the Eusebian Tables

111

Supporting evidence for this argument is the complete lack of any polemical intent in Eusebius’ only comment about the purpose of his apparatus.⁴⁶ Indeed, a striking omission from the Letter to Carpianus is any statement explaining precisely how the corresponding passages are to be interpreted. Eusebius’ language about the parallels he provides is minimal. He calls them ‘concordant sections’ (τὰς ὁμοϕώνους . . . περικοπάς), in which the evangelists ‘were moved by love for the truth to speak about the same things’ (κατὰ τῶν αὐτῶν ἠνέχθησαν ϕιλαλήθως εἰπεῖν). Further on he says that these are places wherein the evangelists ‘have spoken about similar matters’ (τὰ παραπλήσια εἰρήκασιν).⁴⁷ This terminology is sufficiently vague that it does not prescribe any defined way of relating the passages in question, and, in this important respect, Eusebius’ system of cross-references is underdetermined. It simply juxtaposes certain ‘similar’ passages and leaves it to the judgement of the reader what to do with the textual data supplied. This raises the question of what we should call these passages that his apparatus joins together. I have so far used the term ‘parallels’ and will continue to do so, but in saying ‘parallels’ we should not assume that Eusebius meant that these passages were strictly identical in their wording, that they referred to the same historical event, or that the resolution of the difficulties they present should be immediately obvious to the reader. After all, some form of similarity is a necessary condition for an argument that two given passages contradict one another. Therefore, if we are following Eusebius’ own usage, when we speak of the ‘parallels’ that the Canon Tables provide us with, we should understand this to mean ‘passages that are similar’, without any further account of the way in which they are similar. Conversely, those passages listed in Canon X should be understood as not presenting any similarities to passages elsewhere in the fourfold gospel; in Eusebius’ own words, these are passages wherein ‘each of the evangelists has recorded certain matters on his own’ (ἕκαστος αὐτῶν περὶ τινων ἰδίως ἀνέγραψεν).⁴⁸ Hence it is the presence or absence of similarity that justifies placing a passage either in Canons I–IX or in Canon X. Some passages will present a similarity in terms of their wording, and on occasion even verbatim agreement. Other passages may be similar in terms of the themes or topics addressed. The former passages are the more obvious ones and required little interpretive judgement from Eusebius to classify them appropriately. However, with those passages that are more topical or thematic, we ⁴⁶ This lack of polemic is strikingly different from the explicitly anti heretical intent of the Canons on the Letters of the Apostle Paul, a cross referencing paratext created by Priscillian of Avila in the late fourth century, modelled on Eusebius’ Canon Tables. Cf. Lang and Crawford, ‘The Origins of Pauline Theology’. ⁴⁷ ep. Carp. (NA²⁸, pp. 89* 90*). The expression κατὰ τῶν αὐτῶν ἠνέχθησαν is used twice in the letter, as is the expression (λέγω) τὰ παραπλήσια. Cf. McArthur, ‘Eusebian Sections and Canons’, p. 254. ⁴⁸ ep. Carp. (NA²⁸, p. 89*).

112

The Eusebian Canon Tables

can, to some degree, behold his own understanding of how given gospel passages relate to one another by asking what in them he saw as similar.⁴⁹ In the final section of this chapter I want to illustrate this point with a few examples.

CASE S TUDIES IN EUSEBIUS’ PARALLELS

Case Study 1: Divine and Human Origins The most theologically significant parallel in all of Eusebius’ Canon Tables is also the first one that the reader encounters when reading the fourfold gospel from its beginning. Only two of the gospels provide genealogies for their main character (Matthew and Luke, with the latter perhaps inspired by the former), and these two famously diverge from each other, a contradiction that provoked an apologetic response as early as Julius Africanus in the third century, whose proposed solution Eusebius cited both in his Ecclesiastical History and in his Gospel Problems and Solutions.⁵⁰ Hence, when the Gospel of Matthew opens with a genealogy tracing Jesus’ lineage back to Abraham via Joseph and David (Mt 1:1–16 = Mt §1), one might expect that Eusebius would place this passage in Canon XMt, as unique Matthean text. Or, one might suppose that this genealogy would find a home in Canon V, recording parallels between Matthew and Luke, placed alongside the divergent Lukan lineage (Lk 3:23–38 = Lk §14). In fact, however, Eusebius placed the opening verses of the first gospel in Canon III, which lists material held in common by Matthew, Luke, and John. Alongside these two earthly-focused genealogies, divergent already amongst themselves, Eusebius presented three passages from the Johannine prologue, despite the radically different tenor and content of the fourth gospel’s opening. Excluding John 1:6–8 in view of its focus on John the Baptist (listed elsewhere in Canon III) and John 1:12–13 in view of its unique Johannine content (listed in Canon XJn), Eusebius presented the remainder of John 1:1–14 as text parallel to the Matthean and Lukan genealogies (i.e., John 1:1–5, 9–10, 14 = Jn §1, 3, 5; see fig. 6 in the introduction). As argued above, by juxtaposing these passages Eusebius has performed a service for the critic of the gospels, making the contradiction between the genealogies of Matthew and Luke obvious, not to mention the wholly different ⁴⁹ Cf. Watson, The Fourfold Gospel, p. 122: ‘Eusebius’s system does not depend on hypoth eses, although it does require interpretative decisions about what counts as a parallel and what does not’. Similarly, O’Loughlin, ‘The Eusebian Apparatus in Some Vulgate Gospel Books’, p. 7 commented: ‘in those cases where a choice has to be made as to which parallel is placed in the margin, we have what amounts to theological judgement which constitutes a fragment of exegesis in itself ’. ⁵⁰ Eusebius, HE 1.7.1 17; qu. Steph. 4.

Reading the Gospels with the Eusebian Tables

113

perspective of the fourth gospel. However, a more harmonious interpretation of these passages is also possible. No doubt Eusebius was intending to draw the reader’s attention to the theological truth that the Jesus Christ narrated by the fourfold gospel has both human and divine origins, more specifically that he is ‘the son of David’, ‘the son of Adam’, and the ‘Word in the beginning with God’, and that his ‘becoming flesh’ came to pass through a miraculous birth from a certain Mary who was betrothed to a certain Joseph. This association may be traditional and obvious to most readers but it is a different perspective from what one might have arrived at if, for example, the Word’s ‘becoming flesh’ were coordinated with the descent of the Spirit upon Jesus in the opening of Mark’s gospel, suggesting something more akin to an adoptionist Christology. In contrast, Eusebius’ association implies an incarnational Christology that traces the human and divine aspects of the gospel’s main character back to the very beginning of Jesus’ life and, indeed, back to a point before time began. Hence, although Carl Nordenfalk disparagingly referred to this as ‘not one of Eusebius’ most convincing parallels’,⁵¹ it succeeds in highlighting the canonical perspective that Jesus, ‘while yet being God the Word, did not deny that he himself was human’ (ἐπεὶ καὶ Θεὸς λόγος ὤν, οὐκ ἀπηρνεῖτο ἑαυτὸν εἶναι ἄνθρωπον).⁵²

Case Study 2: Jesus’ Passover Journeys to Jerusalem As will become apparent to any close reader of the gospels, the chronology of the fourth gospel noticeably differs from that of the other three, since it records multiple Passover celebrations during Jesus’ years of activity rather than just one, a difference that according to Epiphanius was one reason why the heretical sect known as the Alogi rejected the fourth gospel.⁵³ In the third century Origen similarly acknowledged this chronological divergence between ⁵¹ Nordenfalk, ‘Canon Tables on Papyrus’, p. 37. ⁵² Eusebius, qu. Steph. 1.9 (SC 523, p. 94). See also Eusebius, HE 3.24.13 (SC 31, p. 132) where he asserted that because Matthew and Luke had already recorded Jesus’ human genealogy, the evangelist John began with the ‘divinity’ (τῆς . . . θεολογίας) of the Word. For a theological reading of Matthew’s genealogy, see especially Watson, The Fourfold Gospel, pp. 27 42, and on the coordination of Synoptic and Johannine perspectives in insular manuscripts, see the brilliant study by Jennifer O’Reilly, ‘St. John the Evangelist: Between Two Worlds’, in Insular Anglo Saxon Art and Thought in the Early Medieval Period, ed. Colum Hourihane (Princeton, NJ: Index of Christian Art, Dept. of Art & Archaeology, Princeton University, 2011), pp. 189 218, who observed, ‘The Eusebian canons here give unusual expression to the theo logical rather than literal concordance existing between texts, but the spiritual harmony of these passages was well established in patristic exegesis’ (p. 189). ⁵³ Epiphanius, pan. 51.22.1. For a not entirely convincing argument that Epiphanius’ Alogi were in fact a cipher for Origen himself, see T. Scott Manor, Epiphanius’ Alogi and the Johannine Controversy: A Reassessment of Early Ecclesial Opposition to the Johannine Corpus, VCSup 135 (Leiden: Brill, 2016), pp. 177 99.

114

The Eusebian Canon Tables

the mention of the Passover early in the narrative of the fourth gospel and its very late mention just before the Passion in the three Synoptics.⁵⁴ As one would expect, Eusebius duly recorded all three of the singular Synoptic passages as parallels (Mt 26:2; Mk 14:1a; Lk 22:1 = Mt §274; Mk §156; Lk §260; see fig. 21), but rather than including them in the Synoptic-only Canon II, he placed them in Canon I, thrice repeating this parallel, each time juxtaposed with one of the Passover chronological statements from the fourth gospel, namely John 2:13, 6:4, and 11:55a (= Jn §20, 48, 96).⁵⁵ This juxtaposition illustrates that Eusebius was not interested in recording narrowly historical parallels amongst the gospels. If he had been trying to represent strict historical congruence, he surely would have juxtaposed the Synoptic passages only with the final Johannine Passover mention. The fact that he did not do so implies that he was more interested in creating thematic cross-references, which increases the hermeneutical potential of his system.⁵⁶ Moreover, although these parallels alone by no means solve the difficulties that arise when one tries to coordinate the chronology of the gospel of John with the other three, they do allow the reader to recognize quickly that the fourth gospel is doing something different from the first three and provide the textual material necessary for articulating some kind of account of gospel harmony out of apparent discord.

Case Study 3: The Cleansing(s) of the Temple Immediately following the first of John’s Passover statements is his record of Jesus’ cleansing of the temple in Jerusalem (Jn 2:14–16 = Jn §21). This, of course, is one of the classic instances in which the chronology of the fourth gospel diverges from that of the Synoptics, which all report the temple cleansing in the final week of Jesus’ life (Mt 21:12–13; Mk 11:15b–17; Lk 19:45–46 = Mt §211; Mk §121; Lk §238; see fig. 22). Once again, this divergence had already been recognized since at least the days of Origen, who concluded that the differences were so great that the two chronologies could not be harmonized on the literal level but must be understood in a spiritual sense.⁵⁷ Had he been concerned for strict historical accuracy, Eusebius would probably not have juxtaposed the Johannine cleansing with that of the Synoptics, supposing instead that they are two separate ⁵⁴ Origen, In Jo. 10.112 15. ⁵⁵ Cf. Toda, ‘The Eusebian Canons’, p. 41. ⁵⁶ Cf. McArthur, ‘Eusebian Sections and Canons’, p. 254: ‘There is, therefore, convincing evidence that Eusebius did not limit his system to the identifying of alternative versions of the same incident in the various Gospels. He linked together passages which could not conceivably be identical but which expressed some common concept or activity . . . His system represented a primitive form of Marginal References’. ⁵⁷ Origen, Jo. 10.119 209.

Fig. 21. Mk §156 (Mk §ρνς = Mk 14:1a) in Codex Basilensis A.N.III.12 (ninth c.). The penultimate marginal number written in black is ρνς and the corresponding section begins midway through the third line from the bottom: ἦν δὲ τὸ πάσχα καὶ . . . and ends on the next line with ἡμέρας. Note that in this instance the scribe has made a mistake in the table at the bottom of the page, failing to list the passages from the other three gospels that are parallel to this one in Mark. Basel, Universitätsbibliothek, A.N.III.12, fol. 143r

116

The Eusebian Canon Tables

Fig. 22. Mt §211 (Mt §CCXI = Mt 21:12 13) in the Lindisfarne Gospels (eighth c.). The text reads Et intravit Ih(su)s in templum d(e)i . . . London, British Library, Cotton MS Nero D IV, fol.68v (© The British Library Board)

events in Jesus’ ministry. In fact, however, his system simply does not take into account however he might have wished to resolve this conflict. As with the previous example, rather than determining an outcome to the dilemma, his intertextual references simply present the passages side by side and allow the reader to consider how best to reconcile them.

Case Study 4: Jesus’ Body as Bread The fourth gospel famously omits any account of the institution of the eucharist on the night before Jesus’ crucifixion, an absence no doubt related to the presence of a distinctly Johannine bread-of-life discourse delivered in the Capernaum synagogue following the feeding of the 5,000. Eusebius’ Canons again take note of this divergence between John and the Synoptics while also carefully highlighting what in the Johannine account is similar to the Synoptic Last Supper stories. Much of the sermon in John 6 is placed in Canon XJn, as passages unique to the fourth gospel (i.e., Jn 6:31–4, 35b–37, 39–40, 43–5, 47, 49–50, 52–4, 56–61, 63b–64a, 65–7, 70–1 = Jn §54, 56, 58, 60, 62, 64, 66, 68, 71, 73, 75). However, four passages from John 6 are singled out and

Reading the Gospels with the Eusebian Tables

117

placed in Canon I alongside three passages from the other evangelists, namely Matthew 26:26, Mark 14:22, and Luke 22:19 (= Mt §284; Mk §165; Lk §266).⁵⁸ These verses from the Synoptics are those that record Jesus’ blessing of the bread at the Last Supper and his statement identifying the bread with his body given for the disciples, and do not include the blessing of the cup and the prediction of a future banquet in the kingdom of the Father. In other words, Eusebius has singled out only the ‘bread’ statements from the Synoptics in order to juxtapose them with four selected Johannine passages from the breadof-life discourse (Jn 6:35a, 48, 51, 55 = Jn §55, 63, 65, 67; see fig. 28 in chapter 5). The first two of these Johannine texts are simple declarations that Jesus is the ‘bread of life’, and the latter two identify this bread as Jesus’ ‘flesh’, so the similarity in view here is an identification of Jesus’ person with bread. This parallel is a particularly clear instance of Eusebius striving to find places where the evangelists speak about ‘similar things’, without ignoring the differences between them. His careful treatment of John 6 highlights only this one point of contact between the passage and the Synoptic accounts of the Last Supper, and once again does not tell the reader how the passages should be related to one another. Nevertheless, one theologically suggestive way of reading them is to view the distinctly Johannine idea of gaining eternal life by eating the bread from heaven as a fuller elaboration of the terse Synoptic statements at the Last Supper. Taken by themselves, the Synoptic accounts do not require that one view the eucharist as an impartation of eternal life, though neither do they close off this interpretive option. However, when overlaid with the Johannine bread-of-life discourse, some such theology of the eucharist is inescapable. This was a common enough connection in the interpretation of these passages among early Christian exegetes like Eusebius, so we should not assume that his Canon Tables are the origin of the eucharistic reading of John 6. Still, his juxtaposition of these passages ensured that this reading would be transmitted to later readers who made use of his edition of the fourfold gospel.

Case Study 5: Peter’s Confession Another of the common points of narrative sequence among the Synoptics is Peter’s confession of Jesus as the Christ, followed by the first prediction of the Messiah’s coming suffering, then a discourse on the conditions of discipleship, and finally the transfiguration. The fourth gospel lacks such a sequence but does have a comparable confession that concludes the breadof-life discourse just discussed. In the Synoptics, Peter, following upon Jesus’ questioning of his disciples, declares that Jesus is ‘the Christ, the ⁵⁸ Cf. Toda, ‘The Eusebian Canons’, p. 41.

118

The Eusebian Canon Tables

Son of the living God’, ‘the Christ of God’, or simply ‘the Christ’ (Mt 16:13–16, Mk 8:27–29a, Lk 9:18–20 = Mt §166; Mk §82; Lk §94; see fig. 23),

Fig. 23. Lk §94 (Lk §ϘΔ = Lk 9:18 20) in The Gospel Book of Theophanes (second quarter of the twelfth c.). The corresponding section begins on the fifth line with the words καὶ ἐγένετο ἐν τῷ εἶναι αὐτὸν. Peter’s confession occurs further down on lines 12 13: ἀποκριθεὶς δὲ πέτρος εἶπε τὸν χ(ριστὸ)ν τοῦ θ(εο)ῦ. National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710 5), fol. 153r

while in John, in response to a similarly pointed dominical query, he states that Jesus ‘has the words of eternal life’ since he is the ‘holy one of God’ (Jn 6:68–9 = Jn §74). Eusebius has collected these four passages together, and gone one step further. In addition he has also included the uniquely Johannine scene in which Andrew introduces Peter to Jesus as the ‘Messiah’ and Jesus subsequently renames Peter ‘Cephas’ (Jn 1:41–2 = Jn §17). Here again, it cannot be the case that Eusebius has merely historical congruence in view, because the Synoptic confession accounts take place at a different location from the two Johannine passages. This is therefore yet another topical or thematic parallel, this one grouped around the idea of the confession of Jesus’ identity. Moreover, this parallel once again draws attention to an apparent contradiction amongst the gospels, since in Matthew Peter’s confession is followed by his renaming by Jesus, occurring midway through the overall gospel narrative, while in John as soon as Andrew introduces Jesus to Peter as ‘the Christ’, Peter receives his new name, occurring at the outset of his

Reading the Gospels with the Eusebian Tables

119

following of Jesus.⁵⁹ Alternatively, however, a canonical perspective is able to view these confessions as a diverse but harmonious chorus of statements from across the fourfold gospel proclaiming Jesus’ identity as God’s chosen and anointed one who has the words of life.

Case Study 6: The Petrine Commission The final case study to be considered highlights even more clearly than the previous examples the intricate treatment that Eusebius applied to parallels across the fourfold gospel. The Lukan account of the Last Supper uniquely contains a dominical prediction of Peter’s restoration and commission to ‘strengthen the brothers’ following his denial of Jesus (Lk 22:32b = Lk §274). The user of Eusebius’ gospel Canons is led to recognize that the fourth gospel likewise contains a Petrine commission, though in an entirely different setting, occurring not before but after the resurrection, around a charcoal fire on the shores of a lake following breakfast, far removed from the tense and fearful night on the eve of the crucifixion. In the Johannine post-resurrection episode, Jesus famously queries the leader of the apostles three times regarding his love for the master, and, upon receiving Peter’s increasingly emphatic affirmative replies, each time follows with a command to feed the Lord’s sheep. Eusebius appropriately relegated each of Jesus’ three queries, along with Peter’s three replies, to Canon XJn as uniquely Johannine content (Jn 21:15a–b, 16a–b, 17a–b = Jn §226, 228, 230). However, the thrice repeated command to tend the dominical flock (John 21:15c, 16c, 17c = Jn §227, 229, 231; see fig. 24) he juxtaposed with the Lukan commission to ‘strengthen the brothers’, reinforcing the distinct task given to Peter among the followers of Jesus after the resurrection. The fact that the Petrine commission occurs in a single verse in the gospel of Luke, with no further elaboration, and that Eusebius has so precisely paired it with the similar passages from the fourth gospel illustrates well the complexity of his intertextual system of cross-references and the diligent study of gospel parallels that must have been required to create them.

CO NCLUSION The great challenge presented by the fourfold gospel is how to do justice simultaneously to the similarities and to the differences amongst its component parts. Perpetual temptations are to cut the Gordian knot by giving ⁵⁹ Origen also recognized the differences between the gospels related to the timing of Peter’s call and his renaming (Jo. 10.31 3).

Fig. 24. The Johannine version of Peter’s commission in The Lindisfarne Gospels (eighth c.). This column contains Jn §227 232. The oscillation between Canon IX and Canon X shows how meticulous Eusebius was in his treatment of this passage. e.g., Jn §229 contains only the five words dicit ei pasce agnos meos. With Jn §227 the scribe has made a slight mistake, beginning the section two lines too early. London, British Library, Cotton MS Nero D IV, fol. 258v (© The British Library Board)

Reading the Gospels with the Eusebian Tables

121

privileged place to a single gospel, or by abstracting from the four narratives themselves to create a reconstructed sequence that is identical with none of the four. In light of the challenge this problem presents, the genius of Eusebius’ system is that it provides the fourfold gospel with a tool that enables the reader to attend simultaneously to what is unique and what is common, without disrupting the integrity of any of the four, or prescribing solutions to the many problems that arise when attempting to read the gospels as a coherent canon of Christian scripture.⁶⁰ Eusebius, the ‘impresario of the codex’,⁶¹ ingeniously accomplished this goal by creating a paratext that forged intertextual links and encouraged hypertextual reading within the four-gospel codex, all while leaving up to the reader how best to account for the textual information provided by his apparatus. This was the remarkable achievement that he bequeathed to the various Christian traditions of late antiquity that took up and made use of his gospel Canons. ⁶⁰ For a theological reading of select passages from the fourfold gospel that employs the Canon Tables as a guide to canonical coherence, see Watson, The Fourfold Gospel, pp. 125 65. ⁶¹ The title of chapter 4 of Grafton and Williams, Christianity and the Transformation.

4 ‘The Diversity of Agreement among the Four Evangelists’ Augustine’s Usage of the Canon Tables

One of the puzzling facts in the reception history of the Eusebian apparatus is that the creator of this impressive scholarly tool does not seem to have put it to much use. Of course it is possible that Eusebius might have done so in some work that is now lost to us, and I do not want to claim that there are no passages in his surviving corpus that might reveal dependency upon his Canon Tables, but based on the evidence we do have, it seems that he never thoroughly explored the exegetical implications of the intertextual links he himself created. Eusebius did, of course, directly confront the issue of gospel contradictions in his dual treatise Questions to Stephanus and Marinus, probably the same work that Jerome referred to as De euangeliorum διαϕωνίᾳ.¹ Yet in that work he dealt only with the problems arising from the beginnings of the gospels and their endings, specifically the nativity and resurrection accounts, and the fragments of the treatise that have survived do not clearly demonstrate the usage of the apparatus. To use the analogy proposed in chapter one, it would be as if Ptolemy had simply composed the Handy Tables but never bothered to use them for performing the calculations necessary for constructing a grand astronomical theory accounting for the phenomena his tables represented. Therefore, the task of exploiting the full potential of the Eusebian paratext fell to subsequent generations, and the first figure to take up the challenge was Augustine of Hippo, approximately sixty years after Eusebius’ death. His treatise De consensu evangelistarum represents, in the words of David Laird Dungan, ‘the most sophisticated and comprehensive explanation of the writing of the Gospels in all of early Christianity’, an assessment that is surely on target.² Although ¹ Jerome, vir. 81 (Richardson, p. 43). ² Dungan, A History of the Synoptic Problem, p. 112. However, I must depart from Dungan’s claim to detect in cons. ev. ‘a harsh peremptoriness, a threatening tone, which implies that [Augustine] would not hesitate to bring in the sword of the state if need arose. As a result, The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0005

126

The Eusebian Canon Tables

Augustine’s focus on the problem of contradictions within and amongst the gospels might have initially been prompted by some pagan or Manichaean detractor,³ in this work the bishop of Hippo went well beyond any prior critic or defender of the gospels by analysing all of the possible contradictions throughout the fourfold gospel corpus.⁴ In this chapter I will argue that in the research that led to the writing of his groundbreaking and highly influential treatise, Augustine was directly reliant upon Eusebius’ Canon Tables, indeed, that he would not have been able to accomplish this task in the absence of Eusebius’ marginal apparatus, or some other similar cross-referencing technology. Furthermore, in the last section of this chapter I will argue that based on his exploitation of the Eusebian paratext, Augustine was able to offer a general theory of gospel composition to account for the empirical data of the canonical fourfold text, one that was more sophisticated than anything that had previously been attempted. In other words, it was Augustine who first produced something like the Almagest to complement Eusebius’ version of Ptolemy’s Handy Tables. This hypothesis has been put forward previously but remains without adequate defence. Over a century ago Theodor Zahn, following the work on the Vulgate by F. C. Burkitt,⁵ noted that ‘Augustine had before him Jerome’s Augustine’s writing seems curiously smug, as if he secretly believed that all he needed to do was put forth a convincing show or rational explanations regarding the Gospels, and the imperial police would imprison any who dared dispute with him’ (p. 118). For a recent survey of scholarship on cons. ev., see Brian Dunkle, ‘Humility, Prophecy, and Augustine’s Harmony of the Gospels’, Augustinian Studies 44 (2013): pp. 207 9. ³ In his ret. 2.16 Augustine stated that he wrote the work against those who claimed the evangelists did not agree, though he does not name these opponents. For a brief analysis of the polemical target of the treatise, see Dunkle, ‘Humility, Prophecy, and Augustine’s Harmony’, pp. 210 13, who followed the earlier work of Goulven Madec, ‘Le Christ des païens d’après le De consensu evangelistarum de saint Augustin’, Recherches augustiniennes 26 (1992): pp. 3 67, in seeing the entire treatise as occasioned by a small group of pagans in North Africa. In contrast, Carol Harrison, ‘ “Not Words But Things”: Harmonious Diversity in the Four Gospels’, in Augustine: Biblical Exegete, ed. Frederick Van Fleteren and Joseph C. Schnaubelt (New York: Peter Lang, 2001), p. 157, saw book one as directed against pagan critics like Porphyry and books two through four against the Manichaeans. Merkel, Die Widersprüche, pp. 224 7, also argued that the Manichaeans were Augustine’s opponents. The idea that Porphyry was the unnamed source of the criticisms to which Augustine responded in books two to four goes back at least to Robert Louis Wilken, The Christians as the Romans Saw Them (New Haven, CT: Yale University Press, 2003; reprint, 2nd edn), pp. 144 7 [originally published in 1984], and found its most enthusiastic proponent in Robert M. Berchman, who included extensive passages from cons. ev. in his collection of Porphyrian material (Porphyry Against the Christians, pp. 173 84), even though they have no obvious connection to Porphyry. However, in the most recent assessment of this question, Magny, Porphyry in Fragments, pp. 119 47, argued that none of these criticisms came from Porphyry, and Becker’s new edition of Porphyry’s Contra Christianos contains no material from cons. ev. ⁴ For a survey of the types of argument used by Augustine to resolve apparent contradictions, see Merkel, Die Widersprüche, pp. 227 50; Dungan, A History of the Synoptic Problem, pp. 116 41; Harrison, ‘Not Words But Things’, pp. 159 62; Watson, Gospel Writing, pp. 28 43. ⁵ F. C. Burkitt, The Old Latin and the Itala, Texts and Studies 4 (Cambridge: University Press, 1896).

‘Diversity of Agreement among the Evangelists’

127

revision of the text of the Gospels, which was furnished with the Eusebian canons and sections’. Hence, he inferred that Augustine ‘plainly used’ the apparatus ‘in his harmonistic work’. Zahn, however, gave no specific examples to substantiate his claim, simply noting that Augustine relied on Canons I–VII in books two and three of the treatise, and on Canon X in book four. He also drew attention to a key passage in book one of the work, which might be read as implying that Augustine had open before him a codex with the Canons which he was scanning to gain an overview of the possible relationships among the gospels.⁶ I will return to this passage in due course and largely agree with Zahn’s reading of it. However, as we shall see, Zahn evidently misunderstood Augustine’s purpose in book four, and thereby overlooked the clearest evidence for his usage of Eusebius’ apparatus.⁷ Most importantly, the fact that Zahn’s brief remarks lacked sufficient evidence to substantiate his claim left the hypothesis wide open to critical scrutiny. Scrutiny came only a year later, when Heinrich Vogels argued that in fact Augustine did not use the Vulgate in De consensu and suggested in passing that there was not ‘den geringsten Anhaltspunkt’ for Zahn’s suggestion.⁸ Given that Vogels was even more laconic with his assertion than Zahn had been, the door was left open for the Italian scholar Angelo Penna, half a century later, to take up the challenge of considering ‘in maniera analitica’ whether Augustine had relied upon the Canons.⁹ Penna proceeded to list a series of parallels Eusebius had included in his tables, which Augustine had failed to make note of, and concluded that there was no evidence of dependence, and that the harmonistic purpose of Augustine’s treatise was fundamentally different from the ‘synopsis’ of Eusebius.¹⁰ However, the problem with Penna’s approach was that it was not analytic enough, since he merely surveyed a few negative examples rather than undertaking a thorough assessment of the evidence at hand. Moreover, his methodology was flawed in that he assumed too high a degree of correspondence would be necessary to ⁶ Theodor Zahn, Introduction to the New Testament, Volume II, trans. Melancthon Williams Jacobus et al., 3rd edn (Edinburgh: T. & T. Clark, 1909), p. 422. For the German original, see Theodor Zahn, Einleitung in das Neue Testament, 3rd edn (Leipzig: Deichert, 1907), II.200. ⁷ Dungan, A History of the Synoptic Problem, p. 116, like Zahn, incorrectly assumed that book four ‘considers the material unique to each Gospel’. ⁸ Heinrich Joseph Vogels, St. Augustins Schrift De consensu evangelistarum unter vornehm licher Berücksichtigung ihrer harmonistischen Anschauungen: Eine biblisch patristische Studie, Biblische Studien 13 (Freiburg im Breisgau: Herdersche Verlagshandlung, 1908), p. 52. Vogels acknowledged that the citations in cons. ev. mostly conform to the Vulgate, but posited that this was because they had been altered in the later manuscript tradition. For Burkitt’s response to this theory, see F. C. Burkitt, ‘Saint Augustine’s Bible and the Itala’, Journal of Theological Studies 11 (1910): pp. 258 68, 447 58. ⁹ Angelo Penna, ‘Il «De Consensu Evangelistarum» ed i «Canoni Eusebiani»’, Biblica 36 (1955): p. 6. ¹⁰ Penna, ‘De Consensu Evangelistarum’, pp. 7 19. Penna’s conclusion was followed by Merkel, Die Widersprüche, pp. 229 30 n. 49.

128

The Eusebian Canon Tables

demonstrate dependence. As Francis Watson has argued, ‘Penna’s method does not allow Augustine to be selective in his presentations of parallels’.¹¹ Watson, like Zahn before him, pointed to a central passage from book one which appears to provide a brief summary of information gleaned from Eusebius’ Canons, and concluded from it that Augustine probably used the cross-referencing system.¹² Recently Hugh Houghton, in his study of the text of John in Augustine’s corpus, noted that the Canon Tables ‘would have facilitated Augustine’s composition of De consensu evangelistarum’, though he did not explore this possibility further given that it was outside the scope of his project.¹³ Similarly, in his History of the Synoptic Problem, Dungan assumed Augustine’s usage of Eusebius’ Canons, though without giving any evidence to support this claim.¹⁴ Hence, scholarship on this question over the past century has been divided, though the dominant view, especially in recent decades, has been that Augustine depended upon Eusebius. Nevertheless, a robust argument for this hypothesis has yet to be put forward, owing to the absence of a thorough investigation of the entire treatise with this question in view. The goal of this chapter is to settle the matter by undertaking the ‘analytic’ examination promised 60 years ago by Penna.

AUGUSTINE’ S ACCESS TO JEROME’ S VUL GATE The first question to settle is whether or not Augustine was using the Vulgate in De consensu. If he was relying solely on the Old Latin version, we would have no reason to suppose he would have had access to the Canon Tables, since they were only introduced into the Latin tradition with Jerome’s revision.¹⁵ Fortunately this ¹¹ Watson, Gospel Writing, p. 18 n. 15. ¹² Watson, Gospel Writing, pp. 17 19. The same passage had also been noted by David B. Peabody, who commented that it seems ‘to reflect the data found in the Canons of Eusebius’ (David B. Peabody, ‘Augustine and the Augustinian Hypothesis: A Reexamination of Augustine’s Thought in De consensu evangelistarum’, in New Synoptic Studies: The Cambridge Gospel Confer ence and Beyond, ed. William R. Farmer (Macon, GA: Mercer University Press, 1983), p. 41 n. 7). ¹³ H. A. G. Houghton, Augustine’s Text of John: Patristic Citations and Latin Gospel Manu scripts, OECS (Oxford: Oxford University Press, 2008), p. 14. Thomas O’Loughlin has recently commented, ‘Until a complete comparison of the Eusebian Apparatus and the De consensu is carried out, the question of whether or not Augustine used the Apparatus must be considered unanswered. However, if Augustine did not have access to it, then he must have repeated much of the work of sectioning the gospels already done by Eusebius’ (O’Loughlin, ‘Harmonizing the Truth’, p. 16). ¹⁴ Dungan, A History of the Synoptic Problem, p. 113. ¹⁵ Of course, even if he was using the Vulgate, it is theoretically possible that he had a copy that did not include the Canon Tables, since scribes might copy only the text and omit the complicated paratext. However, the fact that Jerome, in his Novum Opus, presented the tables as a central component of his new edition makes this unlikely, as does the short interval of time between the release of the Vulgate in Rome and Augustine’s use of it.

‘Diversity of Agreement among the Evangelists’

129

is a question that has received a thorough treatment in recent years. Hugh Houghton’s comprehensive investigation of Augustine’s text of the Gospel of John has demonstrated that from around 403 he began using the Vulgate, though at this point he did not exclusively rely on it, but instead still preached from Old Latin manuscripts for the next decade at least. In his analysis of De consensu, Houghton highlighted seventy-nine citations from the fourth gospel that are identifiable as either Old Latin or Vulgate.¹⁶ Of these, seventy-two are clearly Vulgate, with only seven not agreeing with any Vulgate tradition known to us.¹⁷ In other words, among those passages in which the Vulgate differs from the Old Latin, Augustine’s text represents the Vulgate ninety-one per cent of the time. Moreover, Houghton discerned a pattern in Augustine’s usage. His ‘citations are closest to the Vulgate when he follows the sequence of the Gospel’, in contrast to non-sequential passages cited, which tend to be the Old Latin. In other words, the codex he had before him, which he worked through sequentially in the course of the treatise, possessed a Vulgate text, but his ‘mental text’, which he often relied on for citing cross-references from memory, was still Old Latin.¹⁸ Although Houghton’s treatment only focused on the text of the fourth gospel, it is plausible to assume that if his text of John was Vulgate, the same was true for the first three gospels as well. Houghton’s examination therefore confirms the conclusion of Burkitt and Zahn, that the codex Augustine used to carry out his research for De consensu had a Vulgate text. A final indication that Augustine was dependent on the Vulgate when writing De consensu is that in book one he clearly refers to the order of the gospels as Matthew–Mark–Luke–John, and justifies this sequence on the basis of a presumed chronology of writing.¹⁹ Because most surviving Old Latin manuscripts have the order Matthew–John–Luke–Mark, the order given by Augustine probably represents the influence of Jerome’s new version, and the fact that Augustine provides justifications for both orders in the opening paragraphs of the treatise (De consensu 1.1.1–2.3) suggests that the new Vulgate sequence was novel both for him and for his presumed audience.²⁰ Naturally the dating of De consensu is crucial to this discussion. The fact that Augustine’s ‘mental text’ was still decidedly Old Latin implies that the work came from a time not long after he began using the Vulgate. The treatise has been dated to various points between 399 and 415, but the present consensus is around 403 or 404,²¹ notably the same time at which Houghton ¹⁶ He noted a total of 306 distinctly Vulgate readings in the Gospel of John, but only 79 of these passages are cited in cons. ev. ¹⁷ See the helpful chart at Houghton, Augustine’s Text of John, p. 158. ¹⁸ Houghton, Augustine’s Text of John, p. 160. ¹⁹ cons. ev. 1.2.3. ²⁰ Cf. Houghton, Augustine’s Text of John, pp. 14 15. ²¹ Pierre Marie Hombert, Nouvelles recherches de chronologie augustinienne, Collection des études augustiniennes, Série Antiquité 163 (Paris: Institut d’Études Augustiniennes, 2000), pp. 81 7, also followed by Houghton, Augustine’s Text of John, p. 158 n. 2. The argument as I am presenting it here may appear circular, in so far as one might conclude that the reason for dating cons. ev. to 403 is because this is the period when Augustine began citing the Vulgate, and

130

The Eusebian Canon Tables

suggests Vulgate citations begin appearing in Augustine’s broader corpus. Further evidence comes from Augustine’s Epistula 71, also written around 403 and addressed to Jerome. In this well-known letter he acknowledged his gratitude for Jerome’s translation of the Septuagint into Latin, but expressed some annoyance with the newer edition translated from the Hebrew, which lacked the critical signs of the former edition and had caused a disturbance in a nearby church. Towards the end of the letter, Augustine also noted how impressed he was by Jerome’s translation of the gospels, after having checked a number of passages in this new version against the Greek text. Even if the newness of this translation should trouble some, Augustine suggested that ‘so useful a work’ (labori tam utili) deserved praise.²² Thus, the internal evidence in De consensu for the usage of the Vulgate text around the year 403 coincides nicely with Augustine’s claim in Epistula 71 to have been using Jerome’s Vulgate edition of the gospels.²³

METHODOLOGICAL CLARIFICATIONS Before we turn to examine De consensu it will be helpful to lay out a few methodological clarifications. First, Augustine nowhere mentions Eusebius or his Canon Tables anywhere in the treatise. However, neither does he mention Jerome, even though he was clearly using his new translation. Thus, nonuse cannot necessarily be inferred from silence. On a related note, if Augustine was relying on Eusebius’ apparatus, he would likely have read Jerome’s prefatory letter to Pope Damasus, commonly known as the Novum Opus, in which he explained to the reader how to use the system, in a manner similar to certainty about when he began citing the Vulgate depends on the dating of cons. ev. to 403. Hombert, however, makes a convincing case on other grounds, most importantly the close connections between cons. ev. and serm. 51, which was preached in the final days of 403. He also points out a number of parallels with other works dated to 403 and 404. ²² Augustine, ep. 71.4 (CSEL 34.2, p. 254). ²³ It is even possible to speculate about the means by which Augustine may have acquired a copy of the Vulgate gospels, and here I follow closely the observations made by Penna, ‘De Consensu Evangelistarum’, pp. 1 3. According to Epistula 24 in Augustine’s corpus (a letter written some time before 394), Paulinus of Nola had contacted a certain Domnio in Rome to procure a copy of Eusebius’ De cunctis temporibus historiam to send to Carthage, so that a copy could be made and sent on to Alypius at Thagaste (ep. 24.3 (CSEL 34.1, p. 75)). The work in question must have been the Latin edition of Eusebius’ Chronicle produced by Jerome. Jerome’s relationship with this Domnio crops up elsewhere. His translation of Ezra Nehemiah completed in the 390s is dedicated to none other than this same figure, as is his revision of Chronicles, and in his ep. 47.3 he directed a friend in Rome to Domnio from whom he could obtain copies of his works. Hence, in Domnio, a priest in Rome, we have someone to whom Jerome sent his writings from the east for the purpose of dissemination, and we have at least one instance of him sending manuscripts all the way to North Africa for copying. This Domnio is, therefore, a strong candidate for the person responsible for conveying a copy of the Vulgate gospels to Augustine.

‘Diversity of Agreement among the Evangelists’

131

Eusebius’ Letter to Carpianus. However, there are no verbal parallels between Jerome’s preface and Augustine’s treatise that are distinctive enough for us to infer dependence. Jerome speaks of passages that are eadem, vicina, sola, and propria, and asserts that multiple passages can have the same sensus.²⁴ In contrast, Augustine prefers the words sententia, verba, res, and simile to describe the relationship between passages from multiple gospels. He does on occasion use solus but there is no reason to suppose that his usage of this term is due to the influence of Jerome’s preface. Again, however, nonuse cannot be inferred from this lack of terminological overlap, since, as we shall see by the end of this chapter, the rhetorically trained Augustine already had his own conceptual and terminological resources for describing the literary relationships amongst the gospels. Secondly, and more importantly, we need to be clear about what it means to say that Augustine ‘used’ the Canon Tables, especially in view of the argument of the last chapter that the marginal apparatus was open to a variety of uses. In keeping with this principle, the mere presence of the Canon Tables in a codex would not constrain the way in which Augustine chose to interpret any given passage or set of passages. Again, as argued already, the only hermeneutical guide Eusebius’ system provides is the indication that certain passages stand in some kind of relation to other certain passages, that they ‘say similar things’. Thus, the only sort of evidence that can elucidate whether Augustine relied on Eusebius’ guidance is the degree to which he followed the lead of the Caesarean historian when he identified passages standing in relation to one another that required some sort of explanation. How Augustine decided to interpret a set of parallel passages once he settled upon them has no bearing upon whether or not he was reliant upon Eusebius. This clarification is necessary in the light of Penna’s argument that Augustine would not have used the Canon Tables because Eusebius’ apparatus and De consensu had fundamentally different purposes, with the former aiming to present a synopsis of similar passages irrespective of context or historical referent, while the latter attempted to harmonize at the historical level all the contradictions in these accounts.²⁵ This assessment of the difference in the two works is accurate, but does not preclude the possibility of dependence, since the sort of synopsis-like information provided by Eusebius’ paratext is the necessary first step towards a harmonizing project like that of Augustine. Two further useful implications may be drawn from this awareness of the purpose of Augustine’s treatise. First, despite Zahn’s argument that Augustine relied on Canon X in book four, a hypothesis Penna sought to refute, in fact Canon X is the table we should expect to see least used by Augustine. As the bishop of Hippo emphasizes repeatedly throughout the treatise, ‘it cannot be ²⁴ Jerome, nov. op. (Biblia sacra iuxta vulgatam versionem, pp. 1515 16). ²⁵ Penna, ‘De Consensu Evangelistarum’, pp. 6 7, 18.

132

The Eusebian Canon Tables

seen as a disagreement if the one includes what the other omits, and the other records what the one does not’.²⁶ Those passages that are unique to one gospel should, generally speaking, pose no problems of contradiction, since some kind of similarity between two passages is a prerequisite for the existence of contradiction. The only question that might arise with respect to such passages is their relative chronology, which Augustine does on occasion consider,²⁷ but in general Canon X, recording passages unique to each gospel, was the least useful for the goal of his project, and we should accordingly not be surprised to find it little used in De consensu. Secondly, even for those passages that do have parallels in one or more other gospels, we should expect Augustine to restrict himself only to those that pose a possible contradiction. In other words, Augustine was not attempting to explain the relationships amongst the gospels through an exhaustive and comprehensive analysis of their similarities and differences, but was instead operating in a primarily apologetic mode, intending to ward off possible critics of their veracity. Therefore, if Augustine did have the Canon Tables before him, his usage of them would probably not be uniform, but would instead concentrate on those that are most pertinent to the aims of his treatise. Hence, selective usage of the parallels provided by Eusebius is what we should expect to find in a comparison of the two works. Of course, even if the parallel passages examined by Augustine should on occasion line up with those included by Eusebius in his tables, one could always object that Augustine had arrived at his parallels independently, since he was, after all, working with the same limited data set (i.e., the fourfold gospel) as his predecessor. The case for Augustine’s dependence on the Canon Tables must therefore be cumulative and probabilistic, but this need not imply that a convincing argument is beyond reach. If a sufficient number of Eusebian parallels were examined by Augustine, we may conclude that he probably relied upon the tables. Other evidence might also be mounted. If, for example, Augustine agreed with Eusebius in one or more particularly striking or unexpected parallels, we would have further grounds for assuming dependence. Finally, if the sequence of parallels cited by Augustine lined up with a sequence of parallels in the tables, we would have the strongest evidence for Augustine’s usage of Eusebius’ work. There are, therefore, three possible types of evidence that can be examined: 1) the ratio of overlap in the parallels included in the two works; 2) coincidence in identifying unexpected parallels; 3) agreement in a string of parallels. As we shall see, De consensu offers evidence in all three categories. This three-pronged approach looking for signs of positive dependence provides a better way forward than the negative methodology employed by Penna. ²⁶ cons. ev. 2.5.16: nec ideo contrarium uideri potest, quod uel hic dicit, quae ille praetermittit uel ille commemorat quae iste non dicit (CSEL 43, p. 98; trans. Paffenroth, p. 177). ²⁷ See, e.g., cons. ev. 2.5.16, in which he attempts to align the unique episodes of the Matthean and Lukan nativity accounts into a single chronology.

‘Diversity of Agreement among the Evangelists’

133

AUGUSTINE’ S DE P EN D E N C E U P O N T H E C A N O N TABLES IN DE CONSENSU EVANGELISTARUM The first thing to note about the evidence available in De consensu is that, with two exceptions, all of the parallels discussed or alluded to by Augustine had already been noted in the ten Canons. This is to be expected, since Eusebius had done such a thorough job of finding similar passages among the gospels. Moreover, if Augustine was reliant upon Eusebius, these exceptions might simply illustrate that he was not solely dependent on Eusebius but also drew upon his own knowledge of the fourfold gospel, something that we certainly should expect to be true. One of these exceptions pertains to the dominical double love commandment,²⁸ a passage that Augustine knew well and was regularly citing by this point in his career,²⁹ while the other is a correction to one of the few errors made by Eusebius.³⁰ Setting aside these exceptions, a summary of the overlap of Augustine’s parallels with those in the Canon Tables can be found in the tables in Appendix 2. Here I have listed for each section of De consensu the parallel passages discussed by Augustine, and have correlated them with Eusebius’ Canons and enumeration within each gospel. I then tallied the total parallels considered from each Canon and compared it to the total number of parallels included in each Canon. The overall percentages show much greater overlap in some Canons than in others. For example, Augustine mentions 47 per cent of the parallels common to all four gospels (Canon I) and 51 per cent of the parallels among the Synoptics (Canon II), both considerably high figures. However, for the next three Canons there is much less overlap: 14 per cent for Canon III, 28 per cent for Canon IV, and 27 per cent for Canon V. The

²⁸ Cf. cons. ev. 2.73.141 2. Augustine is discussing Mt 22:34 40, in which the Pharisees ask Jesus about the greatest commandment. Eusebius put this passage in Canon VI, noting the parallels Mt §224 and Mk §131. In the context of discussing this passage Augustine points out that Luke too ‘tells something like this, but not in this order and in a totally different place’ (trans. Paffenroth, p. 245), referring to Luke 10:25 8, in which a lawyer, in response to a query from Jesus, named the double love commandment. The latter passage Eusebius had placed in Canon II (Mt §193; Mk §107; Lk §121). Although he observes the similarity that Mt §224 and Lk §121 both name the double love commandment, Augustine concludes that they must be referring to two separate historical events. ²⁹ Cf. conf. 12.18.27; 12.25.35; doc. Chr. 1.26.27. ³⁰ Cf. cons. ev. 3.17.54 where Augustine discusses Mt §342; Mk §222; Lk §323 (= Mt 27:48 9; Mk 15:36; Lk 23:36 7), passages that Eusebius had placed in Canon II. These are the passages about the offering of sour wine to Jesus on the cross just before his death. Augustine realized that John has a similar passage about sour wine (Jn §203 = Jn 19:28 30a) and brought it up in his discussion of the three Synoptic texts. Eusebius had placed Jn §203 in Canon IV alongside Mt §333 and Mk §211 (=Mt 27:34; Mk 15:23), which are passages about Jesus being offered mixed wine. The Johannine text should obviously be aligned with the other three Synoptic passages and placed in Canon I. This error was removed in the Peshitta revision of the Eusebian paratext, on which see my discussion in the next chapter.

134

The Eusebian Canon Tables

highest percentages of overlap occur with respect to Canon VI (70 per cent) and Canon VIII (77 per cent). As expected, Canon XLk and Canon XJn show up very little (17 per cent and 10 per cent respectively), while Canon XMt and Canon XMk are in fact noted relatively often (31 per cent and 32 per cent respectively). Bearing in mind the aforementioned point that we should not expect Augustine to note all of Eusebius’ parallels, since he is primarily interested in those that might pose contradictions, the percentages of overlap with some of the Canons are quite impressive. Either Augustine was using Eusebius, or he, on his own, replicated half of what Eusebius had accomplished with respect to Canons I and II, over two-thirds of Canon VI, and nearly all of Canon VIII. The method employed by Augustine in De consensu partially accounts for the uneven overlap with the various Canons. Book one of the treatise is mostly devoted to deflecting various pagan critics of Christianity who honour Jesus as merely a wise man and who point out that he left behind no writings. The main exegetical portion of the work does not begin until book two. Here Augustine takes up the Gospel of Matthew and uses it as a plumb line against which to study the other gospels. He progresses through the first gospel sequentially, discussing one passage after another by citing a passage’s incipit and explicit, offering an explanation of it, and then moving on to the next passage. This produces a very detailed and thorough treatment, with Augustine identifying and explaining possible contradictions that occur with the other three as he works his way through Matthew, focusing especially on comparing Matthew with the other Synoptics. Hence, Matthew–Mark–Luke and Matthew–Mark comparisons are most common. Moreover, along the way, as he works through Matthew’s text, he on occasion notes Matthew’s unique material even though it is not central to his purpose. Augustine continues in this manner from Matthew’s nativity story up to the beginning of the Passion narrative, covering Matthew 1:1–26:25 in comparison with the other three gospels, and thus concludes book two. In book three he announces a shift in focus. Because the four gospels are much more in unison in the Passion and resurrection accounts, so Augustine explains, he decides to consider all four gospels together in book three, rather than working through a single one using it as a basis of comparison for the others, as he had done in the previous book.³¹ Thus, in book three he investigates the Passion and resurrection accounts of all four gospels simultaneously, and, despite his stated intention to pursue the project otherwise,

³¹ cons. ev. 3.1.1 (CSEL 43, p. 268): iam quoniam omnium quattuor narratio in eo uersatur loco, in quo necesse est eos usque in finem pariter ambulare nec multum digredi ab inuicem, sicubi forte alius aliud commemorat quod alius praetermittit, uidetur mihi expeditius nos demonstrare posse omnium euangelistarum conuenientiam, si ab hoc iam loco omnium omnia contexamus et in unam narrationem faciem que digeramus.

‘Diversity of Agreement among the Evangelists’

135

Matthew still tends to dominate this book in terms of chronology. When he has finished book three by considering all four gospels up through the resurrection appearances, one might suppose that Augustine’s task was complete, but in fact, it is not. He still has not given any attention to the material common to the latter three gospels that is not shared with Matthew, since in book one he used Matthew as his guide. Hence, these passages become the focus of book four. In De consensu 4.1.1–4.7.8 Augustine works through the Gospel of Mark from the beginning noting parallel passages and possible contradictions against Luke. (He does not consider possible contradictions between Mark and John because he has already explained in De consensu 1.2.4 that they share no material that is not also included in the other gospels.) Next, in De consensu 4.8.9–4.9.10 he considers possible contradictions between the Gospels of Luke and John. Finally, when he comes to De consensu 4.10.11, he recognizes that has exhausted all possible comparative combinations, and so notes that ‘John is the last, and there remains nothing with which to compare him’. Having considered all the parallels that could give rise to contradictions between gospels, Augustine concludes the treatise by walking through the Gospel of John and highlighting some of its distinctive aspects. We should observe at this point that Augustine’s working method in De consensu correlates precisely with the sequence of the Canon Tables. By using Matthew as a guide in book two he was able to cover the parallels contained in Canons I–VII. Then, in book four, Augustine turned to Mark to consider parallels with Luke, which Eusebius had reported in Canon VIII, and finally parallels between Luke and John, noted in Eusebius’ Canon IX. One might argue that this sequence was merely suggested to him by the sequence of the fourfold gospel itself, which, after all, begins with Matthew–Mark, then Mark– Luke, and finally Luke–John. However, it is a further striking coincidence with Eusebius that after he has considered Mark–Luke parallels he does not attempt to find any Mark–John parallels, as the canonical sequence would imply, but instead moves on to Luke–John parallels. Thus both the order and omissions in Augustine’s analysis follow Eusebius’ Canons closely. Examining several examples of the overlaps between the Canon Tables and De consensu will add to the body of evidence. Some overlapping parallels are straightforward and sufficiently obvious that they might be mere coincidence. For example, Augustine recognizes that all four evangelists discuss John the Baptist and spends some time noting the differences among the accounts, such as the fact that Matthew and Mark alone include a description of John’s clothing (Mt §9; Mk §3), and that John’s extended speech is recorded only by Matthew and Luke (Mt §10; Lk §8).³² Since the story of John occurs at the beginning of each gospel, the parallel passages would have been easy enough to

³² cons. ev. 2.6.19; 2.12.25 6.

136

The Eusebian Canon Tables

find. Furthermore, these sorts of observations would have been possible if Augustine simply had the four accounts open, perhaps in four separate codices, making his own comparison of them. Another unsurprising parallel is Augustine’s noting that all four evangelists record the feeding of the 5,000 (Mt §147; Mk §64; Lk §93; Jn §49), and his recognition that Matthew, Mark and John include Jesus’ walking on water immediately after this (Mt §150; Mk §67; Jn §51), though only Matthew reports Peter’s walking on water (Mt §151).³³ These sorts of observations about large chunks of narrative, especially of memorable scenes like the feeding of the 5,000 and the walking on water, would not have been too difficult to make in the absence of a crossreferencing system. More important for my argument are parallels identified by Augustine that occur in different narrative sequences in each gospel. For example, when he comes to consider the call to discipleship and the excuses some made in Mt §68 (Mt 8:19–22), Augustine observes that ‘although this account is given similarly in Luke, it comes after many other things and without any real reference to the chronological order’.³⁴ He is referring to Lk §105 (Lk 9:57–60), which appears in an entirely different sequence in the third gospel, and was placed by Eusebius in Canon V. Augustine also notes that whereas Matthew included only two would-be disciples, Luke added a third (Lk §106 = 9:61–2), also recorded by Eusebius in Canon XLk. This is the sort of parallel that would have been difficult to find unless one either had a crossreferencing system or a very thorough familiarity with the sequence of each gospel. Again, Augustine, like Eusebius, notes that both Matthew and Luke record the exorcism of a demon causing muteness, even though Luke includes the story ‘not in this order but after many things’ (Mt §119; Lk §126 = Mt 12:22; Lk 11:14).³⁵ An even more striking example comes from Canon I. When he treats Jesus’ preaching in his hometown of Nazareth in Mt §141–2 (= Mt 13:54–8), Augustine not only recorded the Synoptic parallels (Mk §50–1; Lk §19, 21), which themselves occur in a different sequence from that of Matthew, but was also able to find the Johannine parallels that occur ‘in very different and dissimilar places’.³⁶ Augustine refers to things ‘either spoken to the Lord or by him’ in the fourth gospel parallel to this passage in Matthew, which can

³³ cons. ev. 2.46.95; 2.47.99 100. ³⁴ cons. ev. 2.23.54 (CSEL 43, p. 155; trans. Paffenroth, pp. 202 3): hoc similiter narrat et lucas. sed ille post plura nec ipse sane expresso ordine temporum. ³⁵ cons. ev. 2.37.84 (CSEL 43, p. 186; trans. Paffenroth, p. 217): hoc non isto ordine, sed post alia multa lucas commemorat. ³⁶ cons. ev. 2.42.89 (CSEL 43, p. 192; trans. Paffenroth, p. 220, modified): nam iohannes longe in diuersis et dissimilibus narrationis suae locis uel dicta esse domino talia uel eum dixisse commemorat, qualia hoc loco tres ceteri meminerunt. Paffenroth translates locis as a singular, and so misses the dual reference.

‘Diversity of Agreement among the Evangelists’

137

only be a reference to Jn §59 and §35, since in the former Jesus is spoken to, and in the latter Jesus himself speaks. These Johannine parallels show up in different places in John’s gospel and are brief statements, corresponding to one and two verses in modern reckoning (Jn 6:41–2; 4:44), so Augustine’s success in identifying them is notable. Similarly, Augustine recognizes that although John departs from the Synoptics in not reporting the words of institution of the eucharist on the night of the betrayal, he ‘clearly testifies that the Lord said this much more fully at another time’.³⁷ Augustine no doubt has in mind the Johannine bread-of-life discourse, which Eusebius had included in Canon I as parallel to the Synoptic last supper accounts (Mt §284; Mk §165; Lk §266; Jn §55, 63, 65, 67). Again, when Augustine comes to Matthew’s chronological reference to the Passover at the beginning of the Passion narrative, he observes that Mark and Luke confirm this statement, though John gives a similar temporal indicator ‘three separate times’.³⁸ He thus shows awareness of all of the parallels recorded at this point in Canon I (Mt §274; Mk §156; Lk §260; Jn §20, 48, 96). Examples could be multiplied further but these suffice to indicate the kind of comparisons Augustine makes in the treatise. On occasion Augustine recognizes a Eusebian parallel but concludes that the parallel accounts refer to separate historical events. For example, although Matthew, Mark, and Luke record the healing of a blind man in the vicinity of Jericho (Mt §205; Mk §116; Lk §224), parallels included in Eusebius’ Canon II, Augustine argues that Luke’s incident must refer to another blind man, given the differences in the versions.³⁹ He similarly treats the cleansing of the temple (Mt §211; Mk §121; Lk §238; Jn §21), a Canon I parallel; the parable of the invitation to the banquet or wedding feast (Mt §221; Lk §181), a Canon V parallel; and the anointing of Jesus by a woman, another Canon I parallel (Mt §276; Mk §158; Lk §74; Jn §98).⁴⁰ Penna highlighted Augustine’s treatment of these passages and interpreted it as evidence that he was not using the Canon Tables.⁴¹ However, as suggested above, the mere fact that Augustine shows an awareness that these passages stand in some kind of relation to one another, and so require explanation, suggests possible dependence upon Eusebius. I can find only two passages that might count as genuine negative evidence against Augustine’s reliance on the Canon Tables. In a brief paragraph summarizing John chapters four to six, Augustine mentions the healing of the invalid at the pool in Jerusalem, and explicitly remarks that this is something ³⁷ cons. ev. 3.1.2 (trans. Paffenroth, p. 257). ³⁸ cons. ev. 2.78.152 (trans. Paffenroth, pp. 250 1). ³⁹ cons. ev. 2.65.125 6. ⁴⁰ cons. ev. 2.67.129; 2.71.139; 2.78.153 79.154. ⁴¹ Penna, ‘De Consensu Evangelistarum’, pp. 16 17. About the healing of the blind man at Jericho, Penna states, ‘Nemmeno una parola è spesa per spiegare come mai nei canoni figur assero come paralleli (canone II)’.

138

The Eusebian Canon Tables

John reports that the other three evangelists omit, despite the fact that Eusebius had included this as a parallel in Canon I (Mt §70; Mk §20; Lk §37; Jn §38).⁴² Similarly, Augustine points out that Luke alone recorded Jesus’ post-resurrection eating in the upper room, while Eusebius had made this a parallel in Canon IX, placing it alongside the Johannine post-resurrection eating by the lake (Lk §341; Jn §221, 223, 225). However, neither of these passages necessarily implies that he was not dependent upon Eusebius’ tables. In each case Augustine probably discerned sufficient differences between the various parallel accounts that he regarded them as episodes with distinct historical referents, and so concluded that they did not pose possible contradictions, just as he does with the other parallels noted in the previous paragraph, though in these cases he chose not to even mention the corresponding passages from the other gospels. Thus, it is not necessary to read his silence about these parallel episodes as evidence against his reliance on Eusebius’ system. There is a further type of passage in De consensu that suggests dependence upon Eusebius. On occasion Augustine makes reference very briefly or allusively to parallels without exploring them in depth. It may be that in such cases he himself had flipped the pages in his codex to find the passages in question, or was simply relying on memory, but his comments seem rather to give the impression that he was quickly glancing in the margin to note what possible parallels there were with other gospels. The clearest example of this kind of reference occurs in De consensu 4.6.7. At the opening of the paragraph Augustine identifies the passage he is considering as Mark 9:41–50 (Mk §98–102), a brief passage, much shorter than many others considered throughout the treatise. In his cursory survey of these verses Augustine points out that ‘in this section [Mark] also records some things that are not in any of the other evangelists, and some that are also in Matthew, and some that are also in Matthew and Luke. But these things are in the other gospels in different contexts and in another order’.⁴³ Augustine explores none of these other parallel passages but instead moves directly into a digression about the sacraments. The question is how he became aware of these parallels and why he bothered mentioning them. He could have discovered them on his own by expending a great deal of energy tracking them down, but it is not clear why he would have bothered doing so, since these parallels are irrelevant to his aim in this part of the treatise. Alternatively, he could simply have looked in the margin of his codex and noticed that Mk §98 and 100 were in Canon VI and so shared parallels with Matthew, Mk §99, and 102 were in Canon II and so had parallels in both Matthew and Luke, while Mk §101 was in Canon X and so was unique to Mark. ⁴² cons. ev. 2.45.94. ⁴³ cons. ev. 4.6.7 (CSEL 43, p. 401; trans. Paffenroth, p. 322): contextim commemorat aliqua ponens, quae nullus alius euangelistarum posuit, aliqua uero, quae mattheus quoque posuit, et aliqua itidem, quae et mattheus et lucas, sed et illi ex aliis occasionibus et in alio rerum ordine.

‘Diversity of Agreement among the Evangelists’

139

We come now to the final type of indication that Augustine was dependent upon Eusebius’ Canons—agreement in a string of parallels. This comes from book four and represents the most compelling evidence to be found in the treatise. As already noted, in book four Augustine sets himself the task of analysing the remaining material in the latter three gospels that has not already been considered in relation to the Gospel of Matthew (discussed in book two) and that occurs before the Passion narrative (discussed in book three). He therefore commences by working through the Gospel of Mark looking for parallels with the Gospel of Luke. But the task is actually more difficult even than that. It is obvious to any casual reader that Mark and Luke share much material, but most of this is also shared with Matthew as well, which Augustine has already considered. He therefore has the difficult job of identifying those passages in Mark that have a parallel only in Luke. This was precisely the information conveyed by Eusebius’ Canon VIII, and, as I will argue, Augustine’s dependence upon the Canon Tables is indicated by the fact that he shows an awareness of every Canon VIII parallel prior to the Passion account. Here we must bear in mind Augustine’s method for identifying chunks of text to be considered in each section of his treatise. In the absence of chapters and verses he does so by citing incipits and explicits. Sometimes these are very large sections and sometimes rather small. Either way, in every case we must consider what it is that has made Augustine break his text where he has—why, that is, he has stopped at a particular point. With one exception, in his treatment of Mark he breaks the text when he finds a parallel with Luke. In De consensu 4.1.2 he considers Mark 1:1–21, while Mk §12 (= Mk 1:21) is the first Canon VIII parallel he comes across, successfully identifying it even though it is only a single verse and, as he points out, poses no problem of contradiction. In De consensu 4.2.3 he carries on with Mark 1:22–39, stopping again at the end of a Canon VIII passage (Mk §17 = Mk 1:35–39). This range of verses not only ends with a Canon VIII passage but also includes within it two other Canon VIII passages, Mk §14 and 16 (= Mk 1:23–28, 34b). As he begins his discussion of the section, Augustine explains, ‘in this entire passage there are certain things that [Mark] says with Luke alone’.⁴⁴ The use of the plural quaedam suggests that he has in mind more than just a single parallel with Luke, in other words, that he is aware of both the Canon VIII parallels internal to this section of Mark as well as the Canon VIII parallel that marked the end of the section he is considering with this paragraph. Moreover, he explains that he is not going to deal with these internal parallels with Luke because he has already done so. Indeed, in De consensu 2.17.35 Augustine already noted the Canon VIII parallels Mk §14; Lk §25 and Mk §16; Lk §27, which justifies his skipping of these passages in the survey of Mark–Luke parallels in book four. ⁴⁴ cons. ev. 4.2.3 (CSEL 43, p. 395; trans. my own): in hoc toto loco quamuis sint quaedam quae cum solo luca dixit.

140

The Eusebian Canon Tables

Next, in De consensu 4.3.4 Augustine considers a larger chunk of text, Mark 1:40–3:12. Again, it may appear that he chooses to stop his citation at 3:12 at random, but in fact he does so because this is the next Canon VIII passage in Mark, specifically Mk §28 (= Mk 3:11b–12). Noting that this parallel presents no difficulties, he quickly moves on to the next section, Mark 3:13–5:20, an even larger chunk of text. Once more, his decision to stop at 5:20 strikingly corresponds with the conclusion of the next Canon VIII passage, Mk §48 (= Mk 5:18–20), which is parallel to Lk §84. In the next paragraph Augustine considers Mark 5:21–6:30, once more stopping at a Canon VIII passage, Mk §61, which is equivalent to a single short verse in modern reckoning (Mk 6:30). After identifying the passage he briefly remarks, ‘This last part is also given by Luke without any disagreement, and the rest has already been treated’.⁴⁵ The first half of this sentence refers to Mk §61, where he has stopped his citation. If so, then ‘the rest’ (cetera) alluded to in the second half of this remark must refer to a Canon VIII passage also included in this section but passed over without comment, Mk §56 (Mk 6:12–13). He then turns to consider Mark 6:31–7:37, stopping again at a Canon VIII passage, Mk §75 (= Mk 7:36b–37a), once again explaining that there is nothing in the latter passage to contradict its Lukan parallel (Lk §100).⁴⁶ Next, in De consensu 4.5.6 Augustine deals with Mark 8:1–9:39. By now it should come as no surprise that he has stopped his citation where he has found another Canon VIII passage, Mk §97 (= Mk 9:37–9), which is parallel to Lk §103.⁴⁷ The following paragraph, De consensu 4.6.7, represents the only break with the pattern we have been following. Here Augustine cites Mark 9:40–9, which neither ends with nor includes a Canon VIII parallel. It is unclear why he stops at this point, but even he seems to recognize it as a departure from his stated purpose, because he observes that this chunk of text includes material unique to Mark; material common to Matthew and Mark; and finally material common to Matthew, Mark, and Luke; but nothing that is similar in Mark and Luke alone. Finally, in the last paragraph devoted to the Mark–Luke comparison, Augustine in De consensu 4.7.8 cites Mark 10:1–12:44, another lengthy chunk of text that concludes with a Canon VIII passage, Mk §136 (= Mk 12:40–4), which is parallel to Lk §247. Augustine has not quite reached the start of the Passion narrative in Mark, which commences with Mark 14. However, he explains that between this portion of text and the beginning of the Passion, ‘Mark says nothing that would oblige us to compare or inquire as to whether there seems ⁴⁵ cons. ev. 4.4.5 (CSEL 43, p. 397; trans. Paffenroth, p. 320, modified): hoc ultimum dixit cum luca nihilo discordans, cetera iam ante tractata sunt. ⁴⁶ cons. ev. 4.4.5 (CSEL 43, p. 397): in his cum luca marcus nihil est quod repugnare uideatur et superiora omnia iam considerauimus, quando eos mattheo conferebamus. ⁴⁷ The verse enumeration here is slightly different between the Greek NT and the Vulgate. In NA²⁸ Mk §97 corresponds to Mk 9:38 40 but in the Vulgate it corresponds to Mark 9:37 9, even though the textual content in each section is identical.

‘Diversity of Agreement among the Evangelists’

141

to be any disagreement’.⁴⁸ In other words, the remaining material prior to the Passion has no Canon VIII parallels which would require investigation. From the preceding discussion it should be clear that Augustine has cited or alluded to every Canon VIII passage in sequence from the beginning of Mark’s gospel to the Passion narrative, even though many of these are very short sections and almost none of them even pose any potential contradiction. If further evidence was needed that he was following Eusebius’ system we need only turn to the subsequent section of book four in which Augustine pivots to consider the Luke–John parallels that Eusebius had recorded in Canon IX. Prior to the Passion Eusebius noted only a single parallel, one that was perhaps unexpected, by coordinating the great catch in Lk §30 (= Lk 5:4–7) with the post-resurrection great catch of fish in Jn §219, 222 (= Jn 21:1–6, 11). This was a parallel at the literary or thematic level, but no one would ever construe these two passages as referring to the same historical event, since they occur at completely different points in the narrative sequences of Luke and John and present a variety of other differences, such as the fact that in Luke Jesus is in the boat, while in John he is standing on the shore. Since Augustine was primarily interested in explaining events at the level of historical veracity, we might then expect that he would not bother to consider this parallel given how different the two passages are, but, in fact, in De consensu 4.9.10 he begins his study of Luke by citing Luke 1:5–5:4, stopping where Eusebius began his Canon IX parallel. After arguing that these were not the same event, Augustine remarks ‘Luke did not say anything else similar to John except around the Lord’s Passion and resurrection’.⁴⁹ In other words, Augustine observes that there are no further Canon IX passages prior to the Passion so his analysis of Luke–John parallels is finished. Thus, as with Canon VIII, so also with Canon IX, Augustine notes every Eusebian parallel prior to the Passion, even though in this instance there is only one to consider. The overlap in book four between the passages considered by Augustine and the parallels flagged by Eusebius, in the same precise sequence, goes beyond what could reasonably be accounted for simply by appealing to coincidence. Moreover, the difficulty of the task Augustine has set for himself argues for his dependence on some sort of cross-referencing system. Even for someone who knew the gospels very well, who had much of them memorized, as Augustine no doubt did, identifying the Mark–Luke parallels through purely mental reflection is exceedingly difficult, since these are passages that have to satisfy two conditions, both their presence in Mark and Luke and their absence in Matthew and John. As a result, some sort of offloading of the ⁴⁸ cons. ev. 4.7.8 (CSEL 43, p. 403; trans. Paffenroth, p. 323): non dicit et marcus quod cogat cum aliquo conparari ad inquirendum, ne quid repugnare uideatur. ⁴⁹ cons. ev. 4.9.10 (CSEL 43, p. 406; trans. my own): iam cetera similia iohanni lucas non dixit nisi circa domini passionem et resurrectionem.

142

The Eusebian Canon Tables

cognitive process in written form would be required, precisely the sort of information provided by Eusebius’ paratextual apparatus. If Augustine did not have access to Eusebius’ Canons, then he must have created his own system of notation in the margins of his codex to accomplish this feat. Yet, the simplest explanation for Augustine’s method in book four, and the one that accords best with the external evidence for his access to the Vulgate, is that as he wrote he was flipping through the pages of his codex, stopping when he saw the numeral for the appropriate Canon noted in the margin, turning to the tables at the front to find its parallels in the other gospels, and then either explaining the parallel passages or dismissing them as unproblematic. Thus, the clearest evidence for Augustine’s dependence upon the Canon Tables of Eusebius comes from the end of the treatise, in book four. Moreover, the fact that dependence on the marginal apparatus is, I suggest, undeniable for book four implies that the more probabilistic instances of possible dependence in books two and three, which were noted previously, may plausibly be interpreted along the same lines. The case is similar with respect to the important passage from book one highlighted by Zahn and Watson, but omitted from my analysis thus far. Augustine begins the treatise by giving a brief account of the church’s tradition regarding the origins of the gospels and their authorship, and the respective theological emphases of each. However, in the midst of this rather traditional discussion he inserts a brief, almost parenthetical comment about how the gospels relate to one another. Here is the passage, with the addition of enumeration for each of the relevant propositions included: 1) Mark, as if following [Matthew], seems like his attendant and summarizer. marcus eum subsecutus tamquam pedisequus et breuiator eius uidetur. 2) [Mark] said nothing with John alone; cum solo quippe iohanne nihil dixit, 3) [he said] only a few items unique to himself; solus ipse perpauca, 4) [he said] even less with Luke alone cum solo luca pauciora, 5) But with Matthew [he said] very many things, and many with nearly the same number of words or identical words; whether by agreement with [Matthew] alone cum mattheo uero plurima et multa paene totidem adque ipsis uerbis siue cum solo 6) or also with the others. siue cum ceteris consonante.⁵⁰ ⁵⁰ cons. ev. 1.2.4 (CSEL 43, p. 4; trans. Paffenroth, p. 140, modified). Paffenroth translates the above passage as Mark not ‘sharing’ material with John and ‘having agreements’ with Matthew.

‘Diversity of Agreement among the Evangelists’

143

In one sense, this information serves no purpose at this stage of Augustine’s argument. The point he needs to make is that Mark ‘follows’ Matthew (proposition one), but he has nevertheless gone further into a brief excursus about Mark’s relationship to the other gospels. One might hypothesize that, in order to make his statement about Mark’s relation to Matthew, he has turned to the Canon Tables to see just how much they have in common, and has ended up giving rather more information than was strictly needed. He could easily have discerned proposition two from the absence of a Mark–John Canon, one of the two possible combinations omitted by Eusebius. (And here again, the difficulty of discerning this fact in the absence of Eusebius’ apparatus should be acknowledged, since this is a claim about passages that satisfy two conditions—their presence in John and Mark and absence in the other gospels. John and Mark do have much in common, but all of it is also shared with Matthew, and so is included in Canon III. Hence the word solo in Augustine’s statement is key.) Proposition three would then be drawn from Canon XMk, since Mark here has only 19 unique passages, far fewer than any of the Canon X lists for the other gospels. Proposition four then refers to Canon VIII and is a comparison with Canon XMk. Canon VIII has 13 parallels, so ‘fewer’ than the 19 included in Canon XMk. Finally, with proposition five Augustine returns to his topic, noting that there are ‘very many’ agreements between Matthew and Mark, information he could have gleaned from Canon VI, which lists 47 parallels between these two gospels alone. Proposition six then rounds out Mark’s possible relationship with the other gospels by alluding to the other Canons that also contain Mark and Matthew: Canon I, which encompasses 74 agreements among all four gospels; Canon II, listing 111 parallels among Matthew, Mark, and Luke; and Canon IV which includes 25 parallels among Matthew, Mark and John. One might read this passage as if Augustine were simply reporting these observations on the basis of his own acquaintance with the gospels. However, the precision of his remarks, especially the comparison between propositions three and four, suggests otherwise.⁵¹ It would be impossible to state confidently that Mark has less material in common with Luke alone than he has unique material unless one had actually tabulated the information and calculated the results, precisely the work that Eusebius had done and passed on to posterity in the Canon Tables. Given the strong indication from book four that Augustine was working from a copy of the Canon Tables, we may conclude with some confidence that already in book one he was drawing This obscures the fact that the sole verb throughout is the dixit in proposition two. To translate as Paffenroth does makes it sound as though Augustine envisions source critical relationships amongst the gospels, something that is not at all implied by the dixit. On this point, see the discussion in the subsequent section of this chapter. ⁵¹ Cf. Watson, Gospel Writing, p. 18.

144

The Eusebian Canon Tables

this information from Eusebius’ apparatus as translated by Jerome, giving a prose summary of the visual impression produced by the ordered information in the prefatory tables at the start of his codex. And if this is true for books one and four, there is no reason not to conclude that throughout books two and three Augustine was also using the tables to find parallel passages,. Although Augustine could have found a number of these parallels on his own, it would have been impossible to do such a thorough job of ferreting out potential contradictions without the aid of some sort of cross-referencing tool such as Eusebius provided. Indeed, the ambition of his project was grand. Although it is usually assumed that the work was an apologetic one, intended to stave off either pagan or Manichaean criticisms of the gospels, Augustine’s explicit polemic appears only in book 1, and his opponents disappear from view thereafter. Moreover, there is no indication that he was dealing with a set list of conventional problems posed by an actual opponent, whether contemporary or past. In fact, Augustine worked systematically through the gospels, probably creating far more problems than pagan or Manichaean opponents would even have been aware of, and then solved the problems that he himself had brought to light.⁵² His exhaustive methodology represents the exercise of an inquisitive mind upon the issues that arise from placing four versions of the same story alongside one another, and produces an impressive display of erudition that would have overwhelmed any potential critics of the gospels and assured the consciences of any troubled Christians. His achievement in this work should not be underestimated. Although it is true that ‘a modern reader would likely admit that a careful review of the three exegetical books of the cons. ev. can be tedious and that Augustine’s attempts to treat virtually every discrepancy among the Gospels may lead to a certain scrupulosity that is irrelevant to grasping the pertinent facts of revelation’,⁵³ what he produced was the most extensive and comprehensive study of gospel parallels up to this point in history, one that would not be surpassed for centuries. What we observe in Augustine, therefore, is the way in which an innovative paratextual technology enabled a more thorough and in-depth analysis of the texts in question which exceeded all prior attempts. Hence, when, in writing to Jerome in 403, Augustine described his Vulgate translation of the gospels as labor tam utilis, he may have had in mind not simply his revised version of the text of the gospels but also the new marginal apparatus Jerome had included, which Augustine himself had recently put to

⁵² I am grateful to Francis Watson for suggesting this idea to me. Cf. Magny, Porphyry in Fragments, p. 119: ‘most of the questions present in the text are [Augustine’s] own, and he is merely anticipating the questions from pagans or the shaken faithful’. ⁵³ Dunkle, ‘Humility, Prophecy, and Augustine’s Harmony’, p. 225. Dunkle, however, con tinued by arguing, ‘Yet the rigor of Augustine’s approach reflects his conviction that the reader of the Gospel must imitate its author, as a humble servant’.

‘Diversity of Agreement among the Evangelists’

145

good use in his treatise De consensu evangelistarum.⁵⁴ In fact, in his later Tractates on the Gospel of John, Augustine twice alluded to De consensu, describing it as a ‘laborious’ (laboriosus) work in which difficult questions are investigated ‘in a most painstaking manner’ (operosissime).⁵⁵ After having undertaken the most detailed study of gospel parallels yet attempted, his admission of the difficulty of the task echoes Eusebius’ praise for the ‘hard work and study’ of his predecessor Ammonius. In fact, we should see a direct line of continuity from Ammonius through Eusebius to Augustine (mediated, of course, by Jerome), each building upon the possibilities created by the scholarly endeavours of the former in an attempt to plumb the depths of the intractable challenge inherent within the church’s canonical scriptures.

M E MOR I A RE RU M, M E M O R I A VE R B O R U M, A N D T H E T R U T H O F T H E GO S P E L The preceding analysis should suffice to demonstrate that Augustine had recourse to the Canon Tables in the research he conducted for the purposes of his treatise. Beyond the mere fact of his use, however, the question then arises of how his reliance upon Eusebius’ apparatus shaped his understanding of the gospels. He of course began his treatise with the assumption that these texts were harmonious with one another, a conviction that he maintained to the end. Nevertheless, we can, I believe, observe Augustine, through the process of writing the treatise, coming to grips with the complexity of the gospels’ relationships to one another and formulating the outlines of a rhetorically shaped account of their unity in difference. One strand of research on De consensu has sought to find in Augustine a precedent for a certain view on the source-critical question of the relationships among the gospels. The so-called ‘Augustinian hypothesis’ is that the chronological sequence in which the evangelists wrote corresponds to their ordering in the Vulgate canon (which notably differs from the Old Latin order of the gospels), and, more importantly, that each evangelist wrote with a full awareness of what his predecessor(s) had written. This reading of Augustine goes back at least to the sixteenth century and continues to be presented as one of several possible

⁵⁴ I am grateful to Hugh Houghton for suggesting this possibility to me. ⁵⁵ Io. ev. tr. 112.1; 117.2 (CCSL 36, pp. 633, 652). Agreeing in this assessment, M. B. Riddle asserted that cons. ev. ‘is regarded as the most laborious task undertaken by the great African Father’, and S. D. F. Salmond commented that, ‘Among Augustin’s numerous theological productions, this one takes rank with the most toilsome and exhaustive’ (NPNF, first series, vol. 6, pp. 67, 71).

146

The Eusebian Canon Tables

solutions to the Synoptic Problem.⁵⁶ Key to this interpretation is the passage from De consensu 1.2.4 dealt with in the last section, particularly Augustine’s statement that ‘it is not the case that any one of [the evangelists] can be found to have chosen to write in ignorance of his predecessor, or to have omitted as unknown something which another is found to have written’.⁵⁷ Some have even gone further and proposed that Augustine changed his view on Synoptic relationships as a result of the close study of the gospels he engaged in throughout the course of the treatise.⁵⁸ On this reading, whereas in book one he set out the notion that Mark abbreviated Matthew’s gospel, by the time he reached book four, he had realized that Mark most likely made use of both Matthew and Luke: ‘[Mark] seems to have been Matthew’s companion . . . Or it is more likely to understand that he goes together with both [Matthew and Luke], for, although Mark agrees with Matthew in more places, nonetheless in other places he agrees more with Luke’.⁵⁹ Hence, Augustine becomes the earliest proponent of what is now known as the Griesbach hypothesis, defended in recent times most prominently by William Farmer and his students.⁶⁰ If this interpretation of Augustine’s statements in the treatise is accurate, then we would have grounds for asserting that it was precisely Eusebius’ Canons that enabled Augustine to study the internal evidence of the gospels and arrive at a new source-critical hypothesis regarding the relationships of dependency that exist amongst them. Nevertheless, there are good reasons for questioning whether Augustine is much interested in such source-critical questions. This line of interpretation of De consensu has been seriously called into question in a short study by H. J. de Jonge, who pointed out that throughout De consensu Augustine makes statements that are inconsistent with any rigid source-critical theory of gospel origins. After surveying several illuminating examples, de Jonge concluded, ‘That Augustine’s phrases omittere, praetermittere, addere, and interponere should not be taken as technical terms can also be seen from the fact that in his

⁵⁶ Cf. H. J. de Jonge, ‘Augustine on the Interrelations of the Gospels’, in The Four Gospels (Fs Frans Neirynck), ed. F. Van Segbroeck (1992), p. 2417 n. 22: ‘So far as I know the first author to ascribe the “Augustinian hypothesis” concerning the relationships between the Gospels to Augustine was M. Chemnitz’. ⁵⁷ cons. ev. 1.2.4 (CSEL 43, p. 4; trans. Paffenroth, p. 140). non tamen unusquisque eorum uelut alterius praecedentis ignarus uoluisse scribere repperitur uel ignorata praetermisisse, quae scripsisse alius inuenitur. ⁵⁸ According to Dungan, A History of the Synoptic Problem, p. 442 n. 45, the first scholar in the modern period to draw attention to Augustine’s change of mind was Peabody in ‘Augustine and the Augustinian Hypothesis’. Dungan himself followed this reading (p. 122), as did Watson, Gospel Writing, pp. 20 4. ⁵⁹ cons. ev. 4.10.11 (CSEL 43, pp. 406 7; trans. Paffenroth, p. 325). matthei magis comes uidetur . . . uel, quod probabilius intellegitur, cum ambobus incedit. nam quamuis mattheo in pluribus, tamen in aliis nonnullis lucae magis congruit. ⁶⁰ William R. Farmer, The Synoptic Problem: A Critical Analysis (New York: Macmillan, 1964).

‘Diversity of Agreement among the Evangelists’

147

De consensu evangelistarum each of the four evangelists can be said to have “added” or “omitted” something in comparison with any other evangelist’.⁶¹ In contrast to the source-critical theories that have been ascribed to Augustine, de Jonge argued that he held that ‘each of the four evangelists had access to exactly the same amount of information about Jesus’ words and deeds. It is on this common reservoir of information that, under the guidance of the Spirit, each evangelist individually drew the stories and sayings of Jesus that he recorded in his Gospel’.⁶² I find de Jonge’s arguments compelling. The closest Augustine comes to a source-critical judgement about the gospels is in De consensu 1.2.4, but even here all he claims is that the evangelists wrote with an awareness of their immediate predecessor, which is short of claiming that each evangelist used his predecessor as a source for his own composition. Moreover, even Augustine’s claim that Mark was Matthew’s breuiator is hedged by the verb uidetur, and as such is short of a full affirmation of this relationship. Moreover, even if Augustine envisions literary dependency in 1.2.4 and 4.10.11, these brief comments play no role whatsoever in the actual exegetical research carried out throughout the treatise.⁶³ The Canon Tables themselves might partially account for this lack of interest in source criticism, since the paratextual apparatus creates various links across the fourfold canon, taking into account all possible relations between them, which complicates any attempt at giving a source-critical reading. Yet if he had really thought Mark was an abbreviator of Matthew, Augustine should have simply compared Mark to Matthew to confirm this hypothesis. But in fact he did not do so, and instead hypertextually jumped constantly between various gospels, using the intertextual itineraries created by Eusebius’ apparatus. This is not to suggest that Augustine did not come up with a theory to account for the textual data he found in his codex. In fact he did, but it was one derived from classical rhetoric which does not align with any of the solutions to the so-called Synoptic Problem proposed in modern New Testament source criticism. Given his training and profession, it is hardly surprising that rhetorical traditions informed Augustine’s interpretation of the gospels in De consensu, but this fact seems to have gone unrecognized. The two key ways in which this rhetorical tradition is apparent in the treatise are in Augustine’s persistent emphasis on the evangelists’ memoria and in his repeated distinction between res and verba.

⁶¹ de Jonge, ‘Augustine on the Interrelations of the Gospels’, p. 2413. ⁶² de Jonge, ‘Augustine on the Interrelations of the Gospels’, p. 2416. ⁶³ Cf. de Jonge, ‘Augustine on the Interrelations of the Gospels’, p. 2413: ‘These examples make it sufficiently clear that when Augustine tried to explain the discrepancies between the Gospels, the so called “Augustinian hypothesis” was not his point of departure’. Similarly, Watson, Gospel Writing, p. 22 n. 34, conceded that the opening and concluding remarks in cons. ev. ‘exercise little or no influence on its central argument’.

148

The Eusebian Canon Tables

By at least the first century BCE, memory had become codified as the fourth part of rhetorical theory, as seen in the Rhetorica ad Herennium.⁶⁴ Quintilian later carried forward this tradition, praising memory as that upon which ‘all learning depends’ (omnis disciplina memoria constat).⁶⁵ However, the closest parallels for Augustine’s references to memory in De consensu come from two later sources. Consultus Fortunatianus was a fourth- or fifth-century rhetorician whose three-book Ars Rhetorica used Quintilian as a principal source and mediated much of the rhetorical tradition to later medieval authors. In his discussion of memory, Fortunatianus made a distinction between memoria verborum and memoria rerum: Should we always memorize verbatim? If time permits, yes; if not, we will retain only the substance and later will match the words to it extemporaneously. Semper ad verbum ediscendum est? si tempus permiserit: sin minus, res ipsas tenebimus solas, dehinc his verba de tempore accommodabimus.⁶⁶

The fifth-century author Martianus Capella, who authored an influential survey of the liberal arts, similarly commented, Memory is applied to things and to words, but words do not always have to be memorized; unless there is time for this, it will be enough for anyone to hold the matter itself in the mind, especially if nothing comes naturally from the memory. est quidem memoria rerum atque verborum, sed non semper ediscenda sunt verba, nisi spatium meditandi tempus indulserit, sat erit que res ipsas animo tenuisse, praesertim si nihil eius munere naturali provenerit.⁶⁷

While we need not suppose that Augustine was drawing directly on one of these texts as a source, they represent the common rhetorical understanding of memory in his day, and as such are indicative of the sort of memory pedagogy in which he was trained as a rhetor, and that he probably taught to his own students.⁶⁸ They demonstrate that late antique rhetors were familiar with the limitations of human memory, especially when time is limited, and advised that it is often best therefore to memorize the substance, the

⁶⁴ George A. Kennedy, A New History of Classical Rhetoric (Princeton, NJ: Princeton Uni versity Press, 1994), p. 6. See especially rhet. Her. 3.24.39 for the distinction between memoria rerum and memoria verborum. ⁶⁵ Quintilian, inst. 11.2.1 (LCL 494, p. 58). Cf. Mary J. Carruthers, The Book of Memory: A Study of Memory in Medieval Culture, 2nd edn, Cambridge Studies in Medieval Literature (Cambridge: Cambridge University Press, 2008), pp. 92 3. ⁶⁶ Fortunatianus, rhet. 3.14 (Halm, p. 129; trans. Carruthers and Ziolkowski, p. 297). On the distinction between these two types of memory, see Carruthers and Ziolkowski, The Medieval Craft of Memory, 9 11; Carruthers, The Book of Memory, pp. 92 3, 110 16. ⁶⁷ Martianus Capella, nupt. 5.539 (Willis, p. 190; trans. Stahl, Johnson, and Burge, p. 204). ⁶⁸ Cf. Carruthers, The Book of Memory, pp. 182 3, who noted, ‘Like Martianus Capella, Fortunatianus represents the mnemotechnical pedagogy which Augustine, Marius Victorinus, and other teachers in late antiquity valued’.

‘Diversity of Agreement among the Evangelists’

149

res, of a text rather than to memorize it verbatim. Moreover, Fortunatianus recognizes that one advantage of memoria rerum over memoria verborum is that the former allows the speaker to adapt his wording to the situation at hand, rather than being strictly constrained by the wording of his source. As emphasized by Mary Carruthers in her seminal study of this topic, ‘Memoria ad res compels the recollector to actively shape up material for an occasion, whether as composer or viewer or reader . . . Especially in composition, memory for things is preferred to rote iteration, even when the speaker has accurate command of the original words’.⁶⁹ This is precisely the overlooked rhetorical background upon which Augustine drew in articulating his account of gospel similarity and difference in De consensu. Throughout the treatise Augustine makes passing references to the memory of the evangelists as playing a key role in their method of composition. For example, early in book two, he notes that they ‘have set down these things as they remembered (meminerat) them’.⁷⁰ Further on in book two he explains that Matthew wrote Matthew 8:18 because it was something he ‘recalled’ (recordatum) at this point, and that Luke similarly ‘remembered’ (recordantis modo) Luke 9:57–62 out of ‘chronological sequence’ (ordine temporum), which explains its different placement in Luke’s gospel compared with Matthew’s.⁷¹ He provides the same explanation for the discrepancy amongst the Synoptics over the narration of John the Baptist’s condemnation of Herod’s relation with Herodias, and Herod’s subsequent imprisonment of John. Matthew and Mark both reported this much later in their narratives (Mt 14:3–12; Mk 6:17–29), while Luke gave a briefer mention just prior to Jesus’ baptism (Lk 3:19–20). Augustine explains that this divergence arose because Luke, ‘recalling the incident in connection with the present occasion (ex occasione recordatum), brought it in here by anticipation and put it in his narrative before many other things in his narrative that really occurred before this happened to John’.⁷² These passing references to memory are further explained by a handful of more extensive passages in which Augustine sketches the outlines of a theory of how the gospels were written. Some of these have previously been highlighted in secondary scholarship on the treatise,⁷³ though the central role played by memory seems largely to have been overlooked. First, as noted by de Jonge, Augustine assumes that all of the evangelists have access to the same common fund or reservoir of material about Jesus, whether from first-hand acquaintance with the events in question (Matthew and John) or through

⁶⁹ ⁷⁰ ⁷¹ ⁷² ⁷³

Carruthers, The Book of Memory, p. 93. cons. ev. 2.12.27 (CSEL 43, p. 127; trans. Paffenroth, p. 189). cons. ev. 2.22.53 23.54 (CSEL 43, p. 155; trans. Paffenroth, pp. 202 3). cons. ev. 2.44.92 (CSEL 43, p. 198; trans. Paffenroth, p. 222, modified). See especially Harrison, ‘Not Words But Things’, pp. 162 3.

150

The Eusebian Canon Tables

tradition passed on to them (Mark and Luke),⁷⁴ and that it was from this common source that they each drew material for their compositions. As he writes in book three, ‘there were many statements and answers made, and out of these the writers chose what seemed right for their narratives, putting into them what they thought sufficient’.⁷⁵ The mechanism by which the evangelists stored and later accessed this common fund was their memory. As he writes early in book two, ‘one should not suppose that one of [the evangelists] is deceitful if, in their account of something they saw or heard, several people remember (reminiscentibus) it in not exactly the same way or in not exactly the same words but nonetheless describe the same thing (eadem . . . res)’.⁷⁶ Augustine’s contrast in this passage between memory of ‘words’ and memory of ‘things’ draws directly on the rhetorical distinction between memoria verborum and memoria rerum, and it is the latter that he thinks applies in the case of the evangelists.⁷⁷ As implied in the last passage, Augustine was well aware that this method of composition is constrained by the limitations of the human mind. He, perhaps more than anyone else in late antiquity, explored the intricacies of human memory, and these reflections occasionally surface in De consensu.⁷⁸ For example, he writes in book three, ‘human memory drifts through a variety of thoughts, and no one can control what kind of thought comes into his mind, nor when it does so’.⁷⁹ It is here, however, that Augustine inserts a role for divine agency into the evangelists’ compositional practice. Human memory might be unpredictable but the evangelists’ ‘recollections were governed by the hand of the one who governs the waters however it pleases him’.⁸⁰ This

⁷⁴ cons. ev. 1.1.1 1.2. ⁷⁵ cons. ev. 3.8.35 (CSEL 43, p. 315; trans. Paffenroth, p. 279). multa enim dicta et multa responsa sunt, unde cuique eorum quantum uisum est decerpsit et in narratione sua posuit quod satis esse iudicauit. ⁷⁶ cons. ev. 2.12.28 (CSEL 43, p. 127; trans. Paffenroth, p. 189). ⁷⁷ Furthermore, notice that Augustine does not speak of the evangelists remembering what they read as would be required for a source critical understanding of gospel relationships. Instead, he envisions the evangelists drawing upon either what they saw with their own eyes or what they heard, whether from Jesus or from others who knew him. This again illustrates that he is not thinking in terms of literary dependence as envisioned with respect to the modern Synoptic Problem. ⁷⁸ Cf. Carruthers, The Book of Memory, p. xi: ‘ “Apart from what I have forgotten” [a quotation from conf. x.8]: in the cheerful admission of that phrase lies an essential difference between a modern and a medieval understanding of the cognitive function of memory. To have forgotten things is seen by us now as a failure of knowledge, however ordinary a failure it may be, and therefore a reason to distrust the power of memory altogether. Yet to have forgotten some things was understood in Augustine’s culture as a necessary condition for remembering others’. ⁷⁹ cons. ev. 3.13.48 (CSEL 43, p. 334; trans. Paffenroth, p. 287). fluitat enim humana memoria per uarias cogitationes, nec in cuiusquam potestate est, quid et quando ei ueniat in mentem. ⁸⁰ cons. ev. 3.13.48 (CSEL 43, p. 334; trans. Paffenroth, p. 287). recordationes enim eorum eius manu gubernatae sunt qui gubernat aquam, sicut scriptum est, qualiter illi placuerit.

‘Diversity of Agreement among the Evangelists’

151

divine action governing the evangelists’ memory was not, for Augustine, in any way different from the way in which human memory normally functions: For no one has the ability, no matter how great and trustworthy his knowledge of events is, to determine the order in which he will remember them. For whether something comes into a person’s mind earlier or later is a matter not of our will but of how it is given to us. It is probably enough that each of the evangelists believed it was his duty to narrate events in the order in which God saw fit to suggest to his recollection the things he was narrating.⁸¹

Within the scope of this process, Augustine holds that the evangelists have a significant degree of freedom and are not merely passive in the face of divine action. Rather, as the Holy Spirit ‘guided their minds in the recollection of what to write’ (mentes . . . in recolendo quae scriberent . . . gubernans), he nevertheless ‘lets each one arrange his narrative one way and the other another way. With pious diligence one may investigate why this is so and one may even find an answer with divine help’.⁸² The evangelists therefore have scope for presenting their narrative in the manner that seems most suitable to them, though the fund of material they can draw upon for their creative composition is restricted by what providence prompts them to remember at any given moment. In summary, in De consensu Augustine presents the evangelists as though they composed their works following the principles of late antique rhetoric.⁸³ The actual words and actions of Jesus were the res, which the evangelists had stored in their memory either through first-hand experience or through access to such eyewitnesses. These res are, then, what they sought to communicate to their readers, under the prompting of the Spirit, and as they did so they adapted their verba to what was most suitable to their respective narratives or audiences. Sometimes this meant presenting events in chronological sequence, but often it meant varying the sequence, a compositional practice that had long been recognized in the ancient literary tradition.⁸⁴ Fundamental to Augustine’s understanding of the gospels, therefore, was that the evangelists were intelligent composers of coherent literary works, whose differences one

⁸¹ cons. ev. 2.21.51 (CSEL 43, pp. 152 53; trans. Paffenroth, p. 201). quia enim nullius in potestate est, quamuis optime fideliter que res cognitas quo quisque ordine recordetur quid enim prius posterius ue homini ueniat in mentem, non est ut uolumus, sed ut datur , satis probabile est quod unusquisque euangelistarum eo se ordine credidit debuisse narrare, quo uoluisset deus ea ipsa quae narrabat eius recordationi suggerere. ⁸² cons. ev. 2.21.52 (CSEL 43, p. 153; trans. Paffenroth, p. 201, modified). alium sic, alium uero sic narrationem suam ordinare permiserit, quisquis pia diligentia quaesierit, diuinitus adiutus poterit inuenire. ⁸³ For a parallel argument that Augustine came to understand the Bible as operating accord ing to the principle of dispositio, the second of the five parts of classical rhetoric, see Michael Cameron, ‘ “She Arranges All Things Pleasingly” (Wis. 8:1): The Rhetorical Basis of Augustine’s Hermeneutic’, Augustinian Studies 41 (2010): pp. 55 67. ⁸⁴ Cf. Nünlist, The Ancient Critic, pp. 69 93.

152

The Eusebian Canon Tables

from another were deliberate, and therefore explicable through examination of each evangelist’s unique aims. The corollary to this understanding of gospel-writing is that readers of the gospels should not view the words of each evangelist as an end in themselves, as though they were a completely transparent window into the world about which they speak. Rather, the reader should be attentive to the way in which these words provide but one distinct perspective on the subject matter to which they refer.⁸⁵ Early in book two Augustine points out that the difficulty in deciding on the precise words of John the Baptist demonstrates that ‘what is necessary in order to know the truth is the meaning and not the exact words used’.⁸⁶ Or as he says further on in book two, ‘great variation in the words’ used by the evangelists does not preclude that ‘there is agreement as to the subjects and ideas’.⁸⁷ It is therefore in the res and sententiae, which flow from an author’s voluntas, that truth and harmony reside.⁸⁸ Perhaps most strikingly of all, in book three Augustine concedes that it is impossible to discern the ipsissima verba of Jesus himself.⁸⁹ The most extensive passage on this topic comes from the end of book two. Comparing the inspiration of the gospels with the legend of the seventy translators of the Septuagint, Augustine comments, They wished to show nothing other than that same thing which we now marvel at in the diversity of agreement among the four evangelists, in which it is shown to us that one need not be false if one expresses something in a different way from another, as long as one does not depart from the intention of the One with whom one must be in accord and agreement. Knowing this is advantageous both to morality, for the purpose of guarding against and judging falsehood, and to faith itself, lest we suppose that God would commend the truth to us by giving us not only the thing itself but even the words in which it is expressed, as though they were a kind of deified sound. It is rather the case that the thing which is to be learned is so far above the words by which it is learned that we would not need to

⁸⁵ This theme in cons. ev. has been most thoroughly analysed in Harrison, ‘Not Words But Things’, pp. 163 70. ⁸⁶ cons. ev. 2.12.27 (CSEL 43, p. 127; trans Paffenroth, p. 188): nullo modo hinc laborandum esse iudicat qui prudenter intellegit ipsas sententias esse necessarias cognoscendae ueritati, quibuslibet uerbis fuerint explicatae. Cf. cons. ev. 2.12.28 (CSEL 43, p. 128): quod ad doctrinam fidelem maxime pertinet, intellegeremus non tam uerborum quam rerum quaerendam uel amplectendam esse ueritatem, quando eos qui non eadem locutione utuntur, cum rebus sen tentiis que non discrepant, in eadem ueritate constitisse adprobamus. A similar statement is made at cons. ev. 3.4.14. ⁸⁷ cons. ev. 2.46.97 (CSEL 43, p. 205; trans. Paffenroth, p. 225): ex qua uniuersa uarietate uerborum, rerum autem sententiarum que concordia. ⁸⁸ Cf. Tarmo Toom, ‘Was Augustine an Intentionalist? Authorial Intention in Augustine’s Hermeneutics’, Studia Patristica 70 (2013): pp. 185 93. ⁸⁹ cons. ev. 3.2.8 (CSEL 43, p. 278; trans. Paffenroth, p. 261): si autem quaeruntur ipsa omnino uerba, quae petro dominus dixerit, neque inueniri possunt et superfluo quaeruntur.

‘Diversity of Agreement among the Evangelists’

153

ask about them at all if we were able to know the thing itself without them, as God knows it and his angels know it in him.⁹⁰

As emphasized towards the end of this passage, the reason the verba are relativized in Augustine’s view is because the res to which they refer so far exceeds the capacities of human speech and comprehension. When attempting to refer to a res of this nature, multiple perspectives on the truth are greatly useful, even necessary. Augustine no doubt assumed an unspecified limit with respect to the degree to which the verba can vary from historical reality, and exegetes today might argue for a fair bit more variance than he could countenance. Nevertheless, such statements indicate that he cannot merely be dismissed as the sort of naive literalist that he is sometimes made out to be. An illuminating demonstration of this principle comes in Augustine’s discussion of the divine words spoken at Jesus’ baptism. According to Matthew, the voice from heaven said, ‘This is my beloved son, with whom I am well pleased’ (Mt 3:17), while Mark and Luke both record, ‘You are my beloved son’ (Mk 1:11; Lk 3:22). For an interpreter who insists on exact identification of verba with historical reality this presents a dilemma, and the only way out of a denial of truth would be to suppose that the heavenly voice spoke both sentences in succession. For Augustine, however, it does not matter which of the evangelists records the actual words that were spoken since they all ‘give the same sense’ (eandem . . . sententiam).⁹¹ In fact, this discrepancy among the evangelists, far from being a problem, is in fact a boon: ‘This diversity in speech is also useful for this—that, with only one version of the expression, it might be harder to understand it, or it might be interpreted in a way other than the sense of the thing itself (se res). . . . whichever evangelist has preserved the actual words of the heavenly voice, the others have varied the words (verba) only in order to explain the sense (sententiam) of it more appropriately (familiarius)’.⁹² Read in the light of this principle, the variance between the versions in Matthew and Mark/Luke is intended to reveal that the divine voice was addressing both the

⁹⁰ cons. ev. 2.66.128 (CSEL 43, pp. 230 1; trans. Paffenroth, p. 238): nihil aliud demonstrare uoluerunt quam hoc ipsum quod nunc in euangelistarum quattuor concordi quadam diuersitate miramur, qua nobis ostenditur non esse mendacium, si quisquam ita diuerso modo aliquid narret, ut ab eius uoluntate, cui consonandum et consentiendum est, non recedat. quod nosse et moribus utile est propter cauenda et iudicanda mendacia et ipsi fidei, ne putemus quasi consecratis sonis ita muniri ueritatem, tamquam deus nobis quemadmodum ipsam rem, sic uerba, quae propter illam sunt dicenda, commendet, cum potius ita res, quae discenda est, sermonibus, per quos discenda est, praeferatur, ut istos omnino quaerere non deberemus, si eam sine his nosse possemus, sicut illam nouit deus et in ipso angeli eius. ⁹¹ cons. ev. 2.14.31 (CSEL 43, p. 131; trans Paffenroth, p. 191). The recognition of this point suggests that there is more similarity between the treatment of gospel contradictions by Origen and Augustine than is implied in the assessments of Dungan, A History of the Synoptic Problem, pp. 140 1, and Watson, Gospel Writing, pp. 549 50, both of whom prize Origen’s spiritualizing approach over the literal mindedness of Augustine. ⁹² cons. ev. 2.14.31 (CSEL 43, p. 132; trans Paffenroth, p. 191, modified).

154

The Eusebian Canon Tables

Son and the human onlookers, such that the latter group is drawn into the salvific work accomplished by the Son.

CONCLUSIO N It may seem that we have now wandered some distance from the Canon Tables as a tool for finding parallel passages amongst the four gospels, but I suggest that in fact in Augustine’s formulation of a theory of gospel-writing, we see him responding to the hermeneutical pressure exerted by the Eusebian apparatus. The set of marginal numbers that followed him as he worked through each gospel from start to finish served as a constant reminder that the claim to truth embodied within each text must be understood in relation to three competing accounts. In other words, none of the four could any longer be regarded as an exhaustive or sufficient presentation of the truth of the gospel once they were bound together in a single, indissoluble canon. Augustine characteristically turned this unavoidable conclusion, which others might have regarded as problematic, into a theologically productive assertion: a res as grand as the story of Jesus requires the multiple perspectives presented by the overlapping, interweaving narratives of the four gospels. One of the remarkable features of this account of gospel-writing is the degree of licence that Augustine granted to each of the evangelists. They were neither mechanically recording verbatim transcripts of what was actually said and happened, nor were they mere mouthpieces of the inspiring Spirit. Rather, each of the evangelists crafted his narrative in whatever way seemed to him best suited for conveying the theological res of Jesus’ story, with the Spirit prompting the memory and leaving the human author to compose the actual verba on the page. Hence, when encountering the resulting dialogue of voices in the fourfold canon, the reader should consider the distinct contribution each of the witnesses makes to the conversation.⁹³ It would be claiming too much to say that Augustine’s view of gospelwriting in De consensu evangelistarum is an entirely new formulation. Certainly it echoes a number of his other theological commitments already present before writing the treatise, specifically the distinction in De doctrina christiana between res and signa and his comments on scripture in Confessiones XII.⁹⁴ Nevertheless, it was an innovation to join these theological ⁹³ To this degree, then, Augustine stands in agreement with Werner H. Kelber, who proposed that ‘the project of retrieving the single original saying is contrary to the intentions of the tradition’ (Kelber, ‘The History of the Closure of Biblical Texts’, p. 121). ⁹⁴ As noted by Harrison, ‘Not Words But Things’, pp. 164 7. Cf. Tarmo Toom, ‘Augustine’s Hermeneutics: The Science of Divinely Given Signs’, in Patristic Theories of Biblical

‘Diversity of Agreement among the Evangelists’

155

notions with a rhetorical method of composition in order to account for the complex reality of the relationships of the gospels to one another. Hence, although the origins of the gospels and their sources had been discussed since Papias in the early second century,⁹⁵ the relationship of John to the Synoptics had been theorized since at least Clement of Alexandria in the late second century,⁹⁶ and in the early third century Origen had reflected at length on what it meant to call a writing a ‘gospel’,⁹⁷ in De consensu Augustine offered the first ever general theory of gospel-writing as an attempt to make sense of the empirical evidence of the four gospels, namely their similarities and differences. This creative account was but a synthesis of the textual information encoded in the Eusebian apparatus and would not have been possible apart from the Canon Tables.

Interpretation: The Latin Fathers, ed. Tarmo Toom (Cambridge: Cambridge University Press, 2016), pp. 77 108. ⁹⁵ See Eusebius’ quotation of the relevant fragments from Papias in HE 3.39.14 16, on which see Watson, Gospel Writing, pp. 121 31. ⁹⁶ See the relevant fragment from Clement’s lost Hypotyposes preserved by Eusebius in HE 6.14.7, on which see Watson, Gospel Writing, pp. 432 4. ⁹⁷ See Origen, Jo. 1.27 88.

5 Canon Tables 2.0 The Peshitta Version of the Eusebian Apparatus

With this chapter we shift our focus back to the east, to consider the distinctive way in which the Canon Tables system was passed down in Peshitta gospel manuscripts. Chronologically speaking, this chapter picks up only shortly after the last one left off, since the editorial work we are to consider most likely took place in the fifth century, though evidence for it survives only in manuscripts dating from the sixth century and onwards.¹ The historical sources that are examined in this chapter are also of a different sort from those considered previously. Whereas the last chapter was a close reading of an exegetical treatise that relied upon the paratextual apparatus, the focus in this chapter is on the apparatus itself as it appears in Syriac, specifically Peshitta, gospelbooks. The reason for this is that in the Syriac tradition there emerged an entirely unique version of the Canon Tables that enhanced the potential in Eusebius’ original system to make an even more powerful cross-referencing paratext. We might call this version 2.0 of Eusebius’ technology for ordering the textual knowledge found in the fourfold gospel. This revised system has

¹ An excellent introduction to Syriac manuscripts in this period can be found in Marlia Mundell Mango, ‘The Production of Syriac Manuscripts, 400 700 AD’, in Scritture, libri e testi nelle aree provinciali di Bisanzio, ed. Guglielmo Cavallo, Giuseppe De Gregorio, and Marilena Maniaci (Spoleta: Centro italiano di studi sull’alto Medioevo, 1991), who observed that there are around 300 manuscripts that survive from the years between 400 and 640, in contrast to the roughly 100 contemporaneous Greek manuscripts that are extant. She counted nine gospel codices from this period that preserve Canon Tabes (p. 164). For a provisional list of manuscripts containing the New Testament in Syriac, see David G. K. Taylor, ‘Répertoire des manuscrits syriaques du Nouveau Testament’, in Le Nouveau Testament en syriaque, ed. Jean Claude Haelewyck, Études syriaques 14 (Paris: Geuthner, 2017), pp. 291 313, who helpfully noted which ones include Canon Tables. A list of digitized Syriac manuscripts can be found at http://syri.ac/digimss. For a collection of recent papers on various aspects of Syriac manuscripts, see François Briquel Chatonnet and Muriel Debié, eds, Manuscripta syriaca: des sources de première main, Cahiers d’études syriaques 4 (Paris: Geuthner, 2015). See also the discussion of several famous manuscripts from this period in Sebastian Brock, The Bible in the Syriac Tradition, 2nd rev. edn (Piscataway, NJ: Gorgias Press, 2006), pp. 122 5.

The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0006

Canon Tables 2.0: The Peshitta Version

157

been known about since the nineteenth century,² but very little work has been done examining the differences between it and Eusebius’ original.³ The purpose of this chapter is to demonstrate the way in which this revision accomplished a fine-tuning of Eusebius’ invention, allowing for a closer alignment between text and paratext and a corresponding enhancement in the precision of the system as a reference and citation tool. To set the stage, a brief survey of the history of the gospels in Syriac is necessary.⁴ The earliest version of the gospel in Syriac was probably the socalled Diatessaron of Tatian, based on the fact that it appears to have been the most common gospel text in use in our earliest sources such as Aphrahat and Ephrem, and based on evidence that it influenced later Syriac translations of the gospels.⁵ However, by at least the fourth century there also existed a translation of the four gospels that was known as ‘The Gospel of the Separated Ones’ ( ) to distinguish it from Tatian’s version, which went ).⁶ This under the title ‘The Gospel of the Mixed Ones’ ( translation survives only in three incomplete manuscripts dated to the fourth, fifth, and sixth centuries, though it may have originated as early as the third.⁷ ² G. H. Gwilliam, ‘The Ammonian Sections, Eusebian Canons and the Harmonizing Tables in the Syriac Tetraevangelium’, Studia Biblica et Ecclesiastica 2 (1890): pp. 241 72. ³ The only exceptions are Brock, The Bible in the Syriac Tradition, pp. 120 1, and P. A. Vaccari, ‘Le sezioni evangeliche di Eusebio e il Diatessaron di Taziono nella letteratura siriaca’, Rivista degli Studi Orientali 32 (1957): pp. 433 52. ⁴ For a fuller account of what follows, see Brock, The Bible in the Syriac Tradition, 31 8; Peter J. Williams, ‘The Syriac Versions of the New Testament’, in The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, 2nd edn, ed. Bart D. Ehrman and Michael W. Holmes, New Testament Tools, Studies and Documents 42 (Leiden: Brill, 2013), pp. 143 66. On the wider cultural context in which translations from Greek into Syriac took place in this period, see especially Scott F. Johnson, ‘Introduction: The Social Presence of Greek in Eastern Christianity, 200 1200 CE’, in Languages and Cultures of Eastern Christianity: Greek, ed. Scott F. Johnson, The Worlds of Eastern Christianity, 300 1500, vol. 6 (Farnham: Ashgate, 2015), and Fergus Millar, Religion, Language and Community in the Roman Near East: Constantine to Muhammad, The Schweich Lectures of the British Academy (Oxford: Oxford University Press, 2013). ⁵ For an introduction to the complex study of Tatian’s version of the gospel, see Ulrich Schmid, ‘The Diatessaron of Tatian’, in The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, 2nd edn, ed. Bart D. Ehrman and Michael W. Holmes, New Testament Tools, Studies and Documents 42 (Leiden: Brill, 2013), pp. 115 42, and for a survey of scholarship up to 1994, see Petersen, Tatian’s Diatessaron. Note that Schmid drew attention to important shifts in the field that have occurred since Petersen’s book. For a collection of new studies on the topic, see Matthew R. Crawford and Nicholas J. Zola, eds, The Gospel of Tatian: Exploring the Nature and Text of the Diatessaron, The Reception of Jesus in the First Three Centuries (London: T&T Clark, 2019). ⁶ On the origins of these two titles, see Crawford, ‘Diatessaron, a Misnomer’, pp. 373 7. ⁷ One of the three manuscripts containing the Old Syriac version of the gospels has been discovered only very recently, as a result of the investigation into the palimpsests at St Catherine’s Monastery. For an introduction to the Vetus Syra of the gospels, including an overview of all three surviving manuscripts, see Jean Claude Haelewyck, ‘Les vieilles versions syriaques des Évangiles’, in Le Nouveau Testament en syriaque, ed. Jean Claude Haelewyck, Études syriaques 14 (Paris: Geuthner, 2017), pp. 67 113. The title occurs in

158

The Eusebian Canon Tables

Although Ephrem (d. 373) worked primarily with Tatian’s version, and authored a commentary on it,⁸ he also clearly was aware of the separate gospels as well, demonstrating that both versions must have been in circulation in his day.⁹ Just as the Old Latin was superseded in the West by Jerome’s Vulgate in the late fourth century, so also this Old Syriac translation of the gospels was surpassed by a revised version known as the Peshitta, which was produced some time in the late fourth or early fifth century and became the most commonly used version among all branches of Syriac-speaking Christianity. It was once thought that Rabbula, bishop of Edessa from 411/12 until 435/36 CE, was responsible for this translation, but this idea has since been abandoned in light of the fact that citations of the Peshitta seem to appear earlier than Rabbula.¹⁰ As a result, it is impossible to know what person or persons undertook this translation work, which represents ‘a revision of the Old Syriac on the basis of the Greek’¹¹ rather than an entirely new translation. This is a most unfortunate situation, for the revised version of the Eusebian apparatus that is the focus of this chapter appears for the first time in Peshitta gospel manuscripts, and the consistency with which it occurs in them suggests that the updating of the Greek paratext was undertaken in connection with the new translation, just as Canon Tables entered the Latin world around the same time via Jerome’s Vulgate. In the early sixth century, a further revision of the New Testament was sponsored by Philoxenus of Mabbug (d. 523), though it survives only in quotations and not via direct manuscript evidence. Finally, in the early seventh century, specifically 615/616 CE, yet another revision was produced by Thomas of Harkel, who had been bishop of Mabbug before moving to a monastery in Egypt. The so-called Harklean version ‘marks the zenith of literalism in Syriac representation of Greek’,¹² and the surviving manuscripts of it demonstrate an inconsistency in their presentation of the Eusebian paratext. Some, (‘perhaps two of them, in a colophon in the Sinaitic manuscript and as as a title at the beginning of the Gospel of Matthew in the Curetonian. See F. C. Burkitt, Evangelion da Mepharreshe, 2 vols (Cambridge: The University Press, 1904), vol. 2, pp. 31 3. It also is used in literary sources, such as in can. 43, attributed to Rabbula of Edessa (Vööbus, p. 11). Cf. Petersen, Tatian’s Diatessaron, p. 43. ⁸ For an attempt to recover portions of the Diatessaron from Ephrem’s corpus, see Louis Leloir, Le témoignage d’Éphrem sur le Diatessaron, CSCO 227, Subsidia 19 (Louvain: Secrétariat du CorpusSCO, 1962). ⁹ Matthew R. Crawford, ‘The Fourfold Gospel in the Writings of Ephrem the Syrian’, Hugoye 18 (2015): pp. 9 51. ¹⁰ The idea that Rabbula was responsible for the Peshitta was championed by Burkitt, Evangelion da Mepharreshe, vol. 2, pp. 161 2, and challenged by Arthur Vööbus, Studies in the History of the Gospel Text in Syriac II: New Contributions to the Sources Elucidating the History of the Traditions, CSCO 496, Subsidia 79 (Louvain: E. Peeters, 1987), pp. 4 8. See also Matthew Black, ‘Rabbula of Edessa and the Peshitta’, Bulletin of the John Rylands University Library of Manchester 33 (1951): pp. 203 10. ¹¹ Williams, ‘The Syriac Versions’, p. 151. ¹² Williams, ‘The Syriac Versions’, p. 154.

Canon Tables 2.0: The Peshitta Version

159

the majority’ according to Sebastian Brock¹³) revert back to the original Greek version of Eusebius’ system, though others preserve the revised paratext as it is found in Peshitta manuscripts. In the sole published edition of the Harklean Eusebian apparatus, Samer Soreshow Yohanna makes no mention of the alternate Peshitta version, and comments that the manuscripts used for his edition tend to closely adhere to the version printed in NA²⁸.¹⁴ Presumably, then, Thomas of Harkel, in keeping with his goal of representing the Greek text of the gospels as precisely as possible, reverted back to the original Greek version of Eusebius’ system. The Harklean translation of the gospels never won as wide acceptance as the Peshitta, with the result that the revised version of the Canon Tables remained the more common one in use among Syriacspeaking Christians. The aim of the present chapter is to provide the fullest analysis to date of this Peshitta revision and its relation to the Eusebian original. Three aspects of it must be considered: first, the Peshitta version of Eusebius’ prefatory Letter to Carpianus, which departs in some respects from the Greek original; second, the tabular concordance that typically appears in the lower margin of the page in Peshitta gospelbooks; and third, the additional sections in each gospel that occur in the Peshitta and the corresponding new parallels created from them. By far the most significant departure of the Peshitta version of the Canon Tables from the Greek original, and that which justifies calling it Canon Tables 2.0, is the last of these elements, though I suggest all three features were a part of a single project to overhaul and enhance Eusebius’ edition so that the newest translation of the gospels in Syriac would be equipped with an improved paratextual apparatus.

THE S YRIAC VERSION OF EUSEBIUS’ LETTER TO CARPIANUS The first thing to stress about the Peshitta version of Eusebius’ Letter to Carpianus (see fig. 25) is that it is, on the whole, an accurate translation, and what modifications are present cannot be explained as arising from a misunderstanding of how the marginal apparatus works. It is important

¹³ Brock, The Bible in the Syriac Tradition, pp. 119 20. ¹⁴ Samer Soreshow Yohanna, The Gospel of Mark in the Syriac Harklean Version: An Edition Based Upon the Earliest Witnesses, BO 52 (Rome: Gregorian & Biblical Press, 2015), p. 72.

160

The Eusebian Canon Tables

Fig. 25. First half of the Letter to Carpianus in the Rabbula Gospels (sixth c.). Florence, Biblioteca Medicea Laurenziana, MS. Plut. 1.56, fol. 2v

to emphasize this point in view of the fact that it was not always the case, as can be seen, for example, with the Ge‘ez version of the letter, which departs much more from the original.¹⁵ In contrast, the Syriac mostly follows ¹⁵ Cf. Alessandro Bausi, ‘La versione etiopica della Epistola di Eusebio a Carpiano’, in Aethiopia fortitudo ejus. Studi in onore di Monsignor Osvaldo Raineri in occasione del suo 80°

Canon Tables 2.0: The Peshitta Version

161

closely, often word for word, the Greek original. In what seems to be the only comment on this question in the secondary literature, G. H. Gwilliam observed that ‘the earlier part of [the letter] is a fair rendering of the original, but the latter part has become a paraphrase in the attempt to make the somewhat obscure Greek intelligible’.¹⁶ Gwilliam’s assessment is generally speaking true, but it is also the case that in the additions the Syriac translator made to Eusebius’ original we can see his own understanding of the Eusebian paratext, which represents a further stage of development in the reception of the apparatus. The changes made by the translator have two primary effects, the first related to Ammonius’ work and its relation to Eusebius’, and the second related to a shifting of emphasis from the ten tables to the marginal enumeration within each gospel.¹⁷ First, whereas Eusebius used a single verb to describe Ammonius’ modus operandi, asserting that he ‘placed’ passages from Matthew ‘alongside’ (παραθείς) those from the other gospels,¹⁸ the Syriac translator expands on the original by using four verbs. In this version Ammonius ‘closely attended to the Gospel of Matthew, and he collated the rest of the sections of the three evangelists that were like it [i.e., Matthew]. Those that agreed with one another he cut into pieces and placed them’ ( ).¹⁹ The translator has here provided a much fuller description of the process Ammonius went through to form his edition, making explicit what was implied by Eusebius’ succinct account. The verb is used elsewhere to refer to the scribal collation of multiple manuscripts of the same text,²⁰ and this comparative sense is surely what is in view here—Ammonius sought and collated those passages of the other three gospels that were similar to passages in Matthew and placed them alongside one another. The Syriac text also brings out much more explicitly the unintended, even violent, consequences of his actions. The verb (‘to cut’) is not only cognate with the word here used for ‘section’ ( ; itself a rough equivalent to the Greek περικοπή used by Eusebius), but also ties in with the description of

compleanno, ed. Rafał Zarzeczny, Orientalia Christiana Analecta 298 (Rome: Pontificio Istituto Orientale, 2015), pp. 107 35; McKenzie and Watson, The Garima Gospels, pp. 221 7. ¹⁶ Gwilliam, ‘The Ammonian Sections’, p. 258. ¹⁷ All of the variations between the Greek and the Peshitta versions of the letter that are considered below are absent from the Harklean version, which scrupulously adheres to the Greek original. See Yohanna, The Gospel of Mark in the Syriac Harklean Version, pp. 69 72. ¹⁸ ep. Carp. (NA,²⁸ p. 89*). ¹⁹ ep. Carp. (Pusey and Gwilliam, p. 2). In my translation I have chosen not to follow the punctuation in the Syriac text in one instance. The stop after the word seems to me misplaced, since it reads better if the verb is included with the preceding clause beginning with ²⁰ Cf. the usage of the verb in the colophons printed in W. Wright, Catalogue of the Syriac Manuscripts in the British Museum Acquired Since the Year 1838, 3 vols (London: British Museum, 1870 2), vol. 1, p. 12, col. a, ln. 12; vol. 1, p. 15, col. a, ln. 2.

162

The Eusebian Canon Tables

the text as a ‘body’ ( ; cf. Eusebius’ usage of σῶμα) that occurs in the lines that follow. The translator also made a change that emphasized the relation of Ammonius’ work to that of Eusebius. Whereas Eusebius had said that ‘in order that you might be able to know the particular passages in each evangelist where they were moved by love for the truth to speak about the same things, I took my starting points from the labour of the aforementioned man’ (εἰδέναι ἔχοις τοὺς οἰκείους ἑκάστου εὐαγγελιστοῦ τόπους, ἐν οἷς κατὰ τῶν αὐτῶν ἠνέχθησαν φιλαλήθως εἰπεῖν, ἐκ τοῦ πονήματος τοῦ προειρημένου ἀνδρὸς εἰληφὼς ἀφορμὰς),²¹ the Syriac translator has rendered the phrase φιλαλήθως εἰπεῖν as a statement from Eusebius himself: ‘As a lover of the truth, I say that we took for ourselves the occasion from the labour of that man [i.e., Ammonius] about whom we spoke previously’ ( .).²² Gwilliam pointed out that the translator apparently punctuated the Greek with a full stop after ἠνέχθησαν, beginning a new sentence with the φιλαλήθως εἰπεῖν that follows.²³ This is true, but it is also the case that the translator in fact omitted translating ἠνέχθησαν altogether. In Eusebius’ original, the words beginning with εἰδέναι ἔχοις in the above quotation were a continuation of a ἵνα clause begun earlier describing the purpose of his work, and with the words ἐκ τοῦ πονήματος Eusebius began the main clause describing his own actions to produce the Canon Tables apparatus. The Syriac translator, however, has provided a new main clause on which the ἵνα clause depends (on which more shortly), with the result that φιλαλήθως εἰπεῖν must begin a new sentence. Without the ἠνέχθησαν, the infinitive εἰπεῖν stands alone at the head of a new sentence, an odd construction in Greek, to say the least. Perhaps the translator was working from a defective manuscript that read εἶπον (‘I say’) in place of εἰπεῖν (‘to say’). Alternatively, given that just prior to this sentence the translator departed from his source text by inserting a lengthy new clause without any parallel in Eusebius’ letter, it may be that when he returned to his exemplar this seemed the best way to pick back up following the text of his source. Whatever the exact cause, or whether or not the change was deliberate, the effect of this alteration is to grant greater prominence to the relationship between the works of Ammonius and of Eusebius. Eusebius’ statement in this new version takes on an oath-like quality, defending the legitimacy of his task of composition by appealing to his motivation as a ‘lover of the truth’. The word that the translator used for Eusebius’ ἀφορμάς is , which has much the same semantic range, though he has used the singular in place of Eusebius’ ²¹ ep. Carp. (NA²⁸, p. 89*). ²² ep. Carp. (Pusey and Gwilliam, p. 2). ²³ The Ge‘ez version of the letter similarly disconnects the φιλαλήθως εἰπεῖν from the ἠνέχθησαν, though there is no clear relation between the Peshitta and Ge‘ez recension. See Bausi, ‘La versione etiopica’, p. 125.

Canon Tables 2.0: The Peshitta Version

163

plural, which, as I argued in chapter 2, is crucial for interpreting its meaning.²⁴ Nevertheless, as in the Greek version, it is Ammonius and Eusebius that stand together as collaborators in this version of the letter, though in the Syriac edition the translator has been more explicit about Ammonius’ method and has enhanced Eusebius’ declaration of his debt to his predecessor.²⁵ The Syriac translator similarly inserted a new sentence in the first half of the letter that referred to the enumeration within each gospel, something that

²⁴ Given the absence of vowel pointing, the singular and plural forms of are identical, though if it were plural it would have double seyame dots. Their absence clarifies that the form intended is singular. ²⁵ Traditions about Ammonius were passed on in other Syriac texts, though usually, it seems, with some confusion. One of the earliest of these comes from a short work entitled ‘History of Monasticism’, a part of a collection of texts related to the Council of Nicaea which are attributed to the late fourth and early fifth century bishop Marutha of Maypherqat. The attribution to Marutha is regarded as questionable, though this is immaterial for the present purpose. Rather surprisingly, this ‘History of Monasticism’ (CSCO 439, pp. 13 14) gives a passing reference to Ammonius and to the Canon Tables, though it is difficult to know what to make of it. In narrating the origins of monasticism in the Old Covenant, the author first refers to ‘letters’ written by Philo and sent to James, brother of Jesus. The next sentence then asserts that the same thing can be learned from the letter written by ‘Magna the Alexandrian’ ( ) to ‘Serapion the head of the anchorites’ ( ). The author then explains that ‘this Magna is that one who first ordered [and] set down the Chronicon in the church. Also Eusebius the Caesarean remembers him at the beginning of the canons that he made for the four evangelists, [and] he remembers him also in his other writings that he made’ ( .). Given the association with Eusebius and his ‘Canons…for the evangelists’, there can be no doubt that the ‘Magna’ of this text is the same as the Ammonius mentioned in the Letter to Carpianus. But if so, why is he called ‘Magna’? The text provides no clues but only further questions, since the supposed letter written by this ‘Magna’ to Serapion (bishop of Antioch?) is otherwise unknown, as is the Chronicon that he is said to have composed. Given the other obscure references in this text, such as the aforementioned letter from Philo to James and a letter from Clement of Rome to Dionysius the Areopagite, it is certainly unwise to look here for reliable historical traditions about the works of Ammonius and Eusebius. On Marutha, see Sebastian P. Brock, ‘Marutha of Maypherqat’, in The Gorgias Encyclopedic Dictionary of the Syriac Heritage (Piscataway, NJ: Gorgias Press, 2011), p. 273, and further literature cited there. An earlier version of this text was published in Oscar Braun, De sancta nicaena synod. Syrische Texte des Maruta von Maipherkat nach einer Handschrift der Propaganda zu Rom übersetzt (Münster: H. Schöningh, 1898), and was reviewed by Adolf von Harnack in Theologische Literaturzeitung 2 (1899): pp. 44 8, and later discussed in Bardenhewer, Geschichte der altkirchlichen Literatur. Zweiter Band, p. 202. Later Syriac authors also exhibit confusion between Ammonius and Tatian. Bar Hebraeus, writing in the thirteenth century, asserted that Ammonius had authored ‘the Gospel of the Diatessaron, that of the Mixed’ ( ), the latter being the traditional appellation in Syriac sources for Tatian’s composition (cited from the preface to Matthew in bar Hebraeus’ horr. (Assemani, pp. 57 8)). Slightly later, ‘Abdisho‘ bar Brikha went one step further and directly identified Tatian and Ammonius. In his Catalogue of Syriac Books he referred to ‘the Gospel, which a man from Alexandria, Ammonius that is Tatian assembled and called it Diatessaron’ ( ) (cat. 3 (Assemani, p. 12). This confusion no doubt arose from the fact that Eusebius referred to both Ammonius’ work and that of Tatian under the same title.

164

The Eusebian Canon Tables

Eusebius himself did not mention until the latter half of the letter. Indeed, the structure of the letter makes this the main focus of its first half: in order to preserve the whole body perfectly complete, along with the order of the four evangelists’ words, and in order that you might know also the passages of the words in which they [i.e., the words] agree with one another, there are numbers inscribed for you above each one of the evangelists in the passages that are appropriate.

.

a²⁶

The italicized words in the translation mark the additional material inserted by the translator, and, though they are in keeping with the intent of Eusebius’ original letter, they do subtly shift the emphasis, so that the focus is not primarily upon the ten prefatory tables, but at least equally, if not more, upon the marginal numbers within each gospel. This emphasis on the sectional enumeration comes out again even more clearly at the end of the letter, which represents the most significant change made by the Syriac translator. Once he had completed his rendering of Eusebius’ original, the translator added an entirely new paragraph that recapitulated much of the preceding material: Therefore, these numbers have been appointed in order that the words of the four evangelists might not be cut off one from after another, and so that the sequence of their arrangement might not be destroyed. Rather, only the numbers change in relation to one another, in order to show that the evangelists agree with one another, and in order that the reading of the arrangement of the words of those four might remain intact, that is, Matthew, Mark, Luke, John.

a²⁷

Since this paragraph was not actually a translation at all, but the translator’s expansion on his source text, we can observe here those aspects of Eusebius’ system that seemed to him most significant.²⁸ As he had already highlighted ²⁶ ep. Carp. (Pusey and Gwilliam, p. 2). ²⁷ ep. Carp. (Pusey and Gwilliam, p. 4). ²⁸ Caspar Renè Gregory pointed out that the Greek minuscules 77, 108, and 717 have an additional statement at the end of the usual Letter to Carpianus, which reads: Κανόνες δὲ προσηγορεύθησαν, διὰ τὸ εὐθὲς αὐτῶν καὶ ἀκριβές· ὥσπερ γὰρ ἰθυτενέσι τισὶ γραμμαῖς ταῖς ἰδίαις εὐθύτισι χρώμενοι, οὐκ ἐῶσι τὸν νοῦν περιπλανᾶσθαι τοῦ ἀκροατοῦ· ἀλλὰ πρὸς τὴν ὁμοφωνίαν ἄγουσιν αὐτὸν, τῶν ἐνεχθεισῶν φωνῶν τοῖς εὐαγγελισταῖς. Cf. Caspar Renè Gregory, Textkritik des Neuen Testamentes, 3 vols (Leipzig: J.C. Hinrichs, 1900 9), p. 864. After reproducing the Greek statement, Gregory cited the above Syriac paragraph as translated into French by J. P. P. Martin and implied that they are the same ‘Zusatz’. Although they both come at the same position, as additions at the end of the letter, they seem to me too different to be related to each other. Note, for example, that the Syriac addition focuses on the ‘numbers’, while the Greek addition begins by pointing to the ‘Canons’, nowhere mentioning ‘numbers’.

Canon Tables 2.0: The Peshitta Version

165

with his addition to the first paragraph, here again he points out the danger that the evangelists’ words might be ‘cut’, that violence might be done to the body of the text. He also again repeats the danger that their ‘sequence of arrangement’ ( ) might be corrupted, this phrase having already been used in the first half of the letter to translate Eusebius’ τὸν τῆς ἀκολουθίας εἱρμόν. However, the object of the ‘cutting’ appears subtly different here from the usage of the concept in the first part of the letter. Whereas earlier the translator made explicit that Ammonius’ method involved ‘cutting up’ bits and pieces of individual gospels to align them with Matthew, the threat in this last paragraph has in view the unity of the fourfold gospel as a whole. In other words, the translator wants his readers to know that the Canon Table apparatus is designed to demonstrate the unity of these four texts as a corpus, a ‘body’ (‘the evangelists agree with one another’), as well as the integrity of each individual text in the corpus (‘the reading of the arrangement’). Although the latter concern reflects Eusebius’ own description, nowhere in the original Letter to Carpianus did he claim to demonstrate that the four gospels ‘agree’ with one another and so should not be separated. Rather, he simply provided readers with a tool to find passages where the evangelists ‘have said similar things’.The Syriac reviser therefore claims more for the apparatus than Eusebius himself explicitly did, as though it vindicated the inclusion of these four texts in a canon. Moreover, in this final paragraph the translator explicitly points out the technological advance that has made possible this unique system. Because the evangelists’ words can be represented symbolically by numbers, ‘only the numbers change in relation to one another’. That is, in a gospelbook equipped with this apparatus, the numbers ‘change’ ( in the Pael can even mean ‘move from one place to another’) as one reads through each gospel, showing the relationship of each gospel with its companions from start to finish, and it is this numeric variability that ensures textual stability. Note that the ‘numbers’ in view here are the sectional enumeration within each gospel. In fact, the translator makes no mention in this paragraph of the Canons at all, revealing again that he thought the sectional enumeration was the most important element of the system.

MARGINAL CONCO RDANCE TABLES I N SYRIAC MANUSCRIPTS The heightened emphasis placed upon the marginal enumeration in the Peshitta Letter to Carpianus is no doubt related to another aspect of the presentation of the Eusebian apparatus in Syriac manuscripts. Peshitta gospelbooks consistently include in the bottom margin of each page a small table showing the section numbers for the passages from the other gospels that are parallel to whatever passages are present on the page that you are reading.

166

The Eusebian Canon Tables

Fig. 26. Mt §343 7 (= Mt 26:35ff.) in the Rabbula Gospels (sixth c.). Florence, Biblioteca Medicea Laurenziana, MS. Plut. 1.56, fol. 84r

Take, for example, fol. 84r of the Rabbula Gospels, which contains the Peshitta version of Matthew 26:35–40 (see fig. 26).²⁹ ²⁹ A digitized version of the manuscript is available at http://teca.bmlonline.it/ImageViewer/ servlet/ImageViewer?idr=TECA0000025956&keyworks=evangelia [accessed on 16 January 2018].

Canon Tables 2.0: The Peshitta Version

167

The right-hand column of text (which is read first), contains two Eusebian sections, numbered in red as and (that is, 343 and 344), and below the column is written in red the following table:

Matthew

Mark

John

343

207



344

208

128

In other words, Mt §343 is parallel to Mk §207 (the dots are placed in the column for ‘John’ because this is a Canon VI parallel, and so has no Johannine equivalent), and similarly Mt §344 is parallel to Mk §208 and Jn §128. Similarly, the left-hand column on this same page contains Matthew sections , and (that is, 345, 346, and 347), and beneath the column of text is written the following information:

Matthew

Mark

Luke

John

345

209

321



346

210

322

66

347

211





Once again, the chart in the bottom margin informs the reader that Mt §345 is parallel to Mk §209 and Lk §321, and so on for Mt §346 and §347. This is of course the same information that one could find by turning to the tables at the

168

The Eusebian Canon Tables

beginning of the codex,³⁰ and in fact these tabular cross-references render the ten prefatory tables somewhat redundant since the reader can now turn directly from one gospel to another without needing to consult the Canons.³¹ The diminished need for the prefatory Canons corresponds to the fact that the Peshitta Letter to Carpianus downplays the Canons and highlights instead the marginal enumeration within each gospel. Because it simplifies the steps required to access the parallels created by Eusebius—making the intertextual itineraries between gospels more direct by not requiring that one be routed through the Canon Tables themselves—this additional marginal annotation enhances the usability of the Eusebian paratext.³² Given how early and consistently this feature appears in the manuscript tradition, the marginal tables throughout the gospels were probably a part of the system in Syriac manuscripts from the outset. Was it, however, truly an ‘innovation’ of the Syriac translator, as has been assumed?³³ In fact, a similar development appears in a number of contemporary Latin manuscripts, such as the fifth-century Codex Sangallensis 1395 (see fig. 27).³⁴ In these manuscripts an enhanced version of marginal annotation also appears on each page, though in the Latin tradition the section numbers for the parallel passages in the other gospels are distributed across the page at various places in the margin beneath each of the corresponding section numbers, rather than being collected together in a tabular format, as they are in Syriac codices. Given the likely dating of the Peshitta translation to the late fourth or early fifth century, this is therefore a feature that appeared roughly simultaneously in Latin and in Syriac. Should we then view these as parallel developments that occurred independently of one another or is it possible that they are in some way related? An influence from a Syriac source

³⁰ In one respect the information conveyed in these marginal tables is less than what is found in the ten Canons at the start of the codex. Eusebius sometimes paralleled a single passage from one gospel with multiple passages from other gospels, and, for such instances, the marginal tables only include a single parallel. In the above example, Mt §346 is actually parallel to both Jn §51 and Jn §66, and both parallels are recorded in the first Canon on fol. 4r of the Rabbula Gospels, though only one of the two is listed in the marginal table. ³¹ The ‘somewhat’ here is an important hedge. Commenting on a similar development in the Latin tradition, O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, p. 99, has argued that this additional marginal notation makes the actual Canon Tables ‘redundant’. If one is only interested in finding parallel passages as one is reading a given gospel, this is true. However, the ten prefatory tables still perform an important function of collecting together and classifying passages according to which gospels they appear in, which is not accomplished by the marginal tables. So, for example, if one wanted to study all the passages that appear only in Matthew and Luke, it would be difficult to do so if you only had the marginal tables and not also the ten Canons. ³² Cf. Marchese, ‘Tables and Early Information Visualization’, p. 45: ‘The effect of these cross linkages was to transform a simple linear connection between the indices contained in a canon table and Gospel passages into a nonlinearly interconnected hypertext system’. ³³ Cf. Brock, The Bible in the Syriac Tradition, p. 119. ³⁴ On Codex Sangallensis 1395, see Houghton, The Latin New Testament, pp. 259 60.

Canon Tables 2.0: The Peshitta Version

169

Fig. 27. Mt §354 5 in Codex Sangallensis 1395 (first half of fifth c.). Codex Sangal lensis 1395 is the oldest copy of the Vulgate gospels and was possibly copied within Jerome’s own lifetime. Here we see two sections from the Gospel of Matthew: Mt §354 and §355. Mt §354 is a Canon II passage, parallel to Mk §233 and Lk §338, which are both listed in the margin beneath the red ‘II’. However, Mt §355 is a passage unique to Matthew (note the red ‘X’ for the Canon number), so no parallel passages are listed for it. St Gall, Stiftsbibliothek 1395, p. 132

directly upon a Latin one or vice versa appears unlikely in the period in question, though not impossible. Such a supposition is, however, unnecessary, since there is also evidence of the enhanced marginal annotation in the Greek manuscript tradition, though it has hitherto been largely overlooked. Sebastian Brock, for example, mentioned in passing that a few Greek manuscripts present this same system, but asserted that they are all much later than the earliest Syriac gospel codices.³⁵ For the most part he was correct, in that the handful of relevant Greek manuscripts are mostly later, such as the ninth-century Codex Basilensis A.N. III.12, known as manuscript Ee in the numbering system used by the INTF (see fig. 28).³⁶ The New Testament textual critic Casper René Gregory listed this manuscript, along with seven other Greek ones, which all included such tables in the bottom margin of the page, most of them dated between the ninth and twelfth centuries.³⁷ However, it has recently been pointed out that there is a fragmentary Greek manuscript dated to the sixth century that also preserves evidence of the same kind of marginal table that appears in contemporary ³⁵ Brock, The Bible in the Syriac Tradition, p. 119. ³⁶ On the dating of this manuscript, see Annaclara Cataldi Palau, ‘A Little Known Manuscript of the Gospels in “Maiuscola Biblica”: Basil. Gr. A. N. III. 12’, Byzantion 74 (2004): pp. 463 516. ³⁷ Cf. Gregory, Textkritik des Neuen Testamentes, p. 862. The manuscripts listed by Gregory are E, M, Tb, Wd, 199, 262, 264, 655, and ‘einige andere Kleinschrift Handschriften’. In the enumeration used today by the INTF, these are 07, 021, 083, 0131, 199, 262, 264, and 655.

Fig. 28. Jn §55 8 (Greek numerals ΝΕ, ΝϚ, ΝΖ, ΝΗ) in Codex Basilensis A.N.III.12 (ninth c.). The table in the bottom margin shows the parallel passages from the other three gospels, in the sequence John Luke Mark Matthew. The parallel Jn §55, Lk §266, Mk §165, Mt §284 (Jn §ΝΕ, Lk §ϹΞϚ, Mk §ΡΞΕ, Mt §ϹΠΔ = Jn 6:35a, Lk 22:19, Mk 14:22, Mt 26:26) is discussed in chapter 3. Here Eusebius has singled out the sentence εἶπεν δὲ αὐτοῖς ὁ Ἰ(ησοῦ)ς ἐγώ εἰμι ὁ ἄρτος τῆς ζωῆς (beginning on line 6) and joined it to Jesus’ statements from the Last Supper accounts in the synoptics. Note that the scribe here has made a mistake, writing ϹΞ (= 260) rather than ϹΞϚ (= 266) in the first row of the column for Luke. The correct number is, however, given on fol. 236r where the Lukan passage occurs. Basel, Universitätsbibliothek, A.N.III.12, fol. 269r

Canon Tables 2.0: The Peshitta Version

171

Peshitta manuscripts.³⁸ In addition to the Syriac, Latin, and Greek manuscripts just noted, mention should also be made of Codex Argenteus, a sixthcentury manuscript containing the Gothic version of the gospels in gold and silver ink on purple parchment. It too includes in the bottom margin of each page the same tabular concordance, the only difference being that here the tables are housed within mini-arcades, an additional decorative flourish no doubt inspired by the architectural motifs that housed the now lost Canons at the start of the codex (see fig. 9 in the introduction). It is therefore the case that by the fifth and sixth centuries, there are examples of gospelbooks in Greek, Latin, Gothic, and Syriac that contain the enhanced marginal annotation, so this cannot be viewed as a distinctive feature of the Syriac tradition.³⁹ Peshitta manuscripts are remarkable only for the fact that this became a regular feature of their mis-en-page, whereas it was never universal in the Latin tradition and never seems to have been common in Greek codices. That it appears so consistently in the Syriac tradition is probably due to the fact that, as already suggested, it was an original part of the translation of Eusbius’ apparatus into the language and continued to be copied thereafter as a standard feature, whereas in the Greek and Latin manuscript traditions it was a subsequent development. With the evidence we have at hand, it is neither possible nor necessary to determine in which of these languages the additional marginal annotation first appeared. Greek might seem like the most likely option, with the new concept then radiating out to both the Syriac and the Latin worlds, though it is also possible that this was an innovation that first occurred in a Latin manuscript, and that it then was transmitted back into the Greek tradition, and from there into Syriac.

THE PESHITTA REVISION OF EUSEBIUS ’ SECTIONING AND P ARALLELS In the last section it might have seemed that I was detracting from the uniqueness of the Syriac tradition by placing it in a wider context, demonstrating that the tabular concordances usually found in them are not a unique feature isolated to Syriac codices. However, there is one other respect in which the version of the paratextual system that appears in Peshitta gospelbooks is completely without parallel. All Peshitta codices that have Canon Tables carry a revised version of Eusebius’ sectioning that encodes a more precise citation system, which allows for an even greater degree of comparative analysis of ³⁸ Cf. Matthew R. Crawford, ‘A New Witness to the “Western” Ordering of the Gospels: GA 073+084’, Journal of Theological Studies 69 (2018): pp. 477 83. ³⁹ Similar marginal annotation is found in later Armenian and Caucasian Albanian manuscripts. On the latter, see Gippert et al., The Caucasian Albanian Palimpsests of Mt. Sinai, vol. 1, pp. 32 3.

172

The Eusebian Canon Tables

the similarities and differences amongst the gospels. This revision was accomplished by further subdividing the sections within each gospel. The chart below demonstrates the differences in the original and revised versions: No. of Sections in Eusebius’ Original

No. of Sections in the Peshitta Version

Percentage Change

Matthew

355

426

20%

Mark

233

290

24%

Luke

342

402

18%

232

271

17%

1,162

1,386

19%

John TOTAL:

As the above numbers indicate, the person responsible for this revised version was able to increase the number of sections in each gospel by around a fifth. With a larger number of sections in each gospel, there is greater scope for making parallels between gospels. Accordingly, we find a greater number of parallels listed in most of the ten Canons: No. of Parallels in Eusebius’ Original

No. of Parallels in the Peshitta Version

Percentage Change

Canon I

74

89

20%

Canon II

111

118

6%

Canon III

22

20

9%

Canon IV

25

24

4%

Canon V

82

89

9%

Canon VI

47

57

21%

Canon VII

7

18

157%

Canon VIII

13

23

77%

Canon IX

21

19

10%

Canon XMt

62

76

23%

Canon XMk

19

26

37%

Canon X

Lk

72

86

19%

Canon X

Jo

96

103

7%

651

748

15%

TOTAL:

The first thing that is immediately apparent from the above figures is the uneven distribution of the new parallels created by the more precise sectioning. Quite unexpectedly, three Canons actually present fewer parallels in this new system

Canon Tables 2.0: The Peshitta Version

173

than in Eusebius’ original: Canons III (Mt–Lk–Jn), IV (Mt–Mk–Jn), and IX (Lk– Jn). Moreover, other Canons drastically increased, such as Canons VII (Mt–Jn) and VIII (Mk–Lk), which have 157% and 77% more parallels in the new version, though this high percentage of increase is partially due to the small size of these lists to begin with. It is also striking that the reviser has identified 37% more unique material in Mark than Eusebius had done, a figure no doubt related to the fact that the number of sections in Mark also increased proportionally more than is the case for the other gospels (24%). Analysing examples of the changes made in each of the Canons will give us a better picture of the revision the Syriac scribe has made to Eusebius’ original. I have tried as far as possible to choose examples drawn from a range of types of passages, including narrative and discourse. Note that in what follows I will use a superscript ‘E’ and ‘S’ to clarify whether the sections I am citing belong to Eusebius’ original or to the Syriac revision.

Canon I All four gospels preserve a statement from John the Baptist that draws a contrast between his own status and ministry and that of the ‘one coming after’ him. Eusebius accordingly placed all four of these passages in Canon I as parallel to one another:⁴⁰ Mt §11E

Mk §4E

Lk §10E

Mt 3:11 I baptize you with water for re pentance, but one who is more power ful than I is coming after me; I am not worthy to carry his sandals. He will baptize you with the Holy Spirit and fire.

Mk 1:7b The one who is more powerful than I is coming after me; I am not worthy to stoop down and untie the thong of his sandals. Mk 1:8 I have bap tized you with water; but he will baptize you with the Holy Spirit.

Lk 3:16b e I baptize you with water; but one who is more powerful than I is coming; I am not worthy to untie the thong of his san dals. He will bap tize you with the Holy Spirit and fire.

Jn §12E Jn 1:26

John an swered them, ‘I baptize with water. Among you stands one whom you do not know, Jn 1:27 the one who is coming after me; I am not worthy to untie the thong of his sandal.

⁴⁰ Eusebius actually repeated this parallel three more times, using the same passages from Mt, Mk, and Lk, but three unique passages from Jn: Jn §6E, 14E, 28E (Jn 1:15, 30 1, 3:28). I have focused on the one relevant for the current discussion. The Syriac keeps these sections, as Jn §6S, 15S, 33S. For the English translations of the Greek New Testament included in this chapter I have followed the New Revised Standard Version.

174

The Eusebian Canon Tables

The Syriac version retains this parallel but subdivides it into two Canon I parallels. Thus we have the following:⁴¹ Mt §12S Mt 3:11a c

I baptize you in water for repentance. But he who comes after me is mightier than I, he whose sandals I am unworthy to carry.

Mt §13S Mt 3:11d

He will baptize you with the Holy Spirit and with fire.

Mk §6S Mk 1:7

Lk §10S

And he would preach and say, ‘Be hold, after me comes one who is more powerful than I, the straps of whose san dals I am not worthy to bend down and untie. Mk 1:8a I have baptized you in water,

Lk 3:16a d

John re sponded to them, ‘I baptize you in water, but he who comes is more powerful than I, the straps of whose sandals I am not worthy to untie.

Jn §12S Jn 1:26

John answered them, ‘I baptize with water, but among you stands one whom you do not know. Jn 1:27 This is the one who is coming after me, and he was ahead of me, he whose sandal straps I am not worthy to untie’.

Mk §7S

Lk §11S

Jn §17S

Mk 1:8b but he will baptize you with the Holy Spirit’.

Lk 3:16e He will baptize you with the Holy Spirit and with fire.

Jn 1:33 And I did not know him but he who sent me to baptize with water told me, ‘The one on whom you see the Spirit descend and rest ing, this is the one who baptizes with the Holy Spirit’.

In large measure, these sections are identical to Eusebius’ original. The only exception is that the final lines of the passages from the Synoptic sections have been broken off to form new sections, which now are joined to another new section from the Gospel of John to create a new Canon I parallel. Eusebius had included John 1:33 in a larger section comprising John 1:32–4 (Jn §15E), which formed a Canon I parallel about the descent of the Spirit at Jesus’ baptism. The Syriac reviser excerpted John 1:33 from this Eusebian section because he recognized that it presented a verbal and thematic similarity to a statement from the Synoptics about the sort of baptism that Jesus would bring to pass. This new section therefore highlights the distinctiveness of Jesus’ baptism, that it is accomplished in the Holy Spirit and in fire, though it also demonstrates that further subdivision would have been possible for a reviser who was even more scrupulous about what passes for similarity. While all four gospels state ⁴¹ Here and throughout this chapter I have used the new Antioch Bible translation of the Peshitta: Jeff W. Childers and George A. Kiraz, The Syriac Peshitṭ a Bible with English Translation: Matthew (Piscataway, NJ: Gorgias Press, 2012); Jeff W. Childers and George A. Kiraz, The Syriac Peshit ̣ta Bible with English Translation: Mark (Piscataway, NJ: Gorgias Press, 2012); Jeff W. Childers and George A. Kiraz, The Syriac Peshitṭ a Bible with English Translation: Luke (Piscataway, NJ: Gorgias Press, 2013); Jeff W. Childers, James Prather, and George A. Kiraz, The Syriac Peshitṭ a Bible with English Translation: John (Piscataway, NJ: Gorgias Press, 2014).

Canon Tables 2.0: The Peshitta Version

175

that Jesus will baptize ‘in the Holy Spirit’, only Matthew and Luke add the additional feature of baptism ‘in fire’. This dissimilarity could be taken as grounds for making the prepositional phrases ‘in fire’ in the Matthean and Lukan accounts into two new sections, forming a Canon V parallel, but the Syriac reviser chose not to do so. This example illustrates that, although the Peshitta version is more precise than Eusebius’ original, still further precision would have on occasion been possible, though the additional utility of the greater precision in this instance would have been negligible. It is pertinent at this point to recall the discussion from chapter 3 about the nature of Eusebius’ parallels. Placing two passages alongside each other in a Canon Table implies that they present some kind of similarity to each other, though it does not specify the degree or nature of that similarity. The Syriac reviser often seems to assume that a greater degree of correspondence is necessary for two passages to be set in parallel to each other, as in the above example wherein he highlights the verbal parallel in the statements from John the Baptist across all four gospels. At times, however, the Syriac revision expands Eusebius’ system, not by subdividing and producing greater verbal precision, but instead by adding more passages to a set of thematic parallels originally created by Eusebius. An example is the treatment of Peter’s confession of Christ recorded in the Synoptics. Eusebius had included as a Canon I parallel the following passages: Mt §166E, Mk §82E, Lk §94E, and Jn §17E, 74E (=Mt 16:13–16; Mk 8:27–9a; Lk 9:18–20; Jn 1:41–2, 6:68–9).⁴² In the former three passages Peter confesses Jesus as ‘Christ’, while in the first Johannine passage it is in fact Andrew, Peter’s brother, who makes this confession to Peter. We might then assume that the Peshitta version would split apart the Synoptic and Johannine passages, in view of the difference between them. In fact, however, the Peshitta lists even more passages: Mt §201S, Mk §103S, Lk §119S, Jn §20S, 23S, 39S, 82S, 107S (=Mt 16:16; Mk 8:29b; Lk 9:20b; Jn 1:41–2, 1:49–2:11, 4:42b, 6:68–9, 11:27b). Here the Syriac reviser has added three more Johannine parallels to the passages containing Peter and Andrew’s confession, and these recount the similar confessions made by Nathanael, the Samaritan villagers, and Martha. The Syriac reviser has apparently recognized the thematic similarity of Eusebius’ original set of passages and decided on this occasion to retain and even increase the number of voices confessing Jesus as the Messiah. In other words, strict verbal or historical correspondence was not the only criterion used in revising Eusebius’ system.

⁴² The edition of Eusebius’ apparatus included in NA²⁸ has Mk §83 begin with Mk 8:29b but this seems to be an error. Mk 8:29b is Peter’s confession and belongs with the preceding section, Mk §82. Mk 8:30 begins Jesus’ foretelling of his passion, which is parallel to the other Canon II passages included alongside Mk §83, so Mk §83 should begin with Mk 8:30 rather than Mk 8:29b.

176

The Eusebian Canon Tables

Canon II The story of the healing of a paralytic is a typical example of the kinds of changes made with respect to Canon II. Eusebius had treated these episodes as large chunks of text, presenting them as a Canon I parallel: Mt §70E

Mk §20E

Lk §37E

Jn §38E

Mt 9:1 8

Mk 2:1 12

Lk 5:18 26

Jn 5:1 10

The Syriac, however, subdivides the leading Matthean passage into four sections, alternating back and forth between Canon I and Canon II: Mt §87S

Mk §26S

Lk §46S

Jn §46S

Mt 9:2a

Mk 2:3 4

Lk 5:18 19

Jn 5:5 7

Mt §88S

Mk §27S

Lk §47S

Mt 9:2b 6a

Mk 2:5 10a

Lk 5:20 24a

Mt §89S

Mk §28S

Lk §48S

Jn §47S

Mt 9:6b 7

Mk 2:10b 12a

Lk 5:24b 25

Jn 5:8 9b

Mt §90S

Mk §29S

Lk §49S

Mt 9:8

Mk 2:12b

Lk 5:26

The other portions of Eusebius’ original Johannine section that are not used in the Canon I parallels above (Jn 5:1–4, 9c–10 = Jn §45S, 48S) are now relegated to Canon XJn as material unique to the fourth gospel. The primary goal of these redactional changes is to highlight only those portions of the Johannine episode that have a correspondence with the Synoptic accounts, all of which are otherwise very similar among themselves. The extent of the similarity among the four is simply the introduction of the paralysed man and the successful healing, so these two elements remain Canon I passages, as Eusebius had originally indicated. However, with the removal of the distinctly Johannine elements from John 5:1–10, the remaining portions of the Synoptic passages are shifted from Canon I to Canon II. These changes therefore result in the addition of one new Canon I passage, two Canon II passages, and two

Canon Tables 2.0: The Peshitta Version

177

new Canon XJn passages, illustrating how subdividing a single episode can quickly increase the overall tally of parallels listed in the Tables.

Canon III The last example demonstrated the Syriac reviser making new parallels by removing portions of text that were, in his estimation, not close enough to warrant inclusion alongside the other passages next to which Eusebius had originally placed them. The Syriac redactor also at times added new passages to existing Eusebian parallels, such as with Matthew 8:13, about the healing of a centurion’s servant. Eusebius had originally presented the conclusion to this episode as a Canon V parallel: Mt §66E

Lk §66E

And to the centurion Jesus said, ‘Go; let it be done for you according to your faith’. And the servant was healed in that hour. Mt 8:13

Lk 7:10

When those who had been sent re turned to the house, they found the slave in good health.

The Syriac preserves this parallel but adds to it a new section: Mt §81S Mt 8:13

Jesus said to the centurion, ‘Go. May it be for you as you have be lieved’. And his child was healed at that moment.

Lk §84S

Jn §44S

Then the people who had been sent returned to the house and found that the ser vant who had been sick was well.

‘Go’, Jesus told him, ‘your son is saved’. The man believed in the word Jesus told him, and went away. Jn 4:51 As he was going down, his servants met him and reported the good news, telling him, ‘Your son lives’. Jn 4:52 He asked them, ‘At what time did he get well?’ ‘The fever left him yesterday at the seventh hour’, they told him. Jn 4:53 The boy’s father realized it was in that hour that Jesus had told him, ‘Your son lives’, and he and his entire household believed. Jn 4:54 Now this was the second sign Jesus did after coming from Judea to Galilee

Lk 7:10

Jn 4:50

By adding a Johannine passage alongside the Matthean and Lukan ones, the parallel moves from Canon V to Canon III. The new Johannine passage had been a part of a larger Eusebian section, comprising John 4:46b–54 (=Jn §37E), so this is another example of a new parallel created by subdivision of a Eusebian section. Nevertheless, with his addition of the new Johannine section to the existing Canon V parallel, the Syriac redactor was still following Eusebius’ lead, since Eusebius himself had coordinated Jn 4:46b–54 as a whole with the earlier portions of the healing episodes in Matthew and Luke

178

The Eusebian Canon Tables

(Mt 8:5–10; Lk 7:1–9; Jn 4:46b–54 = Mt §64E; Lk §65E; Jn §37E). What the Syriac author has done, therefore, is to fine-tune Eusebius’ treatment of these passages by more correctly aligning the conclusion to the three episodes as a parallel, rather than having just one large Johannine section parallel to the first half of the Matthean and Lukan episodes.

Canon IV I pointed out above that the number of Canon IV parallels actually decreased as a result of the changes introduced by the Syriac redactor. We can see one example of how such a reduction occurred by examining the treatment of Matthew 12:14 and parallels. Eusebius had created the following Canon IV parallel: Mt §117E

Mk §26E

Jn §95E

Mt 12:14 But the Pharisees went out and conspired against him, how to des troy him.

Mk 3:6 The Pharisees went out and immediately conspired with the Herodians against him, how to destroy him. Mk 3:7a Jesus departed with his disciples to the sea,

Jn 11:53 So from that day on they planned to put him to death. Jn 11:54 Jesus therefore no longer walked about openly among the Jews, but went from there to a town called Ephraim in the region near the wilderness; and he remained there with the disciples.

The Syriac retains these passages as a parallel, but adds a Lukan episode: Mt §139S Mt 12:14

The Phar isees went out and conferred together about him regard ing how they might destroy him. Mt 12:15 Jesus was aware of it and left there.

Mk §36S Mk 3:6

At once the Pharisees and the Herodians went out and conferred to gether about him regarding how they might destroy him. Mk 3:7a Jesus went with his disciples to the lake,

Lk §55S Lk 6:11

But they were filled with resent ment and talked with each other about what to do to Jesus.

Jn §111S Jn 11:53

From that day they planned to kill him, Jn 11:54 so Jesus did not walk openly among the Jews, but went from there to a place near the wilderness, a village named Eph raim, where he spent time with his disciples.

In this instance, Luke 6:11 was originally, for Eusebius, included in a larger section consisting of Luke 6:6–11 (=Lk §42E), but the Syriac redactor has broken off this verse to create a new section, so that he can add it to an existing Canon IV parallel, thereby turning it into a Canon I parallel. Here again,

Canon Tables 2.0: The Peshitta Version

179

however, this is not really a new parallel, but rather a tidying up of Eusebius’ original version. The above verses in Matthew, Mark, and Luke represent the conclusion to the story of the healing of the man with a withered hand. Although all three accounts end with the Pharisees conspiring against Jesus, Eusebius took only the concluding statements from Matthew and Mark to make them a distinct parallel with the Johannine passage, presumably because all three speak of the desire of Jewish leaders to kill Jesus. Although the Lukan ending to the story is not as dramatic, with a vaguer report about the response of Jesus’ enemies, the Syriac reviser decided that it too should be included alongside the other three passages. The impact of this change is that Canon IV loses a parallel, while Canon I gains one.

Canon V The way in which the Syriac author reworked the accounts of Jesus’ temptations is a useful example of the kind of editorial work he undertook. Matthew and Luke uniquely record a series of three temptations, so Eusebius placed these passages in Canon V: Mt §16E

Lk §16E

Mt 4:2 10

Lk 4:2b 13

However, Matthew and Luke, despite recording the same three temptations, disagree on the order in which they occurred, a divergence that Eusebius’ parallel does not reflect. The Syriac redactor makes this divergence explicit by splitting the original two sections in Matthew and Luke into three sections in each gospel: Mt §20S

Lk §19S

Mt 4:2 4

Lk 4:2b 4

Mt §21S

Lk §21S

Mt 4:5 7

Lk 4:9 12

Mt §22S

Lk §20S

Mt 4:8 10

Lk 4:5 8

Once again, this is an example of the Syriac author working with Eusebius’ original sections, further subdividing them to make more explicit what is similar and different in these two accounts.

180

The Eusebian Canon Tables

Another example of a new Canon V section comes from the Sermon on the Mount. Eusebius had made Matthew 5:41–3 a section (Mt §39E) and placed it in Canon XMt as unique to the first gospel. The Syriac reviser, however, recognized that in fact the middle portion of this passage has a verbal similarity with a Lukan passage and he therefore singled out the relevant portion and joined it with a new Lukan section to create a new Canon V parallel: Mt §50S

Lk §68S

Mt 5:42

Give to the person who asks you, and do not refuse the one who wants to borrow from you.

Lk 6:30

Give to everyone who asks you, and if someone takes something of yours, do not demand it back.

The verbal correspondence between the above passages is obvious, so this is an instance in which the Syriac author has corrected what he might have regarded as an oversight on Eusebius’ part. The result is that a single Eusebian Canon XMt passage becomes in the Peshitta two passages in Canon XMt and one passage in Canon V.

Canon VI Several of the new Canon VI parallels arose because the Syriac reviser removed the Johannine portion of what was originally a Canon IV parallel, leaving only the passages from Matthew and Mark, which now formed a Canon VI parallel. One example is the treatment of the accounts of Jesus on the cross being offered wine to drink. Eusebius had divided the text so as to join the following passages together: Mt §333E Mt 27:34 they offered him wine to drink, mixed with gall; but when he tasted it, he would not drink it.

Mk §211E Mk 15:23

And they offered him wine mixed with myrrh; but he did not take it.

Jn §203E Jn 19:28 After this, when Jesus knew that all was now finished, he said (in order to fulfill the scrip ture), ‘I am thirsty’. Jn 19:29 A jar full of sour wine was standing there. So they put a sponge full of the wine on a branch of hyssop and held it to his mouth. Jn 19:30a When Jesus had received the wine, he said, ‘It is finished’.

Canon Tables 2.0: The Peshitta Version

181

In contrast, the Peshitta version presents only the following passages: Mt §397S

Mk §255S

They gave him wine vinegar mixed with gall to drink. He tasted it but did not want to drink it.

Mk 15:23 They gave him wine mixed with myrrh to drink, but he did not take it.

Mt 27:34

The reason for this change is that the Syriac reviser realized that a more plausible parallel for the Johannine passage was available in the other two gospels. In fact, both Matthew and Mark contain two reports of wine being offered to Jesus. The first is that one above, in Matthew 27:34 and Mark 15:23, which occurs at the beginning of the crucifixion account, just after Jesus arrives at Golgotha. They also, however, have a later account of sour wine being offered to Jesus on a sponge just prior to his death (Mt 27:48; Mk 15:36). The latter two passages were included by Eusebius in a separate Canon II parallel, along with a passage from Luke, who only has one offering of wine to Jesus (Mt §342E; Mk §222E; Lk §323E). The Gospel of John, like Luke, only has one offering of wine, and it is sour wine, occurring just prior to Jesus’ death, like the second offering of wine in Matthew and Mark. In other words, this Johannine passage presents a closer similarity to the passages Eusebius had included in the later Canon II parallel than it does to Matthew 27:34 and Mark 15:23.⁴³ The Syriac reviser accordingly removed the Johannine passage from Eusebius’ Canon IV parallel, reducing it to a Canon VI parallel, and added this passage to the Canon II parallel, to create a new Canon I parallel (Mt §406S; Mk §266S; Lk §370S; Jn §233S).⁴⁴ The new version rightly highlights that only Matthew and Mark have an initial refusal of the mixed wine, while all four evangelists have a later offering of sour wine on a sponge.

⁴³ As recognized by Watson, The Fourfold Gospel, p. 149, this appears to be one place in which Eusebius has made a slight mistake. ⁴⁴ In the process of correcting Eusebius’ error, the Syriac reviser seems to have made one of his own. The pertinent Lukan passage about the offering of wine on a sponge is Lk 23:36, which the Syriac labels as Lk §369S, yet the Lukan passage listed parallel to the other accounts is Lk §370S, which is a passage about the soldiers mocking Jesus. Lk §369S, about the sour wine, is instead presented as parallel to Mt §392S, Mk §250S, and Jn §209S (= Mt 27:27 30; Mk 15:16 19; Jn 20:2 3), which are passages about the mocking of the soldiers. It seems that these two Lukan passages (Lk §369S, 370S) were reversed and paired with the wrong parallels from Matthew, Mark, and John, a simple enough mistake. The remarks here are based on the edition of Pusey and Gwilliam. I checked it against the digital version of the Rabbula Gospels available online and the same reversal is evident there (fol. 221r v). Pusey and Gwilliam list no alternative readings from the manuscripts they surveyed for their edition, so this may be a mistake that goes all the way back to the person responsible for revising the system.

182

The Eusebian Canon Tables

One other new Canon VI parallel warrants mention because it highlights a difference between the text of the gospels in Eusebius’ edition and the text of the Peshitta version. Although Eusebius was aware of the longer ending of Mark, and discussed this as a textual problem in his influential work Quaestiones ad Marinum,⁴⁵ there is no evidence that he included these verses in his sectional enumeration. In fact, three later gospelbooks explicitly state, following Mark 16:8, ‘until this point Eusebius Pamphilus canonized’ (ἕως οὗ καὶ Ἐυσέβιος ὁ Παμφίλου ἐκανόνισεν).⁴⁶ It is therefore likely that the edition of the fourfold gospel that he issued with his new apparatus did not include the longer ending, but, even if it did, it must have been unnumbered. The Peshitta, however, did include the longer ending of Mark, and the person responsible for revising Eusebius’ paratext extended the enumeration for Mark to cover these verses, which allowed for the possibility of new parallels between these verses and other passages. One that the reviser highlighted was the following: Mt §426S

Mk §288S

Mt 28:19 Therefore go and make disciples of all nations, and baptize them in the name of the Father and the Son and the Holy Spirit, Mt 28:20 and teach them to obey everything I have commanded you. And behold, I am with you always, until the end of the age. Amen.

Mk 16:15 ‘Go to all the world’, he said to them. ‘And preach my good news throughout creation’.

This passage illustrates well one of the effects of the Canon Tables system that I argued for in chapter 3. By enumerating these verses and listing their section number in the Tables, the Peshitta version thereby incorporates them into the canonical space marked out by the Eusebian paratext, clearly demarcating them from that which is extra-canonical and aligning them with unquestionably authoritative passages such as Matthew 28:19–20. The Canon Tables apparatus thereby integrates this passage from the longer ending of Mark into the unified witness of the fourfold gospel.

⁴⁵ Eusebius, qu. Marin. 1.1 3. On Eusebius’ view on the ending of Mark, see now the following study: Coombs, A Dual Reception. ⁴⁶ The relevant manuscripts are the minuscules 1 (twelfth c.), 209 (fourteenth c.), and 1582 (tenth c.). Images are available on the INTF Virtual Manuscript Room. These colophons were highlighted by Burgon, The Last Twelve Verses, p. 120, and B. H. Streeter, The Four Gospels: A Study of Origins (London: Macmillan, 1924), p. 88. Some Greek manuscripts do include the longer ending of Mark in the Canon Tables apparatus, increasing the number of sections in Mark from 233 to 241. On the variation in the Greek manuscripts, see von Soden, Die Schriften des Neuen Testaments, pp. 394 6.

Canon Tables 2.0: The Peshitta Version

183

Canon VII Sometimes the new parallels created by the Syriac reviser take account of what appear to be trivial narrative details. For example, the introduction to the Sermon on the Mount in Matthew 5:1 was relegated by Eusebius to Canon XMt as unique to Matthew (Mt §24E). The Syriac version, however, juxtaposed this brief statement with a passage from the Gospel of John that appeared in a very different context, prior to the feeding of the 5,000, and thereby created a new Canon VII parallel: Mt §33S Mt 5:1

When Jesus saw the crowds, he went up on a mountain. After he sat down, his disciples ap proached him,

Jn §56S Jn 6:3

Jesus went up a mountain and sat down with his disciples.

The common element here is of course Jesus going up on a mountain with his disciples, though the passages that follow have little in common otherwise. It is not obvious what sort of theological or exegetical point could be made of the new parallel, but it at least demonstrates the close attention to detail exhibited by the Syriac reviser, taking account of relatively brief and seemingly insignificant verbal correspondences. Another example of a new Canon VII parallel is the treatment of the famous statement about Peter as the ‘rock’ following his confession of Jesus as Christ. Eusebius made this a Canon XMt passage (Mt 16:17–19 = Mt §167E) even though, as the Syriac reviser realized, there were parallels with John that might have been exploited. The Syriac version of the apparatus breaks this single section into three new sections, placing the first (Mt §202S) in Canon XMt and the latter two both in Canon VII. Matthew 16:18 (Mt §203S) he joins with John 1:42b (Jn §21S), both statements from Jesus about Peter’s name as ‘Cephas’. Matthew 16:19 (Mt §204S), a promise about the celestial consequence of Peter’s ‘binding’ and ‘loosing’, he paired with a similar Johannine statement about forgiveness of sins spoken to all the disciples in the upper room (Jn 20:23 = Jn §251S). The latter passage had been included by Eusebius in Canon VII alongside a later Matthean statement, also about ‘binding’ and ‘loosing’, directed to all the disciples (Mt 18:18 = Mt §185E = Mt §227S). The Syriac version retains the latter parallel between Matthew 18:18 and John 20:23 but duplicates the Johannine passage alongside the similar statement directed to Peter alone (Mt 16:19). The result is that two different passages from the Gospel of Matthew are now placed in parallel to the same passage from the Gospel of John, thereby linking together all three of the ‘binding and loosing’ and ‘forgiving’ passages in the gospels.

184

The Eusebian Canon Tables

One other new Canon VII parallel should be mentioned, because it reflects a difference in the text of the gospels used by Eusebius and that of the Peshitta version. The standard reading of Matthew 28:18 that appears in the Greek gospel tradition is ‘And Jesus came and said to them, “All authority in heaven and on earth has been given to me”’ (καὶ προσελθὼν ὁ Ἰησοῦς ἐλάλησεν αὐτοῖς λέγων, Ἐδόθη μοι πᾶσα ἐξουσία ἐν οὐρανῷ καὶ ἐπὶ γῆς). The Peshitta, however, adds a subsequent clause to this verse: ‘Jesus drew near and spoke with them, “All authority in heaven and on earth has been given to me”, he told them. “As my Father sent me, I am sending you”’ ( ).⁴⁷ The latter portion of this verse no doubt came originally from the post-resurrection scene in the upper room from the fourth gospel, when Jesus says to the disciples, ‘Peace be with you. As the Father has sent me, so I send you’ (Jn 20:21; Εἰρήνη ὑμῖν· καθὼς ἀπέσταλκέν με ὁ πατήρ, κἀγὼ πέμπω ὑμᾶς). It is possible that the Greek Vorlage from which the Peshitta was made had this rare reading, but it may also have originated in the melding of Matthew and John that took place in the Diatessaron, in which case it might have been passed on to the Old Syriac version of the gospels, and then survived into the Peshitta revision of the Old Syriac.⁴⁸ Whatever the case, the Syriac reviser of Eusebius’ paratext recognized that the version of Matthew 28:18b known to him was verbally identical with John 20:21b, so he subdivided Eusebius’ original sections in each gospel to select the relevant portion (Mt §425S; Jn §249S), and joined them to create a new Canon VII parallel. In this instance the Syriac reviser was not highlighting a potential parallel omitted by Eusebius, since the Caesarean historian could not have been aware of it, but was instead adapting his paratextual system to better suit the text of the Peshitta version. Despite his close attention to detail, there are a handful of instances in which the Syriac revision presents what appear to be errors. An example is the treatment of John 7:43 (=Jn §84E), a single verse that Eusebius had placed in Canon XJn. The Syriac revision creates a new Canon VII passage in the verse immediately prior, John 7:42 (=Jn §94S), aligning it with Matthew 2:5–6 (=Mt §5S), a sensible parallel, since both passages speak about the Messiah being born in Bethlehem. However, the Syriac version extends this new section to include Jn 7:43 as well, despite the fact that it has no correspondence with the Matthean passage. It would have made

⁴⁷ According to the textual apparatus in NA²⁸, this reading only shows up elsewhere in one late Greek ms (Θ). ⁴⁸ Burkitt, Evangelion da Mepharreshe, vol. 1, pp. 172 3, proposed that the Peshitta took over this reading from the Old Syriac. Unfortunately, all of the Old Syriac manuscripts are missing this portion of text.

Canon Tables 2.0: The Peshitta Version

185

more sense for Jn §94S to conclude with John 7:42, and for John 7:43 to remain a Canon XJn passage. Most likely the Syriac reviser, after marking a new section at the beginning of John 7:42, overlooked that Eusebius had broken the passage at 7:42/43, and mistakenly carried his new Canon VII section further than was necessary.

Canon VIII Another means of increasing the overall number of parallels is to split apart some of those that Eusebius had originally joined together. For example, Eusebius had originally presented the following parallel in Canon II: Mt §94E

Mk §86E

Mt 10:33 but whoever denies me before others, I also will deny before my Father in heaven

Mk 8:38 Those who are ashamed of me and of my words in this adulterous and sinful generation, of them the Son of Man will also be ashamed when he comes in the glory of his Father with the holy angels.

Lk §97E

Lk §146E

Those who are ashamed of me and of my words, of them the Son of Man will be ashamed when he comes in his glory and the glory of the Father and of the holy angels.

Lk 12:9 but whoever denies me before others will be de nied before the angels of God.

Lk 9:26

Some of these passages correspond verbally but some are only thematically similar. Note that this is one of the instances in which Eusebius repeated a Matthean passage alongside multiple Lukan passages to create two Canon II parallels. The Syriac version splits the above parallels apart to create new Canon V and Canon VIII parallels: Mt §115S

Lk §175S

Mt 10:33 but whoever denies me before people, I will also deny him or her before my Father in heaven.

Lk 12:9 but whoever denies me before people will be denied before the angels of God.

Mk §107S Mk 8:38 Anyone who is ashamed of me and my words in this sinful and adulterous gen eration, the Son of Man will also be ashamed of them when he comes in his Father’s glory with his holy angels.

Lk §122S Lk 9:26

Anyone who is ashamed of me and my words, the Son of Man will be ashamed of them when he comes in his Father’s glory with his holy angels.

186

The Eusebian Canon Tables

While Eusebius’ original four passages had a common theme but not identical wording, the Syriac has split the four apart into two groups of two that each have a tighter verbal similarity, one about ‘denial’ and one about ‘shame’. Note that in this instance the Syriac reviser has not created any further subdivisions within any of the gospels, but has still increased the overall number of parallels by splitting one Canon II parallel into two parallels, one in Canon V and one in Canon VIII. This alteration achieves greater precision in the cross-referencing system, but also sacrifices some of its utility as a topical index, since the two pairs of passages are now no longer linked. The longer ending of Mark also created opportunities for new Canon VIII parallels. The story of the two disciples who meet the risen Jesus on the road to Emmaus is a unique Lukan passage (Lk 24:13–35 = Lk §339E), so Eusebius accordingly placed in Canon XLk. However, the longer ending of Mark also makes brief reference to this event (Mk 16:12–13), so the Syriac version connects these two passages as a Canon VIII parallel (Mk §285S; Lk §393S). Once again, the longer ending of Mark is integrated into the rest of the gospel witness to the risen Jesus.

Canon IX I noted earlier that the number of Canon IX parallels actually decreased in the Syriac version in comparison with Eusebius’ original, from twenty-one to nineteen. In fact, the Peshitta loses seven of Eusebius’ original twentyone, amounting to a third of Eusebius’ original total, though it then adds five new parallels to mitigate the loss. All of the parallels that are removed have a common feature. On occasion Eusebius presented the same passage in Matthew as parallel to multiple passages in other gospels. The second parallel in Canon I is a good example, with Mt §11E, Mk §4E, and Lk §10E being placed in parallel to Jn §6E, and then repeated for Jn §12E, 14E, and 28E. In Canon IX (Lk–Jn), with Luke rather than Matthew serving as his baseline, Eusebius took this approach to an extreme, using only eight discrete passages from Luke, but repeating them alongside various passages from John to produce the twenty-one parallels included in the table. Five of these Lukan passages in fact are repeated three times each (Lk §274E, 303E, 307E, 312E, 341E). For example, both Luke and John record three statements made by Pilate about Jesus’ innocence (Lk 23:4, 14, 22; Jn 18:38, 19:4, 6), and one would expect each of the three Lukan passages to be aligned in sequence with the three Johannine passages, producing three parallels between these two gospels. Eusebius, however, repeated all three Johannine passages (Jn §182E, 186E, 190E) alongside each of the Lukan passages

Canon Tables 2.0: The Peshitta Version

187

(Lk §303E, 307E, 312E), turning these three parallels into nine and giving the impression of more Luke–John parallels than there actually are. The Syriac reviser, recognizing this superfluity, removed the excess by reducing all but one of these repeated parallels in Canon IX.⁴⁹ For example, statements of innocence one, two, and three from Pilate are now, in the new system, aligned with one another across Luke and John as Lk §349S, Jn §206S; Lk §354S, Jn §210S; and Lk §359S, Jn §214S. Of the new parallels added to Canon IX, some are in fact only more precise subdivisions of parallels Eusebius had already noted. For example, Eusebius had treated the great catch of fish in Luke 5:4–7 (=Lk §30E) as one large block parallel to two portions of the Johannine great catch of fish (Jn 21:1–6, 11 = Jn §219E, 222E). The Syriac, however, divides the chunk of Lukan text into four sections (Lk §36S–39S), now placed alongside three Johannine sections (Jn §256S, 257S, 261S), to produce three Canon IX parallels in place of Eusebius’ original two. There are also genuinely new parallels in Canon IX that Eusebius had previously failed to take note of. The two that stand out most clearly are the fact that both Luke and John report that Jesus’ tomb was previously unused (Lk 23:53d; Jn 19:41c = Lk §382S; Jn §239S) and that both evangelists also uniquely report Peter’s visit to the empty tomb (Lk 24:12; Jn 20:6–10 = Lk §392S; Jn §242S). The former parallel Eusebius had included as a part of a larger Canon I parallel, even though only Luke and John present this detail, and the latter two passages about Peter’s actions on Easter morning he had included in their respective Canon X tables (Lk §339E; Jn §210E). Hence, even though the total number of passages listed in Canon IX decreases in the Peshitta revision, it nevertheless forges new links between these two gospels, and thereby presents the reader with information that was previously unaccounted for in the Eusebian original.

Canon XMt Canon X passages should be the least complicated to create, given that they do not require coordination across multiple gospels but simply the identification of unique material that was previously a part of a larger passage in another Canon. One exemplary passage is Matthew 5:14–16, a dominical saying that occurs in Matthew and Luke. Eusebius treated these three verses together as a section, placed in Canon Table II with the following parallels: ⁴⁹ The only triple parallel that the Peshitta version preserves is Lk §314S; Jn §266S, 268S, 270S (= Lk 22:32b; Jn 21:15c, 16c, 17c), which is the dominical command for Peter to care for the nascent Christian community.

188

The Eusebian Canon Tables Mt §32E

Mk §39E

Lk §79E

Mt 5:14 You are the light of the world. A city built on a hill cannot be hid. Mt 5:15 No one after lighting a lamp puts it under the bushel basket, but on the lamp stand, and it gives light to all in the house. Mt 5:16 In the same way, let your light shine before others, so that they may see your good works and give glory to your Father in heaven.

Mk 4:21 He said to them, ‘Is a lamp brought in to be put under the bushel basket, or under the bed, and not on the lampstand?’

Lk 8:16 No one after lighting a lamp hides it under a jar, or puts it under a bed, but puts it on a lampstand, so that those who enter may see the light.

The Syriac retains these parallels, but significantly pares down the Matthean passage, dividing it into three sections. Matthew 5:14 and 5:16 (Mt §41S, 43S) now are assigned to Canon XMt, while only Matthew 5:15 (Mt §42S) is retained in the Canon II parallel (along with Mk §51S and Lk §99S), obviously because it is the portion of the original with the closest similarity to the other two passages, since it speaks of a ‘lamp’. In contrast, the metaphor of a ‘city’ in Matthew 5:14 and the command to ‘let your light shine’ in Matthew 5:16 are unique to Matthew.

Canon XMk Because so much of Mark is shared by Matthew and Luke, Canon XMk is perhaps the most difficult table to add to. Eusebius had already done a good job of identifying most of the unique Markan passages, but the Peshitta reviser managed to highlight several more, mostly by sectioning off short sentences from larger passages that are shared across the Synoptics. For example, Eusebius had created the following Canon VI parallel: Mt §152E

Mk §68E

Mt 14:32 When they got into the boat, the wind ceased. Mt 14:33 And those in the boat worshiped him, saying, ‘Truly you are the Son of God’. Mt 14:34 When they had crossed over, they came to land at Gennesaret.

Mk 6:51 Then he got into the boat with them and the wind ceased. And they were utterly astounded, Mk 6:52 for they did not under stand about the loaves, but their hearts were hardened. Mk 6:53 When they had crossed over, they came to land at Gennesaret and moored the boat.

Canon Tables 2.0: The Peshitta Version

189

As can be seen, the opening and closing of these two passages are nearly identical, though the middle verse in each presents two very different reactions from the disciples. The Peshitta version of Eusebius’ system therefore keeps the opening and closing of the Markan passage aligned with the two halves of the Matthean text as two separate Canon VI parallels (Mt §183S, Mk §87S; Mt §184S, Mk §89S), and creates a new Canon XMk section for the middle verse of the Markan passage that is unique (Mk §88S).

Canon XLk With respect to Canon XLk we should first observe a peculiarity in the Peshitta revision of Eusebius’ paratext. Luke’s dedicatory preface to Theophilus (Lk. 1:1–4) should belong in this table because it is unique to the third gospel, as indeed Eusebius had done (Lk §1E). The Syriac, however, begins Lk §1S not at Luke 1:1 but instead at Luke 1:5 (see fol. 143v of the Rabbula Gospels). As a result, the opening four verses of the gospel are unnumbered, and therefore not represented in any of the ten Canons. The only other gospel to begin with a Canon X section is Matthew, and here also the reviser began Mt §1S not at Matthew 1:1, but at Matthew 1:2. Perhaps the scribe regarded both passages as prefaces and so separate from the body of the text. Despite the overall increase in the number of sections in this Canon, the Syriac reviser also removed two sections from the table in order to correct inconsistencies in Eusebius’ original design. I pointed out in chapter 2 that, according to the logic of Eusebius’ system, there should never be two Canon X sections in sequence in any gospel, because a new section is only supposed to begin where the text changes to a different relational category. Nevertheless, twice in the Gospel of Luke back-to-back Canon XLk passages occur: Luke §67E–68E (=Lk 7:11–16, 17) and §163E–164E (=Lk 13:1–5, 6–13). In the Peshitta version of Luke, each of these pairs of sections is collapsed into one (Lk §85S, 191S), resolving the inconsistency in the rationale for the sectioning of the text. As with the previous two Canon X tables, the new additions to Canon XLk mostly arise from breaking off details in Luke’s version of events or sayings from larger sections of parallel passages. For example, Eusebius had created a Canon I parallel for the verses from each gospel in which Jesus took bread, blessed it, and identified it with his body (Mt 26:26; Mk 14:22; Lk 22:19; Jn 6:35a, 48, 51, 55 = Mt §284E; Mk §165E; Lk §266E; Jn §55E, 63E, 65E, 67E). However, across these various passages, Luke alone records the dominical command ‘Do this in remembrance of me’. The Peshitta version therefore isolates this short sentence from the larger passage in order to create a new Canon XLk section

190

The Eusebian Canon Tables

(Lk §304S), thereby highlighting Luke’s distinctive contribution to the accounts of the last supper and the origins of the eucharistic ritual. Similarly, although all three Synoptics record Judas’ greeting of Jesus with a kiss on the night of his betrayal (Mt 26:48–50; Mk 14:44–6; Lk 22:47b–48 = Mt §301E; Mk §182E; Lk §286E), only Luke records that Jesus rhetorically asked Judas, ‘Is it with a kiss that you are betraying the Son of Man?’ (Lk 22:48). Accordingly, the Peshitta breaks this verse off from a larger Canon II section in order to make it a new Canon XLk section (Lk §328S).

Canon XJn In Canon XJn we observe one of the very few mistakes in the revised version of Eusebius’ paratext. The fourth gospel uniquely records the wedding at Cana at which Jesus turned water into wine (Jn 2:1–11), so Eusebius placed this passage in Canon XJn along with the immediately preceding verses (1:43–51) to create one large section of unique material (Jn §18E). I noted above that the Syriac reviser excerpted a portion of this large block of text, specifically John 1:49–51, in order to align Nathanael’s confession of Christ with similar confessions across the gospels. One would expect that, after the conclusion of the exchange between Nathanael and Jesus, the Peshitta would revert back to a Canon XJn passage for the wedding at Cana episode. In fact, however, the section that begins at John 1:49 extends all the way through 2:11, and so is placed in parallel with the confessions of Peter, Andrew, and others, even though it is difficult to see how the wedding at Cana presents a similarity with these other passages. The most likely explanation for this oddity is that the Syriac reviser meant to stop his new section at John 2:1, in order to isolate only Nathanael’s confession for the purpose of his new Canon I parallel, but has mistakenly carried on the section until the break in the text that Eusebius had originally created at 2:11. Despite this evident mistake, it is also the case that the Syriac reviser once again corrected an inconsistency in Eusebius’ system. Another pair of back-to-back Canon X passages occurred at John 7:33 and 7:34–9 (=Jn §80E, 81E). In this instance, rather than collapse them together, the Syriac reviser found parallels in Matthew and Mark for the first passage, thereby moving it into Canon IV (Mt §326S; Mk §191S; Jn §90S = Mt 26:11b; Mk 14:7b; Jn 7:33).⁵⁰ Perhaps these were even the passages that Eusebius had in mind when he originally made John 7:33 a distinct section, though for whatever reason it ended up in Canon X in his system rather than Canon IV.

⁵⁰ Actually, the Matthean and Lukan passages here were already a Canon IV passage in Eusebius’ version, though joined with another Johannine passage (Jn §98E). The Syriac reviser added Jn 7:33 to this existing parallel.

Canon Tables 2.0: The Peshitta Version

191

In the immediately preceding example, the Syriac reviser made a change that resulted in the loss of one Canon XJn passage. In another instance, it appears that he again was perhaps correcting Eusebius and in so doing added a new passage to Canon XJn. The dominical promise that a prayerful request made in faith and in the name of Jesus will be granted occurs in Matthew, Mark, and John, though in Matthew and Mark it occurs in the midst of the episode of the cursing of the fig tree (Mt 21:22; Mk 11:24), while in John it is integrated into the farewell discourse in the upper room (Jn 14:13–14). Eusebius made these passages a Canon IV parallel (Mt §216E; Mk §125E; Jn §128E), but inexplicably carried on the Johannine section to include John 14:15–21 as well, even though these verses are unrelated to the Matthean and Markan passages. The Peshitta revision preserves this Canon IV parallel (Mt §259S; Mk §150S; Jn §151S), but correctly makes a break after John 14:14, with the result that John 14:15–21 becomes a new Canon X section (Jn §152S). Another new Canon X section occurs at John 19:5. Eusebius had presented this verse (Jn §187E) in Canon IV in parallel with Matthew 27:27–9 and Mark 15:16–19 (= Mt §329E; Mk §207E), presumably because all three passages make mention of Jesus’ crown of thorns and purple robe. However, the scenes in Matthew and Mark describe the mockery of Jesus by the soldiers in the praetorium, while John 19:5 concerns Pilate’s presentation of Jesus to the crowd with the famous declaration, ‘Behold the man!’ The latter scene occurs only in the fourth gospel, and the Syriac reviser, recognizing this, shifted John 19:5 (= Jn §211S) from Canon IV to Canon XJn.

CO NCLUSION The above examples are representative of the kinds of changes that were made throughout Eusebius’ system by the Syriac scribe. Notably these alterations are not isolated to a single gospel, or to individual parts of various gospels, or only to certain Canons, which suggests that this unnamed person undertook a systematic revision of the entirety of the paratext he found in his Greek exemplar. Given the requirements of this task, he must have engaged in the most thorough study of the similarities and differences amongst the gospels since Eusebius himself. Although he seemingly did not set about interpreting the passages and formulating a theory about gospel writing, as had Augustine, the Syriac reviser stands in continuity with Ammonius and Eusebius in using the methodology of demarcating sections of text for the purpose of comparative analysis. For the most part, his revision represents an upgrade of Eusebius’ original, resulting in an even more powerful reference tool for later users to employ. Though we can recover the rationale behind the revision only through inference, the kind of reasoning it implies is not

192

The Eusebian Canon Tables

fundamentally different from that of the Irish exegetes we will consider in the next chapter who set about classifying parallels in different kinds of relational categories. Because the four gospels present varying kinds of parallels with a range of types of similarities, this process of analysing related passages can never be complete but always remains open to debate and alternative resolution. What is unique about the Syriac tradition is that this scholarly investigation resulted in a completely revised version of the original Eusebian paratext. Future studies of the transmission of the apparatus in other languages might reveal isolated editorial interventions undertaken for similar reasons, but such a thorough overhaul of the original has not survived in any other tradition. It is also striking that the difficult and tedious labour that must have been required to produce the Peshitta Canon Tables does not appear to have been driven by any polemical motivation. The parallels the Syriac scribe revised and the new ones he created reveal no theological or exegetical agenda. The closest we can come to isolating a contextual factor that might have given impetus to this project is the fact that Syriac-speaking Christians continued to use Tatian’s gospel text well into the fifth century, at least according to the report of Theodoret. Syriac scribes accustomed to copying the ‘Gospel of the Mixed’ as well as the ‘Gospel of the Separated’ must have had a greater awareness of the interrelationships among the four gospels than almost anyone else in the world of late antiquity. Perhaps it was this peculiar situation that motivated the anonymous scribe or scribes to expend the energy required for such a laborious task. One further fact about this revised version must be mentioned because it is so obvious that it is easy to miss. The person responsible for it ensured that he would for ever remain hidden in Eusebius’ shadow. The Peshitta Letter to Carpianus makes no mention of the fact that the paratextual apparatus that follows is in fact a revised version of Eusebius’ original, but instead presents it as originating with the Caesarean historian himself.⁵¹ The result of this silence of the scribe about his own work is that a reader of the Peshitta version of the gospels would have no idea that his edition of the Canon Tables differed from that in the Greek gospel tradition. This authorial self-effacement also meant that the stature of Eusebius would remain high for later users of the Peshitta version.

⁵¹ This anonymous updating is analogous to the transmission of Ptolemy’s Handy Tables. Cf. Ossendrijver et al., ‘Tables in Antiquity’: ‘It seems clear that the Handy Tables were never regarded as canonical in the same sense as works like the Elements or the Almagest . . . The scholars who worked with, and transmitted, the Handy Tables throughout the late ancient and medieval periods seem not to have been much concerned with the idea of authorial intention and the tables were regarded [as] objects of contemporary use, not ancient authority’.

Canon Tables 2.0: The Peshitta Version

193

In fact, some later Syriac sources eventually came to attribute an even more significant role to Eusebius than can be maintained on strictly historical grounds. The eighth-century bishop of Mosul, Moses bar Kepha, in answer to the question ‘who collected the four books of the evangelists and set them in order in one book?’ explains that some people, indeed, say that Eusebius of Caesarea, when he saw that Julianus of Alexandria made the Gospel of the Diatessaron, i.e. by means of the Four, and changed the sequence of the words of the Gospels, [and also Tatian] the Greek, the heretic made a Gospel, the one that is called Tasaron, and he too changed the sequence of the words, he, Eusebius, took care and collected the four books of the four Evangelists and ordered them and placed them in one book, and preserved the body of the evangelists’ narration as it was, without taking away anything from their narration or adding anything to them, and he made certain Canons concerning their agreement and disagreement with one another.

Bar Kepha’s opening question shows a striking degree of awareness that the fourfold gospel itself has a history, in his tacit admission that these four texts have not always sat together between the covers of a single codex. He (or whatever earlier Syriac source he was drawing upon) probably came to think that Eusebius was responsible for collecting these four texts together based on the fact that Peshitta gospel codices began with Eusebius’ prefatory Letter to Carpianus, and, in the absence of evidence to the contrary, it was a small step to attribute not just the marginal apparatus but the form of the book itself to the Caesarean historian. Of course, with more evidence at our disposal than bar Kepha had, we now know that the four-gospel codex came into existence at least half a century before Eusebius’ labours. However, bar Kepha’s claim

⁵² Harris inexplicably omitted the phrase , which I have translated as ‘disagree ment’, but Kmosko supplied it. ⁵³ Moses bar Kepha, comm. Mt., intro. §53 (Harris, p. 21). An edition of the fragmentary remains of Bar Kepha’s Commentary on Matthew still has not been published. The passage I cite here was excerpted and published by J. Rendel Harris on the basis of British Library, Add. MS 17,274. Harris did not provide the folio number. The passage was later republished in a slightly more extensive form in M. Kmosko, ‘Analecta Syriaca e codicibus Musei Britannici excerpta’, Oriens Christianus 3 (1903): pp. 104 5. On this passage see also Petersen, Tatian’s Diatessaron, pp. 55 6. Note that Ammonius’ name has been garbled, with the reading (‘Julianus’) probably just a scribal error for

194

The Eusebian Canon Tables

that Eusebius collected these four narratives into a single contains a kernel of truth, since, as argued in chapter 1, Eusebius’ Canon Tables were in fact issued as a component of a new edition of the fourfold gospel, one in which his marginal apparatus both highlighted the canonical boundary between this collection and any potential rival texts and, also, bound these four texts together as a united witness to the story of Jesus.

6 Scholarly Practices The Eusebian Canon Tables in the Hiberno-Latin Tradition

Although this chapter continues the chronological progression that has characterized the last two chapters, we now move from the eastern Mediterranean back to the Latin West, specifically to the furthest edge of western Europe, to an area that had remained beyond the reach of Roman domination even at its fullest extent.¹ Thanks to its incorporation into Jerome’s Vulgate translation, the Eusebian apparatus spread widely in the Latin world from the early fifth century,² usually with Jerome’s dedicatory letter to Pope Damasus

¹ An earlier version of this chapter was published as Matthew R. Crawford, ‘Scholarly Practices: The Eusebian Canon Tables in the Hiberno Latin Tradition’, in Producing Christian Culture: Medieval Exegesis and Its Interpretative Genres, ed. Giles E. M. Gasper, Francis Watson, and Matthew R. Crawford (London: Routledge, 2017), pp. 65 88. ² The starting point for all discussions of the Latin Canon Table tradition remains Norden falk, Die spätantiken Kanontafeln, pp. 166 220, who pointed out that, instead of the ten page layout of Eusebius’ archetype, Latin gospelbooks tend to have either a twelve page ‘shorter Latin series’ or one of two sixteen page ‘longer Latin series’. For a comparison of some of the parallels in the Vulgate tradition with their counterparts in the Greek tradition represented by the Nestle Aland edition, see Thiele, ‘Beobachtungen’, pp. 100 11, and for studies showing how Canon Table versions can be used to establish relationships between manuscripts, see Patrick McGurk, ‘The Canon Tables in the Book of Lindisfarne and in the Codex Fuldensis of St. Victor of Capua’, Journal of Theological Studies 6 (1955): pp. 192 8; Patrick McGurk, ‘The Disposition of the Numbers in Latin Eusebian Canon Tables’, in Philologia Sacra: Biblische und patristische Studien für Hermann J. Frede und Walter Thiele zu ihrem siebzigsten Geburstag, ed. Roger Gryson, Vetus Latina/Aus der Geschichte der lateinischen Bibel 24/1 (Freiburg: Herder, 1993), pp. 242 58; O’Loughlin, ‘The Eusebian Apparatus in Some Vulgate Gospel Books’, pp. 1 92. On prefatory and paratextual material in the Latin biblical tradition, see the two recent reprints of the classic studies by De Bruyne: Donatien De Bruyne, Pierre Maurice Bogaert, and Thomas O’Loughlin, Prefaces to the Latin Bible, Studia traditionis theologiae 19 (Turnhout: Brepols, 2015); Bruyne, Bogaert, and O’Loughlin, Summaries, Divisions and Rubrics of the Latin Bible; and on the Latin New Testament in general, see especially Houghton, The Latin New Testament, who discusses the Eusebian apparatus on pp. 200 2.

The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0007

196

The Eusebian Canon Tables

(known as the Novum opus) serving as a preface.³ In fact, the number of Latin manuscripts with Canon Tables dated between the fifth and ninth centuries far exceeds the number of surviving Greek manuscripts with the Eusebian apparatus from the same period.⁴ Three authors provide supporting evidence for the growing prominence of Eusebius’ paratext in this period.⁵ First, Victor of Capua, in the mid-sixth century, modified the apparatus so that it could function as a navigation system in the harmonized version of the gospels he included in the famous Codex Fuldensis, which in some form goes back to the Diatessaron of Tatian.⁶ Second, later in the sixth century Cassiodorus, in his ³ O’Loughlin, ‘Harmonizing the Truth’, p. 14, detected in the Novum opus several ‘mistaken impressions’. First, Jerome thought the purpose is ‘to provide a cross referencing system’; secondly, he supposed that one should use the Canon number in the margin to determine how many other gospels the passage occurred in; thirdly, he thought the purpose of the system was merely to show that the four gospels agreed; fourthly, his use of the word capitula implies that the sections are chapter like divisions; finally, Jerome’s description suggests to the reader that the system is ‘far too complicated for us to engage with actively’. This is a tendentious and mistaken reading of the Novum opus. On the first point, it is true that Jerome envisions this as a cross referencing system, but then again so did Eusebius, so he is hardly mistaken in this matter. Secondly, Jerome gives explicit instructions in the final paragraph that the reader should turn back to the Canons to find the parallel passages, so it is not the case that he expects the reader to be content with merely noting the Canon number in the margin. Thirdly, Jerome, like Eusebius, speaks of passages that are eadem, vicina, and propria, so it is not at all clear that his understanding of how the gospels relate differs from that of his predecessor. Fourthly, Jerome’s usage of capitula corresponds precisely to Eusebius’ usage of the term κεϕάλαια in the Letter to Carpianus. Finally, as already noted, in the final paragraph Jerome walks his reader through the steps of using the apparatus in the same way Eusebius had done in the Letter to Carpianus, so he clearly expects this to be a tool that is used, and is not attempting merely to impress his readers with something that seems so technical and complex that it is beyond the reach of the average reader of the gospels. ⁴ For a listing of all the manuscripts up to the Carolingian period that have Canon Tables, see Patrick McGurk, Latin Gospel Books: From A.D. 400 to A.D. 800 (Paris: Aux Éditions «Erasme», 1961), pp. 110 11. At some point an unknown scribe translated Eusebius’ Letter to Carpianus into Latin as well, though it remained much rarer than the Novum opus. In the period up to the Carolingian renaissance it appears in only ten surviving manuscripts, in every case alongside the more common Novum opus. Cf. McGurk, Latin Gospel Books, pp. 110 11. As pointed out by McGurk (pp. 64 5), Poitiers, Médiathèque François Mitterrand, MS 17 (65) has a version of ep. Carp. that is different from the one included in the other manuscripts, on which see also Donatien De Bruyne, ‘La préface du Diatessaron Latin avant Victor de Capoue’, Revue Bénédictine 39 (1927): pp. 5 11; McGurk, ‘The Canon Tables’, pp. 194 5. O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, pp. 98 n. 9, 99, incorrectly assumed that Jerome himself translated the Letter to Carpianus. ⁵ On these three authors, see also Elizabeth Mullins, ‘The Insular Reception of the Eusebian Canon Tables: Exegesis and Iconography’ (University College Cork, 2001), pp. 32 9. ⁶ The most recent edition of Codex Fuldensis is E. Ranke, Codex Fuldensis (Leipzig, 1868), and Victor’s preface, in which he discusses his treatment of the Canon Tables, is on pp. 1 3. A new edition of the text is in preparation by Nicholas J. Zola, and a digitized version of the manuscript can be accessed here: http://fuldig.hs fulda.de/viewer/image/PPN325289808/1 [accessed 16 January 2018]. On Fuldensis, see Petersen, Tatian’s Diatessaron, pp. 45 51; Ulrich Schmid, Unum ex quattuor: Eine Geschichte der lateinischen Tatianüberlieferung, Aus der Geschichte der lateinischen Bibel 37 (Freiburg im Breisgau: Herder, 2005); Houghton, The Latin New Testament, pp. 56 8.

Scholarly Practices: the Hiberno-Latin Tradition

197

Institutiones divinarum et saecularium litterarum, gave a brief notice about the Eusebian paratext in the midst of a list of authors who had written commentaries on the gospels.⁷ Finally, in the early seventh century Isidore of Seville, in book six of his massive Etymologiae, provided a short introduction to the Canon Tables, which is heavily dependent upon Eusebius’ Letter to Carpianus and Jerome’s Novum opus.⁸ The inclusion of discussions of the Eusebian apparatus in Cassiodorus’ Institutiones and Isidore’s Etymologiae, two of the most influential texts in the formation of the Latin medieval tradition, illustrates the degree to which the paratext had become a staple feature of Vulgate gospelbooks.⁹ Nevertheless, despite the wide diffusion of the Canon Tables in manuscripts, evidence of their actual use as a cross-referencing tool for studying parallel passages of text is sparse in this period, whether in Latin, Greek, or Syriac sources. The one great exception to this general observation is the Hiberno-Latin exegetical tradition.¹⁰ Early medieval Irish scholars show a ⁷ Cassiodorus, inst. 1.7.2 (Mynors, p. 28). ⁸ Isidore, etymol. 6.15. On Isidore’s dependence upon Jerome, see Stephen A. Barney et al., The Etymologies of Isidore of Seville (Cambridge: Cambridge University Press, 2006), p. 15. Against O’Loughlin, ‘Harmonizing the Truth’, p. 22, who asserted that Isidore was ‘trying to make sense of Jerome without another source of information to assist him’. He attributed Isidore’s ‘lack of concision’ in this section to his poor understanding of the apparatus, based on a merely cursory reading of Jerome’s Novum opus. On the contrary, his ‘lack of concision’ is more likely due to the fact that he is drawing on multiple sources. ⁹ A final, more indirect indication of the importance of Eusebius’ apparatus in the West is the fact that his Canon Tables probably inspired the Canons on the Letters of the Apostle Paul composed in the 380s by Priscillian of Avila, on which see Lang and Crawford, ‘The Origins of Pauline Theology’, pp. 139 44. ¹⁰ On the Hiberno Latin exegetical tradition, see Bernhard Bischoff, ‘Turning Points in the History of Latin Exegesis in the Early Middle Ages’, in Biblical Studies: The Medieval Irish Contribution, ed. Martin McNamara, Proceedings of the Irish Biblical Association 1 (Dublin: Dominican Publications, 1976), pp. 74 160, first published as Bernhard Bischoff, ‘Wendepunkte in der Geschichte der lateinischen Exegese im Frühmittelalter’, Sacris Erudiri 6 (1954): pp. 189 279. The Irish provenance or influence of the works in Bischoff ’s catalogue has been challenged by some. See Clare Stancliffe, ‘Early “Irish” Biblical Exegesis’, Studia Patristica 12 (1975): pp. 361 70; Michael Gorman, ‘A Critique of Bischoff ’s Theory of Irish Exegesis: The Commentary on Genesis in Munich Clm 6302 (Wendepunkte 2)’, Journal of Medieval Latin 7 (1997): pp. 178 233; Michael Gorman, ‘The Myth of Hiberno Latin Exegesis’, Revue Bénédictine 110 (2000): pp. 42 85. However, see the responses to this critique in Michael Herren, ‘Irish Biblical Commentaries Before 800’, in Roma magistra mundi: Itineraria culturae medievalis: Mélanges offerts à Père L.E. Boyle, ed. Jacqueline Hamesse (Leuven: Fédération des Instituts d’Etudes Médiévales, 1998), pp. 391 407; Dáibhí Ó Cróinín, ‘Bischoff ’s Wendepunkte Fifty Years on’, Revue Bénédictine 110 (2000): pp. 204 37; Charles D. Wright, ‘Bischoff ’s Theory of Irish Exegesis and the Genesis Commentary in Munich Clm 6302: A Critique of a Critique’, Journal of Medieval Latin 10 (2000): pp. 115 75. Of the four texts considered below, the first and last come from named Irish authors, and the editors of the critical editions of the two anonymous works are confident of their Irish origin or influence, so I will proceed assuming these may rightly be called ‘Hiberno Latin’, even while acknowledging that in important respects that tradition participated in the wider western exegetical culture. On Hiberno Latin exegesis more broadly, see the collected essays in Thomas O’Loughlin, ed., The Scriptures and Early

198

The Eusebian Canon Tables

degree of interest in the Eusebian apparatus as cross-referencing device that is completely without parallel in the time period covered by this monograph. Recognition of this aspect of the Hiberno-Latin tradition is not new. Bernard Bischoff, in his seminal study inaugurating the study of Hiberno-Latin exegesis over half a century ago, highlighted an interest in the Canon Tables as a distinctive feature of this Irish tradition, but this observation has not received much elaboration in subsequent scholarship, nor has it been integrated into the overall reception history of the Eusebian apparatus.¹¹ In what follows I will consider four representative texts which illustrate the way in which early medieval Irish scholars exploited the potential of the Canon Tables for understanding the gospels. This is not an exhaustive study of the Hiberno-Latin tradition, since other texts, such as gospel commentaries, might also have been included.¹² Rather, I have restricted my focus to four Medieval Ireland: Proceedings of the 1993 Conference of the Society for Hiberno Latin Studies on Early Irish Exegesis and Homiletics (Turnhout: Brepols, 1999), and, on a more introductory level, Martin McNamara, ‘Celtic Scriptures: Text and Commentaries’, in An Introduction to Celtic Christianity (Edinburgh: T&T Clark, 1989). Also important are J. F. Kelly, ‘A Catalogue of Early Medieval Hiberno Latin Biblical Commentaries (I)’, Traditio 44 (1988): pp. 537 71; J. F. Kelly, ‘A Catalogue of Early Medieval Hiberno Latin Biblical Commentaries (II)’, Traditio 45 (1989 90): pp. 393 434. ¹¹ Bischoff comments, ‘The occasional mention of the appurtenance of individual canons to this or than [sic] canon, or the enumeration of the chapters in individual canons . . . , is a further characteristic of the Irish interpretation of the Gospels’ (Bischoff, ‘Turning Points’, p. 84). So also Dominique Barbet Massin, L’enluminure et le sacré: Irlande et Grande Bretagne VIIe VIIIe siècles (Paris: Presses de l’université Paris Sorbonne, 2013), p. 371: ‘L’exégèse irlandaise, en général, s’intéresse aux tables des canons’. The most complete study of this Irish interest in Canon Tables is Elizabeth Mullins’ doctoral thesis from University College Cork, which I cite, where appropriate, in what follows. Only a portion of Mullins’ dissertation has been published: Elizabeth Mullins, ‘The Eusebian Canon Tables and Hiberno Latin Exegesis: The Case of Vienna, Österreichische Nationalbibliothek, lat. 940’, Sacris Erudiri 53 (2014): pp. 323 44, which considers the use of the Canon Tables in an unpublished Commentary on Matthew, a topic that is excluded from the present analysis. ¹² According to Bischoff, the Canon Tables are also discussed in an untitled exposition of Matthew, in the Commentary on Mark by ps Jerome, in an anonymous preface to the Gospel of Mark, and in an anonymous Commentary on Luke (Bischoff, ‘Turning Points’, pp. 115 16, 129, 132, 133; nos. 17 I, 27, 28, and 29 in his list). About the last text on Luke, Bischoff comments, ‘The fact that a section belongs to a certain canon is not only repeatedly alluded to but is in many cases criticised as being based only on an apparent similarity’. Neither the work on Matthew nor the Commentary on Luke has been published, though see the discussions of these texts in Mullins, ‘The Insular Reception’, pp. 130 66; Mullins, ‘The Case of Vienna, Österreichische Nationalbibliothek, lat. 940’. On the Matthean commentary, see also Anthony J. Forte, ‘Some Philological Observations on Codex Vindobonensis 940’, in The Scriptures and Early Medieval Ireland: Proceedings of the 1993 Conference of the Society for Hiberno Latin Studies on Early Irish Exegesis and Homiletics, ed. Thomas O’Loughlin (Turnhout: Brepols, 1999). The short preface to Mark, Praefacio secundum Marcum, was published in CCSL 108B, pp. 220 4. See the discussion of the Canon Tables on p. 221 under §7. The ps Jerome Commentary on Mark has been published in CCSL 82, and translated in Michael Cahill, The First Commentary on Mark: An Annotated Translation (Oxford: Oxford University Press, 1998). Cahill challenged the Irish provenance of the text but hesitated to suggest an alternative. Stancliffe, ‘Early “Irish” Biblical Exegesis’, pp. 365 6, also noted some problems with its supposed Irish character. The commentary

Scholarly Practices: the Hiberno-Latin Tradition

199

texts that are notable not only for their focus on the Canon Tables but also for the unusual genre of each in comparison with the earlier Latin exegetical tradition. My argument will be twofold. First, I suggest that these four texts are similar enough to one another that they form a distinct and identifiable tradition, belonging to the world of the Hiberno-Latin exegetical classroom. Secondly, I propose that one of the distinctive features of this tradition is its scholarly tenor. Augustine, as I argued in chapter 4, had previously made extensive use of the Eusebian apparatus in writing his De consensu evangelistarum, but he did so in service of his overall goal in this treatise and consequently did not think it necessary to alert his readers to his methods. In contrast, the figures I will consider in this chapter took an interest in the paratextual system itself, exploring its intricacies and refining the way in which it orders the textual material of the fourfold gospel. In other words, these authors seem to have had no immediate purpose in view in their discussions of Canon Tables akin to Augustine’s overtly apologetic aim in De consensu, but instead engaged in a more exploratory, open-ended analysis of the Eusebian apparatus and what it revealed about the fourfold gospel. In this respect, the Hiberno-Latin tradition, long noted for its magnificent artistic embellishment of Canon Tables in gospelbooks,¹³ also stands out as unique within the broader Christian world owing to its use of Eusebius’ system as a hermeneutical tool for reading the gospels.¹⁴

shows a marked interest in Canon Tables, and this feature could lend credibility to an argument for its Hiberno Latin identity, though it is inconclusive evidence on its own. In the prologue to the work, the place of Mark in the tables is discussed, and it is pointed out that this gospel has eighteen distinct passages. ¹³ For an introduction, see Nancy Netzer, ‘The Design and Decoration of Insular Gospel books and Other Liturgical Manuscripts, c. 600 c. 900’, in The Cambridge History of the Book in Britain, ed. Richard Gameson (Cambridge: Cambridge University Press, 2011), pp. 225 43. ¹⁴ Similarly Jennifer O’Reilly remarked, ‘The evident concern of exegetes to repeat and supplement material which is set out in the gospel concordance of the Eusebian canon tables and in Jerome’s brief explanation of them in the Novum opus is in some way mirrored in the elaborate decorative embellishment of these prefatory materials in insular gospel books’ (Jennifer O’Reilly, ‘Patristic and Insular Traditions of the Evangelists: Exegesis and Iconog raphy’, in Le Isole britanniche e Rome in éta romanobarbarica, ed. A. M. Luiselli Fadda and É. Ó Carragáin, Biblioteca di cultura romanobarbarica 1 (Rome: Herder, 1998), p. 70). On the Insular interest in quaternities expressing gospel harmony, see also Robert E. McNally, ‘The Evangelists in the Hiberno Latin Tradition’, in Festschrift Bernhard Bischoff, ed. Johanne Autenrieth and Franz Brunhölzl (Stuttgart: Anton Hiersemann, 1971), pp. 111 22; Malgorzata Krasnodebska D’Aughton, ‘The Four Symbols Page in Cracow Cathedral Library MS 140: An Image of Unity’, Peritia 12 (2000): pp. 323 41; Malgorzata Krasnodebska D’Aughton, ‘A Gemmarium for the Recognition of Precious Stones in the Cracow Chapter Library, MS 140: A Study on the Unity of Exegetical Themes’, in Listen, O Isle, unto Me: Studies in Medieval Word and Image in Honour of Jennifer O’Reilly, ed. Elizabeth Mullins and Diarmuid Scully (Cork: Cork University Press, 2011), pp. 54 7.

200

The Eusebian Canon Tables

AILERÁN O F CLONARD, CAN ON EVANGELIORUM The first text I wish to consider is also in some ways the most distinct of the four.¹⁵ In the seventh century the Irish scholar Ailerán of Clonard composed a poem titled Canon Evangeliorum, or ‘The Canon of the Gospels’, which survives in fourteen medieval manuscripts, the most famous being the Augsburg Gospels from the early eighth century (Augsburg, Universitätsbibliothek, Cod. I.2.4°.2; see fig. 29).¹⁶ About Ailerán himself little is known. He was a teacher in the monastery of Clonard, and, in addition to his poem on the gospel Canons, only one other work survives from his hand, a mystical and moral exposition of the Matthean genealogy of Jesus.¹⁷ In the sources he is often given the appellation sapiens, or ecnae, a word referring to the hierarchy of ecclesiastical scholars who functioned as scribes and formed ‘a uniquely Irish institution’.¹⁸ Aside from his association with Clonard, the only other known detail about his life is that he died from plague in 665. His poem on the Canon Tables has received notice primarily due to its reliance upon the well-known evangelist symbols as stand-ins for their human authors, which some have taken to imply that he was working from a gospelbook that included a copy of the Canon Tables decorated with the four

¹⁵ Though it should be noted that there were at least two other poems on the Canon Tables composed by other Latin authors, an anonymous one known as In primo certe canone and a composition by Alcuin titled Textum si cupies canonis (text of both in De Bruyne, p. 186). Both are discussed, along with Ailerán’s composition, at Houghton, The Latin New Testament, p. 202. On the former poem see also David Howlett, ‘Hiberno Latin Poems on the Eusebian Canons’, Peritia 21 (2010): pp. 166 71, who identified it as being Hiberno Latin. ¹⁶ In addition to the Augsburg Gospels, the poem survives in a number of other medieval manuscripts. Cf. David Howlett, ‘Seven Studies in Seventh Century Texts’, Peritia 10 (1996): pp. 11 12; David Howlett, ‘Further Manuscripts of Ailerán’s Canon Euangeliorum’, Peritia 15 (2001): pp. 22 6. ¹⁷ See the translation published in Aidan Breen, trans., Ailerani Interpretatio Mystica et Moralis Progenitorum Domini Iesu Christi, Celtic Studies (Dublin: Four Courts Press, 1995). ¹⁸ So Douglas Mac Lean, ‘Scribe as Artist, Not Monk: The Canon Tables of Ailerán ‘the Wise’ and the Book of Kells’, Peritia 17 18 (2003 2004): pp. 455 68. For the literature on Ailerán, see further Kelly, ‘A Catalogue of Early Medieval Hiberno Latin Biblical Commentaries (II)’, pp. 393 4; Howlett, ‘Seven Studies’, pp. 6 20; O’Reilly, ‘Patristic and Insular Traditions’, pp. 71 2, 80; Mullins, ‘The Insular Reception’, pp. 79 102; Barbet Massin, L’enluminure et le sacré, pp. 368 72; O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, pp. 101 8; Breen, Interpretatio Mystica, pp. 1 13. His authorship of this poem was first noted by Donatien De Bruyne, ‘Une poésie inconnue d’Aileran le Sage’, Revue Bénédictine 29 (1912): pp. 339 40. Howlett made extensive and complex arguments about the internal structure and diction of the poem that I find exceedingly speculative: Howlett, ‘Seven Studies’, 16 20; Howlett, ‘Hiberno Latin Poems’, pp. 162 6.

Scholarly Practices: the Hiberno-Latin Tradition

201

Fig. 29. Ailerán’s Canon Evangeliorum in the Augsburg Gospels (eighth c.). The manuscript was produced around 705 at the monastery of Echternach, a major centre of Irish Anglo Saxon culture on the continent. Ailerán’s poem is the first thing the reader encounters when opening the codex (fol. 1r is blank). The black numerals in the margin indicate the Canon referred to in each stanza and the red numeral indicates the number of parallels in that Canon. Augsburg, Universitätsbibliothek, Cod. I.2.4⁰.2, fol. 1v

heavenly creatures symbolizing the four gospels, a motif most famously exemplified in the Book of Kells (see fig. 30).¹⁹

¹⁹ Cf. Nancy Netzer, ‘The Origin of the Beast Canon Tables Reconsidered’, in The Book of Kells, ed. Felicity O’Mahony (Aldershot: Published for Trinity College Library, Dublin by Scolar Press, 1994), pp. 322 32; Patrick McGurk, ‘The Canon Tables in the Book of Kells’, in New Offerings, Ancient Treasures: Studies in Medieval Art for George Henderson, ed. Paul Binski and William Noel (Stroud: Sutton, 2001), pp. 54 5; Mac Lean, ‘Scribe as Artist, Not Monk’, pp. 437 55. Thomas O’Loughlin, however, was more hesitant to accept the argument (O’Lough lin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, pp. 102 3).

202

The Eusebian Canon Tables

Fig. 30. Canon II in The Book of Kells (c.800). The angelic figures above each column (a winged man, lion, and ox) have replaced the expected titles for the gospels of Matthew, Mark, and Luke. Ailerán’s poem may have been based on an earlier insular gospelbook containing a set of so called ‘beast Canon Tables’. Dublin, Trinity College Library, MS 58, fol. 2v (© The Board of Trinity College Dublin)

Scholarly Practices: the Hiberno-Latin Tradition

203

The poem is short enough that it is worth quoting it in full:²⁰ Ailerani Sapientis Canon Evangeliorum

The Canon of the Gospels of Ailerán the Wise

Quam in primo speciosa quadriga homo leo uitulus et aquila septuaginta unum per capitula de Domino conloquuntur paria

How in the first a beautiful four part team a man, a lion, a calf, and an eagle, through seventy one chapters together speak comparable things about the Lord!

In secundo subsequente protinus homo leo loquitur et uitulus quibus inest ordinate positus decem in se atque nouem numerus.

In the second, following directly after, there speaks a man, a lion, and a calf, among which there is placed in order ten [multiplied] into itself and the number nine.

Tum deinde tertio in ordine homo et bos loquitur cum uolucre numero in quo consistunt antiquae alphabeti Hebraeorum litterae.

Then thereafter in the third order a man and a cow speak with a winged creature in the number in which the ancient letters of the alphabet of the Hebrews consist.

Quarto loco fatentur aequalia una homo leo atque aquila uno ore loquentes capitula Verbi summi sena atque uicena.

In the fourth place there utter equal things at the same time a man, a lion, and an eagle, with one mouth speaking chapters six and twenty of the highest Word.

Quinta uice concordant in loquella homo prudens atque mitis hostia Iesu Christi emicantes agmina Iuda sine Saluatori credula.

In the fifth part there accord together in speech a prudent man and a gentle sacrificial victim, the band of Jesus Christ springing forth believing the Savior, without Judas.

Ecce sexto pari sonant clamore natus Adam cum clamoso leone conputata traditis pro munere sacerdotum oppidis in honore.

Behold in the sixth there sounds with comparable clamour a son of Adam with a clamorous lion things computed from towns handed over as a gift in honor of priests.

En loquuntur septies in septimo homo auis consona de Domino.

Behold, seven times in the seventh a man, a bird, speak consonant things about the Lord.

In octauo nunc leonis catulus Dei uerba profert atque uitulus quorum simul conputatur numerus adïecto Paulo apostolicus.

In the eighth now a lion’s cub brings forth the words of God, and a calf whose number is computed by adding Paul into the group of the apostles.

Nonus ordo in quo duo pariter conloquuntur uitulus et uolucer inspirati sensu spiritaliter proloquuntur ternum septempliciter.

The ninth order in which two speak together at the same time, a calf and a winged creature, inspired spiritually with sense they speak forth three times sevenfold.

Homo nempe uerbum profert proprium sexaginta et per duo numerum Rugientemque leonem audies solum sane decies et nouies Bouem solum fatentem inuenies uerba Dei bis et septuagies Subuolantem ad astra reperies nonagies loqui atque septies.

A man, doubtless, brings forth his own word through the number sixty and two, and the roaring lion you will hear but, certainly, ten and nine times, but the cow you will find uttering the words of God twice and seventy times, the one flying upwards towards the stars you will discover to speak ninety and seven times.

²⁰ The edition I have used is that published in Howlett, ‘Hiberno Latin Poems’, pp. 162 4, and the translation is largely also taken from Howlett as well (pp. 164 5), though with some modifications. I have also consulted the translation in O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, pp. 109 10.

204

The Eusebian Canon Tables

There are four aspects of this poem relevant to the reception history of the Eusebian apparatus. First, at its most basic level, Ailerán’s composition is a way of remembering the appropriate number of passages that occur in the Canon Tables, since each of the ten stanzas of the poem reports the number of parallels that appear in a corresponding table. Thus, for example, the first stanza notes that the first Canon has 71 parallels, the second stanza highlights the 109 parallels for the second Canon, and so on. Consisting essentially of several long lists of seemingly random numbers, Eusebius’ prefatory tables were ripe for scribal error, as plenty of surviving manuscripts attest. A slight mistake in the copying—skipping a single number, for example—would render the rest of the given table useless. Ailerán’s poem therefore provides a way for the scribe copying these lists to check that his finished product has the correct number of parallel passages shown in each table. In other words, whatever else its intent may have been, the poem at least serves a prophylactic purpose which testifies to a concern that accurate and so functional copies of the Canon Tables be produced.²¹ This concern for a text’s being properly received by subsequent readers is of course the function of a paratext, and we could then say that what Ailerán has created is a new paratext for the original Eusebian paratext. This suggests that Ailerán came from a scribal culture in which Canon Tables were now viewed as an essential component of the four-gospel codex, so much so that a senior scholar such as himself should create a new composition to facilitate the proper transmission of the Eusebian apparatus. Secondly, it is striking that Ailerán decided to comment upon the Canon Tables in the form of poetry rather than in some other genre. It is a choice that probably strikes the modern reader as peculiar, since something as mundane as a cross-referencing system or a numeric matrix hardly seems an appropriate subject for the creative expression afforded by poetry. One possible reason for the use of this form is that Ailerán probably intended for scribes-in-training to memorize the poem, and the rhythm and rhyming of poetry of course aid the memory.²² However, there is more than merely this utilitarian purpose at work, since the poem operates at multiple levels of signification. One signal that Ailerán is interested in more than merely providing an aide-memoire is the way in which he plays with numbers in the poem. Only in four of the stanzas (1, 4, 7, 10) are the numbers for each Canon reported straightforwardly. In two others (2, 9), the numbers are given as mathematical calculations.

²¹ On the practical utility of the poem, see also O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, pp. 105 7. Its mnemonic value is also recognized in Heidi C. Gearhart, ‘From Divine Word to Human Hand: Negotiating Sacred Text in a Medieval Gospel Book’, Word & Image 32 (2016): p. 435. ²² As recognized also by O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, p. 101.

Scholarly Practices: the Hiberno-Latin Tradition

205

Even more striking is that in the four remaining stanzas (3, 5, 6, 8) the numbers are not given directly at all, but instead are presented as puzzles requiring an expert knowledge to solve. So, for example, in the third stanza the poet refers to the number of letters in the Hebrew alphabet (22), a datum surely available only to the highly educated in seventh-century Ireland.²³ In stanza five, the solution to the riddle depends upon both biblical knowledge and mathematical calculation. Both Matthew and Luke report Jesus’ sending out of messengers to proclaim his good news, though they differ over the number of persons sent forth, with Matthew reporting only the twelve apostles and Luke referring to a much larger anonymous group of seventytwo (Mt 10:1–42 = Mt §79–100; Lk 10:1–24 = Lk §107–20). If one adds together these two groups as the complete ‘band of Jesus Christ springing forth’, and then subtracts Judas (presumably because he was a false apostle who did not believe), then one arrives at the number of parallels in Canon V (83). Notably this riddle depends upon knowledge of gospel parallels of precisely the sort that the Canon Tables make apparent. The riddle in the sixth stanza is even more arcane. The ‘towns handed over as a gift’ to the priests refers to the cities designated for the Levites when the Israelites took possession of the land, which according to Numbers 35:7 totalled forty-eight, the same number as there are parallels in Canon VI. Finally, the riddle in stanza eight is the easiest to solve, requiring knowledge of the number of apostles (12) with Paul added in for good measure to arrive at a total of thirteen.²⁴ The terms in which these riddles are cast reveal a scribal culture that placed a high value on scriptural expertise, indeed even upon what might be considered Bible trivia (see fig. 31). As such, they are in keeping with the close

²³ It is possible that Isidore served as Ailerán’s source here, since he discusses the Hebrew alphabet at etymol. 1.2.4. Cf. Michael Herren, ‘On the Earliest Irish Acquaintance With Isidore of Seville’, in Visigothic Spain: New Approaches, ed. Edward James (Oxford: Clarendon Press, 1980), pp. 243 50, who argued that Isidore’s writings were being read in Ireland by the middle of the seventh century, so within Ailerán’s lifetime, though Marina Smyth proposed that ‘one should be very cautious’ about positing any Isidorian influence before the late seventh century (Marina Smyth, ‘Isidorian Texts in Seventh Century Ireland’, in Isidore of Seville and His Reception in the Early Middle Ages: Transmitting and Trans forming Knowledge, ed. Andrew Fear and Jamie Wood, Late Antique and Early Medieval Iberia (Amsterdam: Amsterdam University Press, 2016), pp. 111 30, at p. 112). See also Robert E. McNally, ‘The “Tres Linguae Sacrae” in Early Irish Bible Exegesis’, Theological Studies 19 (1958): pp. 395 403. ²⁴ The numbers reported by Ailerán agree with the numbers in the Stuttgart Vulgate edition of the Canon Tables in all but four instances. The numbers for Canon I, IV, V, and XJn vary by one each. On the variations and their possible relation to certain gospel manuscripts, see Nancy Netzer, Cultural Interplay in the Eighth Century: The Trier Gospels and the Making of a Scriptorium At Echternach, Cambridge Studies in Palaeography and Codicology 3 (Cambridge: Cambridge University Press, 1994), p. 61; O’Loughlin, ‘The Eusebian Apparatus in the Lindis farne Gospels’, pp. 110 11.

206

The Eusebian Canon Tables

Fig. 31. A word square facing Ailerán’s poem in the Augsburg Gospels (eighth c.). If you begin in the centre of the square and read in a straight line, making 90 degree turns to move towards a corner of the square, the letters always spell euangelia ueritatis, no matter what path you take. The rows of letters, which can be read up, down, or sideways, mimic the columns and rows of the Canon Tables that follow (fol. 7r 12v, after the two standard Hieronymian prefaces, Plures fuisse and Novum opus) and prepare the reader to interpret the Eusebian numbers as revealing the ‘gospel of truth’. The puzzle like format of the word square also coincides with the riddling nature of Ailerán’s poem, which is written on the page that faces the square when the codex is open. Augsburg, Universitätsbibliothek, Cod. I.2.4⁰.2, fol. 2r

attention to the biblical text that the Canon Tables themselves are intended to foster. Moreover, in at least one other respect, the presentation of this information in poetic form might be related to the specific dynamics at work in the Eusebian apparatus. Recognition that the cross-referencing system functions on the basis of the principle that numbers can symbolically represent passages of text may have inspired Ailerán to experiment with further symbolic signification, perhaps on the assumption that numbers in scripture were not mere

Scholarly Practices: the Hiberno-Latin Tradition

207

incidental details but rather were significations of higher truths. On such a view, it is a short step to the notion that the numbers in each of the Canon Tables are also laden with an excess of meaning, and as such must resonate with similar numbers elsewhere in scripture not by coincidence but rather by divine intent. If so, then Ailerán’s poem might be expressing in textual form the same set of assumptions that justified the elaborate and intricate symbolic artwork with which insular scribes decorated Canon Tables in gospel manuscripts.²⁵ The third striking feature of Ailerán’s Canon Evangeliorum is the way in which it presents the Canon Tables as a series of successive speech acts not by the evangelists but rather by their heavenly authorizers or patrons,²⁶ the traditional beast symbols.²⁷ Eusebius himself, in his Letter to Carpianus, had referred to the text of the gospels as the evangelists ‘speaking similar things’ (τὰ παραπλήσια εἰρήκασιν), and Jerome had used similar language in his Novum opus (loca in quibus vel eadem vel vicina dixerunt),²⁸ so Ailerán’s discourse here is in keeping with those precedents.²⁹ However, by shifting the actor responsible for the speech to the heavenly patrons he calls attention to the ultimate divine origin of the gospels, which has the effect of projecting the irreducible diversity amongst the four into the heavenly realm, specifically the liturgical space around the throne of God. In contrast to Eusebius’ relatively restrained description of his apparatus in the Letter to Carpianus, which makes no grand claims for the divine origin of his system and only alludes to the evangelists’ inspiration once in rather vague terms (ἠνέχθησαν ϕιλαλήθως εἰπεῖν),³⁰ Ailerán has foregrounded the four creatures around the throne of God who provided the evangelists with their respective words, and whose divinely inspired harmony is now communicated through the ten tables (see fig. 32). This

²⁵ See, e.g., Heather Pulliam, ‘Painting By Numbers: The Art of the Canon Tables’, in The Lindisfarne Gospels: New Perspectives, ed. Richard Gameson (Leiden: Brill, 2017), pp. 112 33, who argued that the abstract decoration of the Canon Tables in the Lindisfarne Gospels ‘conveys meaning through both numeric and mathematical forms of expression, portraying divine perfection through measure and proportion’ (pp. 112 13). ²⁶ I use the term ‘patrons’ here under the influence of Watson, Gospel Writing, p. 567: ‘At much the same time as these images from the first half of the fifth century, the four cherubim acquire books and are henceforth rarely depicted without them. They become the heavenly patrons of the four evangelists. Yet they do not cease to be cherubim, the four living creatures who encircle the throne of Christ with their song of praise. The new images do not replace the cherubim with mere “symbols of the evangelists”, rather they represent the comprehensive liturgical context within which the fourfold gospel has its being’. ²⁷ Therefore, David Howlett was not quite correct when he asserted, ‘Ailerán presents the correspondences in more than competent narrative verse as conversations among the Evangel ists’ (Howlett, ‘Seven Studies’, p. 12). On the contrary, the evangelists are nowhere to be seen in the poem, an observation that could be related to the fact that early Irish gospel books, such as the Book of Durrow, seem not to have had evangelist portraits, but instead simply carried the evangelists’ symbols, unlike slightly later books such as the Lindisfarne Gospels and the Book of Kells that included both evangelist beast symbols and evangelist portraits. ²⁸ Eusebius, ep. Carp. (NA²⁸, pp. 89* 90*); Jerome, nov. op. (Biblia sacra iuxta vulgatam versionem, p. 1516). ²⁹ Cf. Mullins, ‘The Insular Reception’, pp. 92 8. ³⁰ ep. Carp. (NA²⁸, p. 89*).

208

The Eusebian Canon Tables

Fig. 32. Christ in Majesty in Livre d’Evangiles de l’abbaye Sainte Croix (eighth c.). Ailerán’s poem occurs on fol. 26r v, following a series of Canon Tables on fol. 17v 25v. After several prefatory texts, the maiestas image then occurs and next the text of Matthew begins. Note that in the image the heavenly beasts surround Christ, just as in Ailerán’s poem they together speak forth the gospel. The poem thus works together with the image to highlight the relation among the individual evangelists, their respective gospels, and the diverse but harmonious message they proclaim. I am grateful to Lynley Herbert for drawing my attention to this image. Poitiers, Médiathèque François-Mitterrand, MS 17 (65), fol. 31r

Scholarly Practices: the Hiberno-Latin Tradition

209

theological assumption about the inspired origin of the tables is probably what justified Ailerán’s search for numerical resonances across the canon considered previously.³¹ Finally, in a further layer of symbolic signification, Ailerán also uses these traditional evangelist beasts as metaphors for Christ himself.³² To some degree this move had been anticipated in the preface to Jerome’s Commentary on Matthew, which was often excerpted and used as a preface for medieval gospelbooks (known by its opening phrase Plures fuisse), since there he had associated each of the beasts with the distinctive beginning of its respective gospel. Gregory the Great later carried forward this line of interpretation by ‘show[ing] how each gospel, as epitomised by its opening lines and characterised by its symbolic beast, reveals a particular aspect of the redemption of humanity wrought by Christ’.³³ Like Gregory, Ailerán playfully explores how each of the creatures gives a particular perspective on Christ, though parallels between his poem and Jerome are also evident. As suggested above, in this respect the poem reflects in textual form the same imaginative world that gave rise to the artwork that typically filled the very same insular gospelbooks in which Ailerán’s poem was copied. It is now widely accepted that insular artwork functioned in a polyvalent manner,³⁴ and Ailerán’s poetry clearly was intended to be read similarly.

³¹ I am grateful to Clift Ward for suggesting this point to me. ³² Mullins, ‘The Insular Reception’, pp. 86 92, also saw multiple levels of meaning in the poem and highlighted the way in which the four evangelist symbols function Christologically. ³³ O’Reilly, ‘Patristic and Insular Traditions’, pp. 57 8. This tendency in Hiberno Latin sources had previously been noted by McNally, ‘The Evangelists in the Hiberno Latin Tradition’, p. 116. ³⁴ Michelle Brown referred to ‘the fundamental tenet that multivalent reading is a key feature of Insular art’ (Michelle Brown, ‘Embodying Exegesis: Depicting Evangelists in “In sular” Manuscripts’, in Le Isole britanniche e Rome in éta romanobarbarica, ed. A. M. Luiselli Fadda and É. Ó Carragáin, Biblioteca di cultura romanobarbarica 1 (Roma: Herder, 1998), p. 124). So also Jennifer O’Reilly, who noted that ‘it seems unlikely’ that the temple image in the Book of Kells ‘was intended as a static and didactic representation of theological or ecclesiological truths, or that it was intended to be read, in a manner more suited to some later medieval works, as an onion layered series of fixed meanings denoting successively the body of Christ, of the individual faithful and of the Church. Patristic exegesis, in contrast, is concerned to show the simultaneity of such truths as aspects of a single truth and to show the continuing significance of the spiritual interpretation of the literal text for the present reader who is provided with images for meditatio and imitatio’ (Jennifer O’Reilly, ‘Exegesis and the Book of Kells: The Lukan Genealogy’, in Scriptural Interpretation in the Fathers: Letter and Spirit, ed. Thomas Finan and Vincent Twomey (Dublin: Four Courts Press, 1995), p. 353). O’Reilly further suggested that the image of the evangelist John in the Book of Kells ‘is shown without an identifying Evangelist symbol, so that the hieratic figure more readily evokes associations with Christ himself ’ (O’Reilly, ‘St. John the Evangelist’, p. 193). Similarly, Michelle P. Brown argued that in the evangelist portraits of the Lindisfarne Gospels, two evangelists are depicted youthfully to connote the immortality of Christ (Mark and John), and two are shown aged to evoke his mortality (Matthew and Luke) (Brown, The Lindisfarne Gospels, p. 349).

210

The Eusebian Canon Tables

This aspect of the poem is signalled by the variety of descriptions the poet uses for these creatures and begins to become apparent in the fourth stanza. Throughout the composition several synonyms are used in referring to agreements between the gospels’ heavenly patrons. In stanza one the creatures ‘speak together comparable things’ (conloquuntur paria), in stanza five they ‘accord together in speech’ (concordant in loquella), in stanza six they ‘sound with comparable clamour’ (pari sonant clamore), in stanza seven they ‘speak consonant things’ (loquuntur . . . consona), and in stanza nine they ‘comparably speak together’ (pariter conloquuntur). However, in stanza four a unique word is used to express this point. Ailerán in this instance says that the patrons for Matthew, Mark, and John ‘utter equal things’ (fatentur aequalia). The term for ‘equal things’, aequalia, is very likely a play on words with the term for eagle, aquila, which is the symbol of the fourth gospel, and which follows in the next line of this stanza. In the light of the puzzle-like intent of the poem, it is probably not a coincidence that in this, the fourth stanza, Ailerán has chosen to use a word that specifically calls to mind the fourth gospel. This particular emphasis on the fourth gospel is also apparent in the last line of this stanza, in which the poet refers to the ‘chapters of the highest word’ (capitula Verbi summi).³⁵ The fourth gospel, of course, stands apart from the previous three with its unambiguous ascription of deity to Jesus and its description of him as the ‘Word’. A long patristic tradition going back to Clement of Alexandria had recognized the uniqueness of the fourth gospel, and often expressed its significance through the use of superlative adjectives. Jerome, for example, in the Plures fuisse text, spoke of the fourth evangelist as drinking from the ‘purest streams of doctrine’ (purissima doctrinarum fluenta potauit).³⁶ In the light of this precedent, to the attentive and informed reader of Ailerán’s poem, this reference to the summum Verbum in the fourth stanza would undoubtedly have called to mind the fourth gospel’s witness to the Word that was in the beginning with God.

³⁵ The use of the term capitula to refer to the Ammonian/Eusebian sections follows the precedent of Jerome, who used it in this sense in nov. op. (Biblia sacra iuxta vulgatam versionem, p. 1516). Jerome, in turn, was simply translating Eusebius’ term κεϕάλαια (ep. Carp. (NA²⁸, p. 90*)) with its closest Latin equivalent. Thomas O’Loughlin thought that Ailerán’s use of the term is a ‘weakness’ of the poem and may represent confusion on the part of its author, since the same word was used for chapter like divisions that demarcated the text of the gospels according to sense units (O’Loughlin, ‘The Eusebian Apparatus in the Lindisfarne Gospels’, p. 106). O’Loughlin’s comment implies that capitulum at this period referred only to such chapter headings, but it is not clear that it had such a restricted meaning, so we should not suspect confusion on the part of Ailerán. ³⁶ Jerome, comm. Mt., praef. (SC 242, p. 62). O’Reilly, ‘Patristic and Insular Traditions’, pp. 52 3, asserted that the Plures fuisse is ‘the single most influential patristic source’ for insular traditions of the evangelists, and Ailerán’s poem certainly serves as corroborating evidence for her claim.

Scholarly Practices: the Hiberno-Latin Tradition

211

In contrast to the emphasis on the deity of Christ that is apparent in stanza four, stanzas five and six use the heavenly patrons to lay stress on his humanity. Matthew’s symbol of the man is now specified as a homo prudens, a prudent man. Multiple associations here are possible. Matthew’s gospel famously records the Sermon on the Mount, by virtue of which Jesus may be regarded as a prudent or wise man.³⁷ Moreover, Jesus himself is described with the word prudens in the parable of the master who leaves for a time and returns either to reward or to punish his servants for their actions while he was away. In Matthew’s version of this story, it is the servant who is said to be ‘faithful and wise’ (fidelis servus et prudens) (Mt 24:45 = Mt §265), but in Luke’s version it is the master himself, that is, Christ, who is so described (fidelis dispensator et prudens) (Lk 12:42 = Lk §157), a parallel that is included in the very Canon to which this stanza refers. Furthermore, in stanza five Ailerán seems to have also exploited Luke’s beast symbol, the calf, in order to draw the reader’s attention to Jesus. It is not simply a calf that he speaks of here, but specifically a mitis hostia, a ‘gentle sacrificial victim’, which is no doubt an allusion to the crucifixion. In addition, it is possible that the description of the calf as mitis, gentle, also is intended to echo descriptions of Jesus from the gospels, since in the gospel of Matthew Jesus describes himself as ‘gentle and humble of heart’ (mitis sum et humilis corde) (Mt 11:29 = Mt §113). This emphasis on the humanity of Jesus that we have observed in stanza five carries over into stanza six. Now Matthew’s symbol is not simply a man, but specifically a ‘son of Adam’. When hearing this phrase someone familiar with the gospels could not help but think of the Lukan genealogy which traces Jesus’ lineage back to Adam and presents him as a new start to the human race.³⁸ The reference to Christ is somewhat less straightforward when it comes to the Markan lion in stanza six, but is probably still present. Ailerán describes Mark’s heavenly patron as a ‘clamorous lion’ (clamoso leone), and again in stanza ten he calls it a ‘roaring lion’ (rugentiem leonem). Once again, these descriptions echo those of Jerome in his Plures fuisse. The earlier Latin exegete had noted that in Mark’s gospel one hears a ‘lion roaring in the wilderness’ (vox leonis in heremo rugientis), a description that he then explained with reference to the Isaianic prophecy that opens Mark’s gospel, which presents John the Baptist as ‘the voice of one crying in the desert’ (uox clamantis in deserto).³⁹ Hence, in Jerome’s usage the lion functions primarily as an allusion to the preparatory mission of John, rather than directly as a statement about ³⁷ Mullins, ‘The Insular Reception’, p. 88, pointed out that other Hiberno Latin texts, including the Irish Reference Bible, also associate Matthew with prudentia. ³⁸ Mullins, ‘The Insular Reception’, pp. 88 9, likewise highlighted the theme of Christ as the second Adam, a notion that in the insular context came to have cosmological associations as well, since the letters ADAM form an acrostic using the names of the four cardinal directions. ³⁹ Jerome, comm. Mt., praef. (SC 242, p. 64).

212

The Eusebian Canon Tables

Jesus, and this might be what Ailerán has in mind in stanza six. However, it is clear that he also intends for Mark’s lion to be viewed as a reference to Jesus, since in stanza eight he refers to Mark’s symbol as a ‘lion’s cub’ (leonis catulus), which is undoubtedly a reference to Genesis 49:9, a passage whose description of a future king from Judah as a ‘lion’s cub’ (catulus leonis) was widely read by patristic authors as a prophecy of Christ, including in Jerome’s Hebrew Questions on Genesis.⁴⁰ Therefore, Ailerán’s reference to Mark’s heavenly patron as not just a leo, but a catulus leonis, indicates that, as with the symbols of Matthew and Luke, so also with Mark’s, he is drawing his reader’s attention to aspects of Christ that may be observed in the four heavenly creatures, in this instance specifically his fulfilment of Old Testament prophecy and probably also his kingly rule. Stanza ten recapitulates many of the themes that we have observed thus far and can even be read as providing a narrative summary of the life of Christ. Matthew’s homo, ‘man’, recalls Jesus’ human birth; Mark’s rugientem leonem, ‘roaring lion’, suggests Jesus’ fulfilment of Hebrew prophecy as the king who came from Judah; Luke’s bouem, ‘cow’, once again recalls the sacrifice of Jesus as the fulfilment of Israel’s cult; and, finally, John’s subvolantem ad astra, ‘one flying upwards to the stars’, implies Jesus’ resurrection and ascension, and, more generally, his deity, which these events demonstrated.⁴¹ Here, more clearly than anywhere else in the poem, these heavenly creatures can be read as symbols not only of the individual gospels, but also of the Christ about whom those gospels speak. The hints in this direction in the earlier stanzas have prepared the attentive reader to see the deeper significance laid out in sequence here at the end. In the light of the multiple layers of meaning evident in this poem, it is worth considering briefly the sense of its title, particularly the fact that it alludes not to canones in the plural, as had Jerome in the Novum opus, but rather to canon in the singular.⁴² The use of the singular suggests that the canon of the title refers not to the ten tables created by Eusebius, but to some other canon that performs a similar function for the Canon Tables, a ‘canon’ for the Canon Tables as it were. Canon in the title must, therefore, not mean ‘table’ or ‘list’, as Eusebius had originally used the word κανών in the Letter to

⁴⁰ Jerome, QHG 49:8 9 (Hayward, p. 84). So also Mullins, ‘The Insular Reception’, p. 89. Moreover, as pointed out by Robert McNally, the Gelasian Sacramentary states that when the Gospel of Mark is read to those going to be baptized, the Genesis prophecy is to be cited (McNally, ‘The Evangelists in the Hiberno Latin Tradition’, p. 117). Cf. sac. Gel. XXXIIII, 305 (Mohlberg, p. 47). ⁴¹ Mullins, ‘The Insular Reception’, p. 92, highlighted a parallel between this description of John and a passage from the Carmen Paschale (PL 19, col. 591): More volans aquilae, verbo petit astra Joannes. ⁴² Mullins, ‘The Insular Reception’, pp. 80 3, also noted the usage of canon in the singular in the poem’s title.

Scholarly Practices: the Hiberno-Latin Tradition

213

Carpianus, but instead likely means something more like ‘guiding principle’ or ‘rule’.⁴³ In this sense the poem itself functions like a canon, since it serves as a plumb line for the scribe engaged in copying the ten tables. However, given the way in which the evangelist symbols function in the poem as polyvalent representations of Christ, the title of the poem more allusively also calls attention to the true canon of the gospels, namely Jesus Christ himself, the one who, though eternally existing as the highest Word, was born as a son of Adam, fulfilled prophecy as the lion of Judah, was killed like a gentle sacrifice, and finally ascended into heaven, flying to the stars like an eagle. It is this singular canon of the gospels that is instantiated in the ten individual canones which provide the key to unlocking that diverse, yet unified message. One of the most remarkable things about Ailerán’s poem is how starkly original it appears when seen against the prior reception history of the Eusebian apparatus in Latin. One goes from silent use of the Canon Tables by Augustine, to brief passing comments on the apparatus by Victor, Cassiodorus, and Isidore, and then finally to a poem exhibiting such a high level of sophistication and creativity in its exploration of the meaning and significance of the prefatory tables. This truly marks a new approach to the Eusebian apparatus that emerged in seventh-century Ireland in the midst of a scribal culture that, in this respect at least, was wholly unique at the time. Though Ailerán’s poem is probably the earliest of the four texts I will consider in this chapter, the next three sources provide further glimpses into the distinct scholarly context that must have already existed in some form when he composed his Canon Evangeliorum.

Pauca de Libris Catholicorum Scriptorum in Evangelia Excerpta The next text is of an entirely different genre than that of Ailerán’s poem, but it too stands out against the earlier Latin tradition, and also serves as a precursor to the final two texts to be considered. The manuscript Clm 6235, held in Munich and dated to the middle of the ninth century, contains a short work carrying the title Pauca de libris catholicorum scriptorum in evangelia excerpta. Distinctive Irish features in the manuscript, including some Old Irish glosses, identify the texts within it as belonging to the Hiberno-Latin tradition, and Robert McNally, who has edited the anonymous work, proposed that Pauca de libris may date to as early as 750–75.⁴⁴ The content of this short text consists of a thorough, almost scholastic, examination of the euangelia in terms of ⁴³ So also the Irish Reference Bible, a text which will be discussed below: ‘quaeritur cui lingua canon. Isidorus dicit: canon aut greci dicitur latine regula’ (Paris MS lat. 11561, fol. 130v); citing Isidore, etymol. 6.16.1. ⁴⁴ McNally, Scriptores Hiberniae Minores, Pars I, p. 209.

214

The Eusebian Canon Tables

fourteen categories: locus, tempus, persona, lingua, regula, ordo, auctoritas, causa, figura, demonstratio, conventio, qualitas, numerus, and documenta. Such categorization in some form no doubt goes back to classical and late antique grammatical and rhetorical training, but the consistent use of such categories and the increased number of them are distinctive Irish features.⁴⁵ Given its compressed and systematic nature, as well as the use of the questionand-answer format, McNally plausibly suggested that Pauca de libris represents ‘the digest of a lecture’, in keeping with the school-like nature of other Irish biblical expositions.⁴⁶ The author, or, perhaps more accurately, lecturer, considers the issue of the Canon Tables under the heading De numero euangeliorum.⁴⁷ The analysis under this category is almost entirely taken up with the Eusebian apparatus, and the whole of it accounts for approximately a fourth of the entire text. The teacher first points out that there are six relevant numbers for the gospel, including the number of gospels, the number of Canons, the number of capitula within each gospel, the number of brevia causarum, and the number of lines in each gospel.⁴⁸ Following this list he then offers a series of samples from each Canon: • Canon I: • Canon II:

• Canon III: • Canon IV:

• Canon V: • Canon VI:

Non est propheta sine honere nisi in patria sua (Mt 13:57; Mk 6:4; Lk 4:24; and Jn 4:44 = Mt §142; Mk §51; Lk §21; Jn §35) Amen dico, sunt quidam de hic stantibus qui non gustabunt mortem, reliqua (Mt 16:28–17:9; Mk 9:1–9; Lk 9:27–36 = Mt §172; Mk §87; Lk §98) Non est discipulus super magistrum (Mt 10:24–25a; Lk 6:40; Jn 13:16–17/15:20a = Mt §90; Lk §58; Jn §118, 139) Omnia quaecumque petieritis in oratione credentes accipietis (Mt 21:22; Mk 11:24; Jn 14:13–21a; 15:7, 16b; 16:23b–24 = Mt §216; Mk §125; Jn §128, 133, 137, 150) Non sum missus nisi ad oues domus israel quae perierunt (Mt 15:24; Lk 19:10 = Mt §158; Lk §226) quecumque ligaueritis super terram, et reliqua (Mt 18:18; Jn 20:23 = Mt §185; Jn §215)

⁴⁵ See the similar categories listed in various texts in McNally, Scriptores Hiberniae Minores, Pars I, p. 210. So also O’Reilly, ‘Patristic and Insular Traditions’, p. 69. ⁴⁶ McNally, Scriptores Hiberniae Minores, Pars I, p. 211. ⁴⁷ For what follows see Pauc. lib. cath. §21 3 (CCSL 108B, pp. 218 19), on which see also Mullins, ‘The Insular Reception’, pp. 103 29. On the Irish interest in numbers, see also Charles D. Wright, ‘The “Enumerative Style” in Ireland and Anglo Saxon England’, in The Irish Tradition in Old English Literature, Cambridge Studies in Anglo Saxon England 6 (Cambridge: Cambridge University Press, 1993), pp. 49 105. ⁴⁸ The sixth and final item in the series is obscure: Numerus concentuum iiii uel iii uel ii uel propria uniuscuiusque, quae non habentur in aliis, in Matheo lvi, in Marco xx, in Luca liiii.

Scholarly Practices: the Hiberno-Latin Tradition

215

• Canon VII: Postquam crucifixerunt eum, diuiserunt sibi uestimenta eius (Mt 27:35–36; Mk 15:24; Lk 23:34b–35a; Jn 19:23–24b = Mt §334; Mk §212; Lk §321; Jn §201) • Canon VIII: Discendit iesus in capharnaum in ciuitate galilea ibi que docebat eos sabbatis (Lk 4:31; Mk 1:21 = Lk §23; Mk §12) • Canon IX: Intrauit satanas in iudam qui nominatur scarioth unus de xii (Lk 22:3; Jn 13:2, 26b–27a = Lk §262; Jn §113, 124) Mt • Canon X : Beati mites, reliqua (Mt 5:5 = Mt §26) • Canon XMk: Seorsum autem deserebat discipulis suis omnia (Mk 4:34b = Mk §46) Lk • Canon X : Exi a me, quia homo peccator sum, domine (Lk 5:8–10a = Lk §31) Jn • Canon X : Si quis mihi ministrauerit, me sequatur (Jn 12:26 = Jn §106) After the list of sample passages, the author concludes his section on the Canon Tables by noting briefly their divisio, that is, which gospels are included in each table, and cites a passage from Isidore’s Etymologiae regarding the origin of the system in the work of Ammonius of Alexandria and Eusebius.⁴⁹ The astute reader will have noticed that in two instances our author gives incorrect examples. For Canon VI, which lists parallels in Matthew and Mark, he has provided a set of passages in Matthew and John, which should properly be in Canon VII. Moreover, for Canon VII he has listed a passage common to all four gospels, which should therefore be in Canon I. These kinds of mistakes are of the sort one might expect in a lecture, and should not be taken to indicate someone who was merely a novice in the use of the apparatus. On the contrary, these examples demonstrate someone actually using the apparatus as Eusebius intended, namely, as a reading tool. The samples are not drawn merely from the first set of parallels in each Canon, but rather are taken from throughout the fourfold gospel, implying that this is a person who has made a thorough exploration of Eusebius’ intertextual links. Moreover, this short text evinces a subtle shift in the use of the apparatus in comparison with Eusebius’ own description of the purpose of his paratext. In his Letter to Carpianus providing instructions for the use of the system, Eusebius envisioned a reader who was reading linearly through one of the gospels, who then became interested in whether or not a given passage appeared in other gospels as well. In other words, Eusebius imagined someone starting from the text of one gospel and going to another, in the context of a normal sequential reading. In contrast, the author of this short text has begun to use each of the Canons as a category of passages which can then be ⁴⁹ Elizabeth Mullins likewise observed that Hiberno Latin authors drew upon Isidore as a source for understanding the Canon Tables (Mullins, ‘The Insular Reception’, p. 32).

216

The Eusebian Canon Tables

investigated on their own terms. It is a more scholastic, perhaps even encyclopedic exploration of the relationships that exist among the gospels, one that is further removed from a sequential reading of any of the four. The short text Pauca de libris represents the earliest surviving evidence of this kind of engagement with the parallels presented in the Canon Tables, and as such marks another important milestone in the reception history of the Eusebian apparatus.

The Irish Reference Bible The next of the four texts I wish to consider is, like the last one, formally anonymous, and it carries the title Pauca problesmata de enigmatibus ex tomis canonicis, but has been known, since Bischoff ’s seminal study, as the Irish Reference Bible.⁵⁰ This work consists essentially of a commentary on the entire Bible, covering select topics on various books, and thus far only the portion pertaining to the Pentateuch has been published in a critical edition.⁵¹ In what follows I therefore rely on my own observation of one of the most important manuscripts preserving this text, Paris, Bibliothèque nationale, MS lat. 11561, dated by Bischoff to the middle of the ninth century.⁵² In the section on the gospels, the Reference Bible repeats much of the material contained in Pauca de libris, and, like that text, it too ‘has a rather schoolroom air’.⁵³ MacGinty, who has edited the section on the Pentateuch, follows Bischoff ’s dating of this work to around 750, though he suggests that an earlier date is possible, and he also agrees with Bischoff on the Irish origin, or at least Irish influence, upon the work.⁵⁴ We may be fairly confident, then, that we have here a further ⁵⁰ See Bischoff, ‘Turning Points’, p. 88, who pointed out that it ‘in a certain measure brings the older Irish commentary literature to an end’ and that it presents ‘a new encyclopaedic tendency’. ⁵¹ G. MacGinty, The Reference Bible * Das Biblewerk: Pauca Problesmata de Enigmatibus ex Tomis Canonicis: Praefatio et Libri de Pentateucho Moysi, CCCM 173 (Turnhout: Brepols, 2000). ⁵² Bischoff, ‘Turning Points’, p. 97. Cf. MacGinty, The Reference Bible, XIV XV, who agreed in the dating. Two other witnesses to the text used by MacGinty for his edition of the Pentateuchal section are Vatican, Apostolic Library, Vat. Reg. lat. 76, fols. 1 106; and Munich, Bayerische Staatsbibliothek, Clm 14276 + 14277. The entire manuscript Paris BNF lat. 11561 has been digitized and may be viewed online at http://gallica.bnf.fr/ark:/12148/btv1b90668240 [accessed on 29 January 2014]. ⁵³ MacGinty, The Reference Bible, IX. Bischoff, ‘Turning Points’, p. 111, also highlighted the relationship between Pauca de libris and The Reference Bible. ⁵⁴ MacGinty, The Reference Bible, X. On the Reference Bible, see further Joseph Kelly, ‘Das Bibelwerk: Organisation and Quellenanalyse of the New Testament Section’, in Irland und die Christenheit: Biblestudien und Mission = Ireland and Christendom: The Bible and Missions, ed. Próinséas Ní Chatháin and Michael Richter (Stuttgart: Klett Cotta, 1987), pp. 113 23; Martin McNamara, ‘Plan and Source Analysis of Das Biblewerk, Old Testament’, in Irland und die Christenheit, pp. 84 112; G. MacGinty, ‘The Pentateuch of the Reference Bible: The Problem Concerning Its Sources’, in The Scriptures and Early Medieval Ireland: Proceedings of the 1993

Scholarly Practices: the Hiberno-Latin Tradition

217

example of Irish or Irish-influenced scholars engaging in biblical exegesis using Eusebius’ paratextual system. In Paris MS lat. 11561 the prefatory material related to the gospels covers fol. 126v-137r, and within this section the Canon Tables are the topic of discussion from 130v through 131v, covering six columns of text, by far the most extensive discussion of Eusebius’ system up to this point in history. Just before this section, in §9 as it is noted in the manuscript (fol. 130r), the author begins to consider the category of numerus as it relates to the gospel, and under this rubric he provides the same six items as in the Pauca de libris. Then, starting on fol. 130v, noted in the manuscript as §10, the discussion turns to De numero canonum, and here the author goes well beyond Pauca de libris.⁵⁵ He first highlights the number of capitula in each gospel, referring thereby to the individual sections into which Eusebius divided the gospels. He also answers the question of why some capitula are shorter and others longer. Since the section divisions do not always follow natural breaks in the narrative of each gospel, this question would surely have been one of the first to arise when someone was learning how the apparatus functioned. In reply the author points out that the magnitudo capitulorum is measured according to the size of the agreements among multiple gospels or by the distinct elements within each gospel (brevitas concentuum . . . vel unius cuiusque proprium). The text then goes on to define canon by quoting Isidore, who traced it back to Greek and equated it with regula in Latin. Isidore elaborated with several further explanations of how regula derives its name: because it is that which ‘draws in a straight line’ (recte ducit), or because it ‘rules’ (regat), or because it ‘provides a norm of living correctly’ (normam recte vivendi praebeat), or finally because it ‘corrects anything twisted and crooked’ (distortum pravum quid corregat).⁵⁶ Applying this definition of canon to the subject matter at hand, the author of the Reference Bible asserts that the Canon Tables are designed to make plain the ‘harmony’ (concentus) of the gospels, and to correct the ‘corrupted mixture’ (pravitatem commixtionis) that resulted from the work of inexperienced scribes (apud inperitos). A citation from Jerome’s Novum opus follows to support the latter point, noting that the Canons remove the error confussionis by showing the proper place for each passage in the gospels. Next comes a further citation from Isidore, the same passage that was also included, albeit more briefly, in the Pauca de libris to explain the source of the Eusebian apparatus in the work of Ammonius.

Conference of the Society for Hiberno Latin Studies on Early Irish Exegesis and Homiletics, ed. Thomas O’Loughlin (Turnhout: Brepols, 1999), pp. 163 77. ⁵⁵ On this section see also Mullins, ‘The Insular Reception’, pp. 103 29. ⁵⁶ Citing Isidore, etymol. 6.16.1.

218

The Eusebian Canon Tables

After this introductory material the author moves into the content of the Canons, listing the number of parallels in each one, and giving examples of each. The examples for Canons III, IV, V, VIII, IX, XMt, and XLk are identical to those given in Pauca de libris, while original passages are provided for Canons I, II, VI, XMk, and XJn.⁵⁷ Moreover, the parallel that the author of Pauca de libris had incorrectly placed in Canon VI is, in the Reference Bible, correctly placed into Canon VII. Next, the author considers the divisio of the tables, noting that there are four, since either all four evangelists agree, or three, or two, or a single evangelist speaks alone. Then, the tables are considered according to their modi and ordines. The former comprises three modi, but it is rather unclear what the author has in mind with this category, while the latter is subdivided into two elements: the ordo descensionis and the ordo discretionis. The ordo descensionis refers to the fact that the tables begin with a collection of four parallels, then collections of three, then two, and finally distinct passages without parallels. Once again, it is difficult to discern what the author means by the ordo discretionis, owing to the compressed, outline-like format of the text. In the final section of the work devoted to the Canon Tables, also the longest and most exegetically interesting portion, the author considers the qualitates of the Canons. For the author, the qualitates refer to the types of gospel parallels presented in the tables. First, there are passages that are the ‘same’ (eadem), then passages that are not the same but nonetheless ‘similar’ (vicina), and finally passages which are ‘unique’ (sola). As the author uses the terms, eadem and vicina seem to sit on a continuum according to degree of correspondence, with eadem being nearly exact or identical, and vicina being something rather looser. Furthermore, our author notes that both of these categories may be further subdivided according to sensus and superficies. Bischoff pointed out that in Hiberno-Latin texts, sensus often was contrasted with historia and comprised the moral and allegorical sense of a text.⁵⁸ The term as it is used in the Reference Bible seems to have a broader meaning and is not contrasted with history but is instead a way of referring to the ‘meaning’ of a passage of text, whether understood in historical or theological terms. In contrast, superficies seems to refer to something more like the ‘surface’ of a text, in the sense of its actual wording. In short, the distinction seems to be between what a text says (superficies) and what it means (sensus). ⁵⁷ The examples given that are not identical to those stated above for Pauca de libris are: Canon I: Mt 3:3 = Mt §8; Mk 1:3 = Mk §2; Lk 3:3 6 = Lk §7; Jn 1:23 = Jn §10 Canon II: Mt 4:1 = Mt §15; Mk 1:12 13b = Mk §6; Lk 4:1 2a = Lk §15 Canon VI: Mt 18:8 = Mt §180; Mk 9:43 7 = Mk §100 Canon XMk: Mk 1:45 = Mk §19 Canon XJn: Jn 1:28 = Jn §13 ⁵⁸ Bischoff, ‘Turning Points’, 87. On the distinction between these terms, see further Mullins, ‘The Insular Reception’, pp. 159 66, who drew attention to their use in earlier patristic sources, in Bede, and in Irish grammatical texts.

Scholarly Practices: the Hiberno-Latin Tradition

219

Thus, for example, passages can be ‘the same’ (eadem) in two ways (modis duabis): either by both sensus and superficies, or by sensus alone or superficies alone. As an example of a passage that is the same by both sensus and superficies, the author adduces Jesus’ saying about a prophet being without honour in his hometown, recorded in Canon I (Mt 13:57=Mt §142; Lk 4:24 = Lk §21; Jn 4:44 = Jn §35).⁵⁹ With only minor variations, the wording of these passages is nearly identical, the historical event at which they were spoken is presumably the same, and the underlying meaning is the same as well. Hence they agree in both sensus and superficies. Next come those things that are ‘the same’ on the level of ‘sense’ (eadem sensu) but ‘different’ with respect to the ‘surface’ (dissimilia superficia).⁶⁰ To illustrate this category the author selects one of the Eusebian parallels I discussed in chapter 3, namely the passages in Luke and John about Peter’s restoration and commission. On the night before his crucifixion, Luke’s Jesus predicted Peter’s restoration following his denial and commanded him to care for his brethren (tu aliquando conversus confirma fratres tuos) (Lk 22:32b=Lk §274). This passage has no parallel in the other two Synoptics, but Eusebius insightfully recognized that the fourth gospel has a post-resurrection scene in which Christ also restores and commissions Peter, using a thrice-repeated charge to care for the flock (si diligis me pasce oves meas) (Jn 21:15c, 16c, 17c = Jn §227, 229, 231). The author of the Reference Bible cites these two passages from Canon IX to highlight the fact that they do not exhibit verbatim agreement, nor do they refer to the same historical event, so they are not the same on the ‘surface’ level. However, on a deeper level they refer to the same reality, namely the restoration of Peter and the dominical charge to shepherd the church, so they have the same sensus.⁶¹ The ‘neighbouring’ (vicina) passages receive a similar analysis, since they too can be related ‘in two ways’ (modis), either by sensus or by superficies, and examples of each are again provided. As a parallel that is similar (similia) by superficies the author adduces Luke 24:41–3 (= Lk §341) and John 21:13 (= Jn §225). In the Lukan passage the risen Jesus eats fish and honeycomb before sharing it with his disciples in the upper room, while in the Johannine parallel the resurrected Christ on the shore of the Sea of Galilee offers bread and fish to his disciples. The author comments that these two texts are ‘only similar on ⁵⁹ The author does not list the parallel passage that occurs in Mk 6:4 6a = Mk §51, presumably because the other three were sufficient to make his point. ⁶⁰ The manuscript reads desimilia, which I have corrected to dissimilia. ⁶¹ The text mistakenly says this is a Canon VIII parallel, either by slip of the memory or from scribal error. Moreover, further textual corruption in this passage is evident. The manuscript reads haec dissimilia superficia sunt sed eadem sensu et idcirco uno id est VIII canon non iunguntur. That is, the text says that these two passages are not joined together in Canon VIII, which is beside the point. I suspect that the non iunguntur is meant to be coniunguntur, since the latter verb is used in the following paragraph.

220

The Eusebian Canon Tables

the surface’ (superficie tantum similia) yet ‘dissimilar by sense’ (dissimilia a sensu), since they report neither ‘the same historical matters’ (eadem historiae) nor ‘events that happened at the same time’ (eodem tempore gesta). This might therefore be the most tenuous of parallel passages in the gospels, since in the author’s view there is no deeper unity of meaning (sensus), and even at the surface level (superficies) the correspondence is not a close one, presumably merely the idea of Jesus consuming food.⁶² Next, the Reference Bible considers vicina passages that are similar by sensus, and not on the level of wording. Here the chosen exemplum is Eusebius’ pairing of the Matthean and Lukan genealogies with the Johannine prologue. The author, initially ignoring the Lukan parallel, cites Matthew 1:1–16 (= Mt §1) alongside John 1:1–5, 9–10 (= Jn §1, 3). These passages are ‘very different’ (longe dissimilia) with respect to their ‘surface’ (superficie), yet they nevertheless retain ‘a certain proximity of senses’ (vicinitas sensuum),⁶³ though not enough to conclude that the senses are ‘the same’. As with the other examples, the author does not elaborate further on why he classifies this parallel in this manner. Obviously the two passages have no verbal correspondence, which must explain their perceived lack of ‘surface’ similarity. One wishes that he had said more about how the sensus of the two passages is ‘similar’, but in view of his silence all we can do is speculate that he understood the passages as making similar assertions about the incarnation or dual natures of Christ. Finally, although the author began by asserting that there were only two ways in which passages could be vicina (either by ‘surface’ or by ‘sense’), for the sake of completeness he adds at the end a further way in which passages could fall into this category, namely by being similar in terms of both the ‘surface’ and the ‘sense’ of the text. The texts chosen here are the Matthean and Lukan genealogies. Both consisting of a list of fathers and sons, these two passages are clearly ‘similar’ with respect to the ‘surface’ of the text, though their divergence in many of the names listed precludes their classification as ‘the same’. Yet they also are ‘similar’ with respect to their ‘sense’, since they both narrate the genealogy of ‘the same family’, tracing the Davidic lineage of Jesus. Hence, these passages are vicina in both categories of relation. Whether the author of the Reference Bible intended his work to be a reference tool for students or other scholars or whether we have here his own lecture notes or a transcript, it is clear that he has provided a much more extensive engagement with the Canon Tables than we have seen thus far. Some ⁶² One could, nevertheless, view these passages as having more in common than the author of the Reference Bible suggests if they were read as parallel events in which the resurrected Christ consumed food and so demonstrated that his resurrected body was real. In other words, theologically speaking they may have the same sensus, even if they refer to events that happened at different times and places. ⁶³ Reading sensuum for sensum in the manuscript.

Scholarly Practices: the Hiberno-Latin Tradition

221

of his analysis overlaps with that of the Pauca de libris text considered previously, but the Reference Bible also adds much original material, especially by considering the different types of parallels in the tables. Eusebius, in his original letter explaining the system, had alternately said that the evangelists spoke κατὰ τῶν αὐτῶν (‘concerning the same things’) and τὰ παραπλήσια (‘similar things’ or literally ‘neighboring things’), seeming to use the two phrases as basically synonymous.⁶⁴ In his letter to Damasus, Jerome took over this terminology, referring to the loca vicina vel eadem in the gospels, and by presenting the words as a pair he implied that they referred to two classes of parallels. Moreover, those passages that are ‘the same’ he implied could be either simply eadem without qualification, probably suggesting exact verbal correspondence, or it might be that one evangelist has ‘expressed the same sense [as another evangelist] in a different manner’ (eundem sensum alius aliter expressit), implying not agreement in terms of wording, but some kind of deeper correspondence at the level of meaning.⁶⁵ Jerome, however, made this statement in passing, merely as an account of how gospel manuscripts had been corrupted in his day, and he thus gave no extended explanation of how these two classes might differ, nor did he provide any examples of either. The Irish Reference Bible, therefore, appears to be the first surviving account of someone exploring how these words could function as technical terms to classify the kinds of corresponding passages in the gospels, and as such it marks a significant moment in the history of the scholarly study of the relationships amongst the four canonical gospels.

SEDULIUS SCOTTUS The final source to be considered in this chapter is undoubtedly Irish and also clearly carries forward this exegetical tradition in an even more systematic form. Little is known about the early career of Sedulius Scottus other than that he came from Ireland and spent most of his life in Francia in a circle of Irish peregrini in the mid-ninth century. His corpus comprises a variety of different genres and demonstrates a high degree of learning. Eighty-three poems are attributed to him, as is the Collectaneum, a work consisting of a series of extracts from a wide range of classical and Christian sources, and also the De rectoribus Christianis, a presentation of the ideal Christian ruler probably directed towards Charles the Bald. Finally, Sedulius composed grammatical commentaries on Eutyches, Priscian, and Donatus, and biblical commentaries on the Pauline corpus, on the Gospel of Matthew, and on the series of ⁶⁴ ep. Carp. (NA²⁸, p. 89* 90*).

⁶⁵ Biblia sacra iuxta Vulgatam versionem, p. 1516.

222

The Eusebian Canon Tables

prefatory texts that by this point were regularly copied in gospelbooks. A surviving Greek psalter with interlinear Latin in Sedulius’ hand attests his linguistic capabilities (Paris, Bibliothèque de l’Arsenal, MS 8407).⁶⁶ Moreover, aside from his Irish origins, there is telling evidence to tie him directly to the earlier Irish tradition I have considered thus far. In addition to the aforementioned works, he was also responsible for providing a recension of Ailerán’s Interpretatio mystica et moralis progenitorum Domini Iesu Christi, so it is reasonable to suppose that he was also aware of Ailerán’s Canon Evangeliorum.⁶⁷ This small detail nicely ties together the end points of the two centuries I am considering, and also serves as an inclusio around the two anonymous works in between.⁶⁸ If Ailerán considered the Canon Tables in the genre of poetry, and the authors of the Pauca de libris and the Reference Bible did so in lecture format, Sedulius explored their intricacies through the medium of commentary. The biblical commentary was of course commonplace by this point, but Sedulius applied this format to gospel prefaces, composing line-by-line expositions of the Priscillian prologues (argumenta), of Jerome’s Novum opus and Plures fuisse, of the ps-Jerome text known as Sciendum etiam, and finally of the Latin translation of Eusebius’ Letter to Carpianus. Discussion of Canon Tables comes up in the commentaries on Novum opus and Sciendum etiam, and, of course, in the one on the Eusebian letter. There are indicators in the texts that this was the order in which Sedulius composed these works, so I will progress through each of them briefly. Roughly halfway through his explanation of Jerome’s letter to Damasus, the Novum opus, Sedulius turns to consider the Canon Tables, following Jerome’s own description of them. He immediately takes up the issue of types of parallels, picking up right where the Reference Bible left off. Like that earlier text, he asserts that there are three kinds of passages in the gospels, eadem, vicina, and sola. However, his analysis of these terms differs from that of the earlier Irish author. Whereas the Reference Bible stated that eadem passages could be either the same in sensus and superficies or possibly in only one or the other, Sedulius defines eadem capitula as those that agree both in verbis and in sensu, such that ‘there is no difference between them’. In other words,

⁶⁶ See Luned Mair Davies, ‘Sedulius Scottus (fl. 840x51 860x74)’, in Oxford Dictionary of National Biography (Oxford: Oxford University Press, 2004) [http://www.oxforddnb.com/view/ article/50134, accessed 29 Jan 2014]. For further information on Sedulius, see the introductions in Edward Doyle, On Christian Rulers and the Poems (Binghamton, NY: Medieval & Renaissance Texts & Studies, 1983), and R. W. Dyson, De Rectoribus Christianis (Woodbridge: Boydell & Brewer, 2010). ⁶⁷ Sedulius’ recension is included in Breen’s translation of Ailerán’s work noted above. ⁶⁸ So also Mullins, ‘The Insular Reception’, p. 104, who proposed that Sedulius’ usage of earlier Irish sources demonstrates ‘a continuous tradition of interpreting the Eusebian series from the seventh to the ninth century’.

Scholarly Practices: the Hiberno-Latin Tradition

223

Sedulius’ definition of eadem is more restrictive than that of the Reference Bible. Those capitula that are vicina, as one would expect, have a lesser degree of correspondence, merely a ‘certain likeness’ (similitudinem), whether in verbis or in sensu or both.⁶⁹ Sedulius next enters into a discussion of the various kinds of errors that had crept into the gospel tradition. Given that the four gospels overlap so much, yet are distinct, and considering that those responsible for copying the gospels were intimately familiar with them from the liturgy and study, the situation was ripe for inattentive scribes bringing the four books into conformity with one another. Sedulius summarizes the possible mistakes that might befall a ‘defective translator’ or a ‘sleepy copyist’ as adjectio, immutatio, detractio, and transmutatio.⁷⁰ He, like Jerome before him, emphasizes the power Eusebius’ system has for reigning in what threatened to become a wildly chaotic textual tradition. The Canons showed the proper place of each passage, as well as the interrelationships between the four, and thereby enabled the scribe to maintain the diversity of the gospel tradition rather than letting the four narratives collapse together. Like the earlier Irish scholars, Sedulius next considers the four modi of the Canons, which, he points out, are understood by the titles above each table providing their ‘proper definitions’ (propriis definitionibus). The four modi correspond to either four evangelists agreeing, or three, or two, or just one, each of which may be further subdivided into species or formulae, according to the kind of correspondence in a given parallel. Sedulius notes that the ‘question arises’ (quaestio oritur) why the order of the Canons proceeds from four agreements down to individual passages, rather than in reverse, which might be regarded as a more natural, or perhaps ‘better’ ordo. His answer to this query is in keeping with the emphasis on the unity of the gospel tradition noted earlier in Ailerán’s poem. Sedulius says that the agreements between the four gospels are placed first to highlight ‘the excellence of the more numerous agreements’ (excellentia numerosioris concordantiae). In other words, the accent lies on what the gospels say together and their overall harmony. Moreover, Sedulius points out that the ten Canons correspond with the decalogue, thereby showing the ‘greatest agreement’ (concordissimam) between the law and the gospel.⁷¹ The harmony displayed numerically by the Canon Tables is thus extended to all of the Bible as Christian scripture. There is not space at present to comment in detail on all of Sedulius’ exposition, but a few other pertinent features should be highlighted. In addition to providing an extremely detailed account of how to use the Canon Tables, he also focuses on each gospel individually. Although the ⁶⁹ exp. in praef. Hier. 12 (PL 103, col. 340). ⁷⁰ exp. in praef. Hier. 13 14 (PL 103, col. 340 1). ⁷¹ exp. in praef. Hier. 15 16 (PL 103, col. 341 2).

224

The Eusebian Canon Tables

enumeration of capitula within each gospel begins with the number one and carries on through to the end, the sequence of the Canons noted for each capitulum appears completely random. Thus, none of the gospels begins with a capitulum corresponding to Canon I, nor does any gospel include capitula in all ten Canons. Moreover, although the numbering of the capitula reaches several hundred within each gospel, the Canon number for each capitulum never goes beyond ten, because there are only ten Canons.⁷² While such observations may seem to border on a concern with mere trivialities, they are also indicative of a scholarly attempt to analyse in minute detail the inner workings of the Eusebian paratext and the way in which it provides a unique perspective onto the text of the fourfold gospel. Sedulius also places special emphasis on the fact that Jerome says that each passage in the gospels should have only its own capitulum number written in the margin, distinct to that individual gospel. The reason he may have been so concerned with this issue is that, as noted in the last chapter, by his day Latin gospelbooks often carried an enhanced version of Eusebius’ system, which included the capitula for the parallel passages from the other gospels copied directly into the margin. Sedulius recognizes a problem here, since Jerome’s letter only mentions a single capitulum number written in the margin, whereas the manuscripts he knew had the corresponding capitula as well. His explanation for this innovation is that either ‘some kind person’ (aliquis benevolus) may have been trying to remove the ‘labour and difficulty’ of constantly turning back to the tables at the front, or some scribe may have had a manuscript which lacked the tables, so he inserted the parallel capitula numbers in the margins to preserve the functionality of the system.⁷³ The emphasis on functionality in Sedulius’ exposition is not to be missed, along with his interest in the transmission history of the Eusebian apparatus itself and awareness of the changes to it that had already occurred prior to his day. Sedulius’ shorter commentary on the ps-Jerome preface Sciendum etiam is largely taken up with a single issue.⁷⁴ Strictly speaking, every capitulum does not appear in the Canon Tables only once. Although every capitulum only appears in a single Canon, certain capitula are repeated multiple times in a single Canon, if more than one parallel to that given capitulum shows up in another gospel(s). So, for example, in Canon I, presenting parallels between all four gospels, Mt §11 is shown alongside Mk §4 and Lk §10, repeated four times, in each instance joined with a separate capitulum from John (§6, 12, 14, 28).

⁷² exp. in praef. Hier. 16 18 (PL 103, col. 342 3). ⁷³ exp. in praef. Hier. pp. 20 1 (PL 103, col. 343 4). ⁷⁴ For the listing of early manuscripts that have the Sciendum etiam text, see McGurk, Latin Gospel Books, pp. 110 11. He lists five manuscripts from the period before 800. Cf. Mullins, ‘The Insular Reception’, pp. 76 7; Houghton, The Latin New Testament, p. 157. The Sciendum etiam text can be found in De Bruyne, Bogaert, and O’Loughlin, Prefaces to the Latin Bible, p. 158.

Scholarly Practices: the Hiberno-Latin Tradition

225

Such repetition would have been especially liable to mistaken copying, so the short text Sciendum etiam draws attention to this feature of the Eusebian apparatus. Moreover, the user of the Canons might initially be puzzled by the repetition in the tables, and wonder whether there are in fact multiple capitula in Matthew numbered §8. Sedulius accordingly explains that these peculiar instances in the tables indicate nothing more than that ‘what the three evangelists spoke one time in their individual capitula, the fourth evangelist recounted multiple times’ (illud quod tres evangelistae in singulis capitulis semel dixerunt, quartus evangelista toties multipliciter narraverit).⁷⁵ Sedulius’ commentary on Eusebius’ Letter to Carpianus need not detain us long. Having given such a detailed explanation of how the Canon Tables function in his exposition of the Novum opus, the commentary on the Eusebian letter is largely taken up with the new information added by the letter pertaining to the origins of the system, and with resolving certain grammatical issues that the text presents, not a surprising topic of concern given Sedulius’ other grammatical commentaries. He begins by discussing the earlier work of Ammonius of Alexandria, whose labour Eusebius had relied upon, and he follows Eusebius in pointing out the problematic nature of this earlier attempt to sort out gospel parallels. As Sedulius observes, Ammonius’ method resulted in what could only be called ‘one gospel’, eliminating the diversity of the four by dissolving the integrity of the latter three gospels. Eusebius’ solution, by contrast, is superior, since it allows one to know what is eadem, vicina, and sola, while preserving the ordo capitulorum of each gospel. Nevertheless, Eusebius’ work stands indebted to Ammonius, since, as Sedulius plausibly conjectures, Eusebius used the earlier work to find seven different kinds of agreements with Matthew, corresponding to his first seven Canons, along with an eighth category of uniquely Matthean passages, which became Canon XMt. Sedulius supposes that Eusebius then conducted further research beyond what Ammonius had done and thereby added two more categories of passages, which became Canon VIII and IX.⁷⁶ In the remainder of the commentary, Sedulius provides several more examples of parallels from the Canon Tables, some of which were not noted earlier in the Reference Bible. Along the way he comments briefly on music and number theory, tipping his hat to Augustine’s De musica,⁷⁷ and attempts to clarify the prolixitas and obscuritas of Eusebius’ difficult syntax.⁷⁸

⁷⁵ exp. in praef. Hier. 25 6 (PL 103, col. 346 7). Sedulius mistakenly says that Mk §5 is repeated four times, rather than Mk §4. However, only two paragraphs later he gives the correct number. Such a mistake illustrates well the ease with which a scribe might miscopy a number, especially when using Roman numerals. ⁷⁶ ex. Eus. 2 3 (Espositio, pp. 83 5). ⁷⁷ ex. Eus. 8 (Espositio, p. 88). ⁷⁸ ex. Eus. 10 (Espositio, p. 90).

226

The Eusebian Canon Tables

Clearly Sedulius draws upon the same exegetical tradition as the previous three texts. What he adds to them is the genre in which he deals with the material, accompanied by his notable thoroughness, and his explicit recognition that the Canon Tables ensure that the integrity of each gospel remains intact by acting like reinforcing girders preventing the four gospels from collapsing in on one another through scribal inattention. In other words, Sedulius recognizes that the Canon Tables perform an important structuring function within the gospelbook, precisely the paratextual effect that I argued for in chapter 1. Moreover, Sedulius shows an awareness of the hermeneutical implications of the Eusebian apparatus: with each passage securely in its proper place, the reader is enabled to ‘grasp the force and the sense [of the capitula] in the context of [each] gospel’.⁷⁹ Finally, the degree of interest Sedulius takes in the history of the Eusebian apparatus is remarkable, both in terms of its relation to Ammonius’ work and the changes that had occurred to it subsequent to Eusebius. As the reader may recall, my own assessment of the former issue in chapter 2 is largely in keeping with the conclusions of the Irish scholar.

CONCLUSIO N Despite the wide and rapid adoption of the Canon Tables in late antique gospelbooks across a variety of languages, few authors, it seems, explored the possibilities presented by the innovative paratext. Of course future discoveries may change this picture considerably, but on available evidence the Irish exegetical tradition stands out against this backdrop for taking up the baton and considering further the relationships amongst the gospels, specifically the different types of parallels between them. Even though they were not the first to use the Eusebian apparatus systematically—recall Augustine’s De consensu— still the lack of polemical intent in their treatments provides a different tenor to the texts I have considered in this chapter. Rather than simply trying to show that there are no contradictions amongst the gospels, these Irish authors engaged in a more open-ended exploration of the way in which the gospels related to one another and the way in which they together witnessed to the one Jesus Christ. As I argued in chapter 1, the main function of the Canon Tables is to order the material in the fourfold canon according to various categories of relation amongst the gospels. What is most remarkable about these Irish texts is the way in which they extend this ordering function by creating more specific ⁷⁹ exp. in praef. Hier. 23 (PL 103, col. 345). Sed quid prodest in canonum tramitibus numerum capitulorum solummodo dignoscere, nisi eorum vim ac sensum in contextu ipsius Evangelii ad purum comprehendere possimus?

Scholarly Practices: the Hiberno-Latin Tradition

227

classifications to account for the variety of kinds of agreement and disagreement amongst the Eusebian parallels. Though their culture had always existed on the periphery of the Greco-Roman world of antiquity, Irish exegetes between the seventh and ninth centuries stood at the forefront of the scholarly analysis of the literary problem presented by the fourfold gospel, exploring this issue in poetry, lecture, and commentary.

7 Seeing the Salvation of God Images as Paratext in Armenian Commentaries on the Eusebian Canon Tables

The present chapter extends the focus on the symbolic aspect of the Canon Tables seen in the last chapter, particularly in Ailerán of Clonard’s poetry, but now we move geographically back across the Mediterranean to the eastern fringes of the ancient Roman world and move forward chronologically from late antiquity into what is incontestably the Middle Ages. The reason is that the reception of the Canon Tables among Armenian authors from the eighth century onwards was as distinctive as that among the Irish between the seventh and ninth centuries, albeit in a different respect. Whereas the Irish scholars considered in the last chapter used the numbers in the ten tables and the intertextual relationships they represented to investigate the meaning of the fourfold gospel, the Armenian authors to be considered below explored how the artistic imagery around the numeric matrices could function symbolically in relation to sacred scripture. Recall that Ailerán used the Canon Tables in conjunction with the traditional evangelist symbols to exploit these four heavenly patrons as representations of different aspects of Christ, who himself was the true referent of the fourfold gospel and so also the ‘canon’ of the Canon Tables. In the sources examined here this same desire to find multiple layers of meaning in the Eusebian paratext is also present, although the scope of the symbolic import expands drastically to encompass all of salvation history and is viewed as hidden in a more complex and elaborate artistic scheme. The artistic aspect of the history of the Eusebian paratext is the one area that has been the focus of a reasonable amount of scholarly attention. In his seminal work on the topic published eighty years ago, Carl Nordenfalk explained that he had initially intended to write a monograph on the entire field of late antique book illumination, but, when he found this too ambitious, decided to focus on the Canon Tables because they ‘promised the richest and

The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0008

Seeing the Salvation of God

229

overall most important results for late antique art history’.¹ Thus, Nordenfalk’s monograph was an art-historical study, employing source-critical methodology to reconstruct a Eusebian ‘archetype’ in order to illumine late antique book decoration more broadly. A number of studies since Nordenfalk have extended this line of analysis, and more are surely on the horizon.² The primary goal of this chapter is not to contribute another art-historical study showing the transmission of certain motifs in various manuscript traditions or providing an iconographic analysis of the imagery. Rather, in keeping with the focus of the present study, the aim of this chapter is to draw attention to another way the Eusebian Canon Tables could function as a paratext in relation to the fourfold gospel. Indeed, the primary argument in what follows is that the artwork that grew up around the Canon Tables was not extrinsic to their function as a paratext, but rather extended their utility as a means of ordering knowledge and presenting the text of the four gospels to a prospective reader.³ Nordenfalk himself, in several later studies, touched upon this issue in passing, arguing that the decorated pages ‘function as a propyleum through which we approach the sanctum sanctorum of the Holy Writ’,⁴ a description that is strikingly close to Genette’s claim that a paratext is a ‘threshold’ or ‘vestibule’ ‘that offers the world at large the possibility of either stepping inside or turning back’.⁵ The difference between these two statements is that for Genette this description was a metaphorical way of highlighting the function of a paratext, whereas Nordenfalk was describing actual images of architectural facades that commonly frame the ten numeric tables and so serve as a monumental preface to the fourfold gospel. Thus, on Nordenfalk’s interpretation of Canon Table ornamentation, these decorated pages visualize their paratextual function using the same conceptual metaphor that Genette would later realize was an apt description of paratexts in general. ¹ Nordenfalk, Die spätantiken Kanontafeln, p. 13: ‘Ich habe mich dann für die Behandlung der spätantiken Kanontafeln entschieden, weil dieses Material die reichsten und für die spätan tike Kunstgeschichte im allgemeinen wichtigsten Ergebnisse versprach’. ² For a general survey that is now admittedly somewhat dated, see Wessel, ‘Kanontafeln’, pp. 927 68. ³ Genette, Paratexts, p. 406, acknowledged that the ‘illustration’ is one of three paratextual practices that he omitted from his study in the light of the fact that the expertise required to treat it ‘exceeds the means of a plain “literary person” ’. ⁴ Nordenfalk, ‘Canon Tables on Papyrus’, p. 30. Cf. Carl Nordenfalk, ‘The Beginning of Book Decoration’, in Beiträge für Georg Swarzenski, ed. Oswald Götz (Berlin: Verlag Gebr. Mann, 1951), p. 17: ‘the motif [of arcades] was also adopted as a solemn introduction to the Gospels, appearing as a celestial aqueduct carrying the Water of Life’; Carl Nordenfalk, ‘The Apostolic Canon Tables’, Gazette des Beaux Arts 105 (1963): p. 18: ‘the Canones, as Eusebius called them, were housed in a set of arcades, forming as it were an impressive atrium at the entrance of the sacred text itself ’. A similar conclusion, specifically with respect to the archways in the Garima gospels, was reached by McKenzie and Watson, The Garima Gospels, p. 159. ⁵ Genette, Paratexts, pp. 1 2.

230

The Eusebian Canon Tables

The argument pursued in this chapter is fundamentally in agreement with Nordenfalk’s basic insight, but attempts to add greater depth and complexity to it by highlighting texts that fell outside the purview of his study. Although Canon Tables in Armenian gospelbooks were central to his research, and in fact, in his view, represented most faithfully the Eusebian archetype,⁶ he did not consider certain texts that also exist in Armenian that comment directly on this manuscript imagery. These texts in fact support his claim that the arcades functioned as a gateway leading to ‘the sanctum sanctorum of the Holy Writ’, since they present the Canon Table decorative scheme as a visual preface to the fourfold gospel, which was divinely inspired for the purpose of fostering the spiritual contemplation required of a reader before encountering the sacred text. Subsequent scholars have mentioned these Armenian sources in passing, though most have done so in the context of attempts to interpret the iconography of decorated sets of Canon Tables and only a few have considered the assumptions they present about the way text and image relate to each other.⁷ Taking up the latter line of enquiry, in what follows I emphasize what these sources reveal about the way readers of the fourfold ⁶ Cf. Nordenfalk, Die spätantiken Kanontafeln, 109, referring specifically to the Etchmiadzin Gospels. ⁷ The two texts that I consider in this chapter were first brought to the attention of western scholars in Sirarpie Der Nersessian, Manuscrits arméniens illustrés des XIIe, XIIIe et XIVe siècles de la Bibliothèque des pères mekhitharistes de Venise (Paris: E. de Boccard, 1937), pp. 58 61; Sirarpie Der Nersessian, Armenian Manuscripts in the Freer Gallery of Art, Oriental Studies 6 (Washington, DC: Freer Gallery of Art, 1963), pp. 100 1. After that point they were noted in several subsequent studies: Underwood, ‘Fountain of Life’, pp. 77 80; Günter Bandmann, ‘Beobachtung zum Etschmiadzin Evangeliar’, in Tortulae: Studien zu altchristlichen und byzantinischen Monumenten, ed. Walter Nikolaus Schumacher, Römische Quartalschrift für christliche Altertumskunde und Kirchengeschichte. Supplementheft 30 (Rome: Herder, 1966), pp. 11 n. 2, 15 n. 16, 22 n. 63; Dickran Kouymjian, ‘Armenian Manuscript Illumination in the Formative Period: Text Groups, Eusebian Apparatus, Evangelists’ Portraits’, in Il Caucaso: Cerniera fra Culture dal Mediterraneo alla Persia (secoli IV XI), Settimane di studio del Centro italiano di studi sull’alto Medioevo 43 (Spoleto: Centro italiano di studi sull’alto Medioevo, 1996), pp. 1031 2; Vrej Nersessian, Treasures From the Ark: 1700 Years of Armenian Christian Art (London: The British Library, 2001), pp. 79 82; Vrej Nersessian, The Bible in the Armenian Tradition (London: The British Library, 2001), pp. 70 4; S. Peter Cowe, ‘The Bible in Armenian’, in The New Cambridge History of the Bible. Volume 2: From 600 to 1450, ed. Richard Marsden and E. Ann Matter (Cambridge: Cambridge University Press, 2012), p. 159; McKenzie and Watson, The Garima Gospels, pp. 141 4. The most extensive discussion in a western language is to be found in Thomas F. Mathews and Avedis K. Sanjian, Armenian Gospel Iconography: The Tradition of the Glajor Gospel, Dumbarton Oaks Studies 29 (Washington, DC: Dumbarton Oaks Research Library and Collection, 1991), pp. 166 76, though see the thorough treatment of this topic in Vigen H. Ghazaryan, ԽՈՐԱՆՆԵՐԻ ՄԵԿՆՈՒԹՅՈՒՆՆԵՐ = Commentaries of Canon Tables (Yerevan: Sargis Khachents, 1995). The present chapter builds upon Judith McKenzie’s remark that these texts explain ‘how viewing the beauty of the tables is preparation for reading the gospel texts’ (McKenzie and Watson, The Garima Gospels, p. 142). McKenzie in turn was following the lead of Mathews and Sanjian, who commented, ‘By focusing the viewer’s attention on largely abstract forms and colors, the canon tables were meant to focus the powers of his soul on the central mysteries of Christian revelation’ (Mathews and Sanjian, Armenian Gospel Iconography, p. 172).

Seeing the Salvation of God

231

gospel used this imagery paratextually, and, in order to shed light on the process they envision, I draw on analogies with discussions of memory in the classical rhetorical tradition and the way in which that tradition was received in the medieval Latin West. The upshot is that these Armenian authors utilized the decorative scheme as an elaborate program of mnemonic symbolism which summarizes the meaning of the fourfold gospel and in so doing provides the viewer with the proper theological context within which an interpretation of the four-gospel codex should take place. As will become clearer in what follows, my interpretation of these sources is inspired by the work of Mary Carruthers, who, in her insightful study The Book of Memory, drew attention to the mnemonic potential of the usual gridded format of a manuscript page containing Canon Tables.⁸ Carruthers, however, made this observation in passing and did not explore the Eusebian apparatus any further, and, at any rate, the Armenian material that most clearly confirms her hypothesis fell well beyond the scope of her study.

A R T I S T I C A D O R N M E N T OF C A NON TAB LE S I N LAT E A N TI Q U E AN D M E D I E V A L G O S P E L BO O K S To set the stage for my analysis of these Armenian texts, it is necessary first to give a broad overview of surviving Canon Table artwork and the secondary scholarship on it. The first thing to note is that artistic embellishment is not strictly speaking necessary for Eusebius’ apparatus to fulfil its purpose as a cross-referencing system. The design of the tables requires, at a minimum, that a scribe maintain proper alignment of the columns and rows in each matrix, so that the relationship between the adjacent numerals and the gospel to which each numeral belongs is clearly discernible. This need may be fulfilled by a simple grid pattern, which is indeed all that is found in a few gospelbooks such as the late seventh-century Book of Durrow from the British Isles, which presents the section numbers between intersecting horizontal and vertical lines surrounded by a decorative rectangular frame along the border of the page (see fig. 33). This grid is usually maintained even in more ⁸ Carruthers, The Book of Memory, p. 118: ‘It has been suggested that, in this context, an arcade motif may derive from the ancient mnemonic advice to use buildings including intercolumnia, the spaces between columns as backgrounds for things to be remembered. Certainly intercolumnia is one of the most enduring types of memory locus. Within each rectangular space made by the columns in the Eusebian Tables, the name of the gospel is written at the top, and then the chapter numbers of the synoptic passages are recorded. Horizontal lines, sometimes colored, are drawn between every four numbers (in the Greek text) or five (in the Latin); the effect is to divide the page into a series of small rectangular bins, none holding more than five items. Such a layout is clearly designed for mnemonic ease’.

232

The Eusebian Canon Tables

Fig. 33. Canons V, VI, VII, and VIII in The Book of Durrow (seventh c.). Dublin, Trinity College Library, MS 57, fol. 9v (© The Board of Trinity College Dublin)

Seeing the Salvation of God

233

elaborate versions, and Nordenfalk is no doubt correct that it is a design element that goes back to Eusebius himself, particularly since, as was pointed out in chapter one, the same feature was employed in the presentation of Ptolemaic astronomical tables prior to Eusebius.⁹ Although this basic grid is all that is necessary to ensure proper copying of the tables, in most copies across all linguistic traditions the tabular grid is housed within an architectural frame, a feature that Nordenfalk again plausibly supposed went back to Eusebius himself in the light of its broad geographic diffusion (see figs 34–36).¹⁰ The architectural frames that are used to house the Canon Tables typically consist of two pillars, one on each side of the page, which rest on a simple lower base extending along the bottom of the page, and which support an arch or arches that connect the external pillars. The arch spanning the pillars creates a semicircular panel beneath it, which is often decorated with geometric patterns or with flora and fauna and typically also contains the titles for each of the columns below. Beneath this semicircular panel and between the outer pillars the numbers for a given table are listed, and in some examples, such as the Lindisfarne Gospels, intervening pillars across the page create several discrete spaces (i.e., ‘intercolumniations’), in each of which is written a single column of numbers (see fig. 35). Recently Judith McKenzie has put forward a compelling argument that this design ‘probably originally represented semi-domes’, such as the one in the House of the Large Fountain in Pompeii (VI 8, 22).¹¹ Although ornamental architectural frames appear widely in the classical and late antique tradition,¹² their use specifically to house text in a manuscript is rare in our surviving evidence, at least prior to the fourth century. The one pre-Eusebian example Nordenfalk drew attention to was a papyrus roll from the third century  which contains a number of school texts presented within an architectural framework (P. Cairo 65445), although the frame in this papyrus is comparatively simple and it is surely significant that this is a school text

⁹ Nordenfalk, Die spätantiken Kanontafeln, pp. 74 5: ‘Es besteht kein Zweifel, daß es zu der ursprünglichen Ausstattung der eusebianischen Kanones gehört’. ¹⁰ Nordenfalk, Die spätantiken Kanontafeln, p. 73. See also Nordenfalk’s full analysis of the ‘Die Kanonbögen’ on pp. 73 126. ¹¹ McKenzie and Watson, The Garima Gospels, pp. 85 8. See her full analysis of the architectural frames on pp. 83 94, 105 16, which is the most extensive discussion of the topic since Nordenfalk. Although McKenzie is primarily focused on three Ge‘ez gospelbooks, she brings in much comparanda from the classical tradition as well as from Syriac and Armenian manuscripts. ¹² Cf. Verity J. Platt and Michael Squire, eds, The Frame in Classical Art: A Cultural History (Cambridge: Cambridge University Press, 2017).

234

The Eusebian Canon Tables

Fig. 34. Canons XLk and XJn in Abba Garima III (fifth seventh c.). Ethiopia, Abba Garima Monastery, AG II, fol. 260v (© Michael Gervers, 2004)

rather than a luxury item.¹³ A further example, slightly later than Eusebius, is the famous Codex Calendar, made in 354 by the Roman calligrapher Filocalus, which used architectural frames to house both text and illustrations ¹³ Cf. Nordenfalk, ‘The Beginning of Book Decoration’, pp. 11 13, including images of the manuscript on p. 13. See also McKenzie and Watson, The Garima Gospels, pp. 83 5, who also reproduce images of the manuscript.

Seeing the Salvation of God

235

Fig. 35. Canon V in the Lindisfarne Gospels (eighth c.). Here the more realistic marble columns found in earlier manuscripts have been stylized and filled in with geometric patterns and birds. London, British Library, Cotton MS Nero D IV, fol. 15r (© The British Library Board)

236

The Eusebian Canon Tables

Fig. 36. Canon XJn in The Gospel Book of Theophanes (second quarter of the twelfth c.). The enclosure of the arch within a rectangular headpiece is a later development. The inclusion of figures standing atop the columns, which represent the months of the year and the virtues, is also very unusual, occurring only in this manuscript and a handful of related ones. National Gallery of Victoria, Melbourne, Felton Bequest, 1960 (710-5), fol. 7v

Seeing the Salvation of God

237

(see, e.g., the Natales Caesarum on fol. 7 and the portrait of Emperor Constantius II on fol. 13, Barb. lat. 2154, Biblioteca Vaticana, Rome).¹⁴ Given the common use of architectural frames for inscriptions,¹⁵ their appearance in the fourth century in manuscripts, both to house text and images, could be an adaptation of an existing design element to a new medium. Nordenfalk hypothesized that the use of these architectural frames to enclose the numeric grids of the Eusebian apparatus was a ‘compensationphenomenon’ (‘Kompensationserscheinung’). Supposing that a ‘scientific numeric concordance’ (‘wissenschaftliche Ziffernkonkordanz’) like the Canon Tables was somehow at odds with the sanctity of the gospels, Nordenfalk suggested that the architectural frames ‘counterbalance the inner lack of solemnity with magnificent external splendor’ (‘inneren Mangel an Feierlichkeit durch äußere Prachtentfaltung aufzuwiegen’).¹⁶ Although Nordenfalk’s assumption of a tension between ‘Wissenschaft’ and ‘Feierlichkeit’ is more at home in modernity than in late antiquity, his main point is no doubt on target, namely that the architectural frames increase the gravitas of what would otherwise be a simple numeric grid. We will see later in this chapter that our Armenian authors pick up on precisely this point, seeing a profound depth of mystical symbolism specifically in the architectural adornment of the tables. The Canon Table decorative motif that has generated the most amount of scholarly discussion is the full-page illustration of a round building that concludes the series of tables in some manuscripts (see figs 37–40). This structure, which is usually called a tholos in the secondary literature, typically has a domed roof supported by four or more pillars and often also includes one or more curtains hanging within the structure between the pillars.¹⁷ Although it does not show up in any early Latin examples and is rare in Greek manuscripts, Nordenfalk argued that this image was a component of the Eusebian archetype, given that it fits so well with the architectural ¹⁴ Fol. 7 is reproduced in Nordenfalk, Die spätantiken Kanontafeln, p. 119, and fols. 7 and 13 in Michele Renee Salzman, On Roman Time: The Codex Calendar of 354 and the Rhythms of Urban Life in Late Antiquity, The Transformation of the Classical Heritage 17 (Berkeley, CA: University of California Press, 1990), fig. 7, 13. ¹⁵ Cf. Sean V. Leatherbury, ‘Writing, Reading and Seeing Between the Lines: Framing Late Antique Inscriptions as Texts and Images’, in The Frame in Classical Art: A Cultural History, ed. Verity J. Platt and Michael Squire (Cambridge: Cambridge University Press, 2017), pp. 544 81. On p. 566 Leatherbury drew attention to the framed dedicatory frontispiece in the Codex Calendar of 354, which is an adaptation of a type of framed inscription (called a tabula) that was originally hung in votive contexts. An image of the frontispiece can be seen in Salzman, On Roman Time, fig. 1. ¹⁶ Nordenfalk, Die spätantiken Kanontafeln, p. 125. For another interpretation of the mean ing of the architectural imagery, see Gearhart, ‘From Divine Word to Human Hand’, pp. 430 3, who followed the work of Reudenbach, ‘Der Codex als heiliger Raum’. ¹⁷ On the motif of curtains, which occur not only on the tholos but also frequently on the architectural frames housing the ten numeric tables, see Reudenbach, ‘Der Codex als heiliger Raum’, pp. 66 71, who argued that they signify the revelation of the Word of God.

238

The Eusebian Canon Tables

Fig. 37. Tholos image in Abba Garima I (sixth seventh c.). Ethiopia, Abba Garima Monastery, AG II, fol. 258v (© Michael Gervers, 2004)

Fig. 38. Fountain of life image in the Gospels of St Médard de Soissons (before 827). Paris, Bibliothèque nationale de France, MS lat. 8850, fol. 6v

Seeing the Salvation of God

239

Fig. 39. Tholos with a hypothesis inscription, serving as a frontispiece for the sequence of Canon Tables that follow in a Greek gospelbook (ninth c.). Venice, Biblioteca Nazionale Marciana, MS gr. I 8, fol. 3r

Fig. 40. Tholos image in the Adishi Gospels (897), fol. 5v. The tall grilles that have been added between the outer and inner columns on either side of the central section conform the image more closely to the representations of Christ’s tomb found on late antique pilgrims’ flasks. Mestia, Svaneti Museum, Georgian National Museum

240

The Eusebian Canon Tables

frames housing the preceding numeric tables, repeating the motif of the pillars used throughout the sequence.¹⁸ He also highlighted the wide diffusion of this motif, which shows up in Armenian manuscripts from the tenth and eleventh centuries,¹⁹ as well as the earliest surviving gospelbook in Georgian, the Adishi Gospels (897 ),²⁰ and a rare Greek example of likely Constantinopolitan origin in which the tholos prefaces rather than concludes the sequence of tables (Venice, Bibl. Marc. Cod. gr. I, 8, fol. 3r; ninth century).²¹ Unknown to Nordenfalk at the time, the structure also appears in the Ge‘ez gospelbook known as Abba Garima I, dated to c.600 , which is now the earliest surviving example.²² The companion volume from the same monastery, which is known as Abba Garima III, has a full-page illustration of the Jerusalem temple in place of the tholos image, probably a subsequent development but one that may carry a similar symbolic significance to the more common tholos.²³ Numerous later Ethiopian manuscripts from the fourteenth and fifteenth centuries also contain the tholos page.²⁴ Judith McKenzie has recently argued that the style of tholos found in a number of these manuscripts ‘goes back to Ptolemaic Alexandrian architecture, as carved from the rock in Petra and depicted in Roman wall-paintings’.²⁵ Yet despite its obvious similarities to this Ptolemaic structure, the tholoi that appear in sequences of Canon Tables came to carry a variety of distinctly Christian associations. Nordenfalk was the first to recognize the similarity of the tholos image with depictions of the aedicula built by Constantine over the Holy Sepulchre in Jerusalem, especially on late antique pilgrims’ flasks (ampullae),²⁶ and he pointed out that, since the tables are a symbolic summary of the fourfold gospel, it would

¹⁸ Nordenfalk, Die spätantiken Kanontafeln, p. 103: ‘Die Hinzufügung des “Rundtempels” weist eine innere Folgerichtigkeit auf, die mit aller Entschiedenheit dafür spricht, daß diese Schlußdekoration auf den eusebianischen Archetypus selbst zurückgeht’. See Nordenfalk’s full analysis of this feature on pp. 102 8. ¹⁹ See the examples from the Etchmiadzin Gospels and the Second Etchmiadzin Gospels on McKenzie and Watson, The Garima Gospels, p. 133. ²⁰ See McKenzie and Watson, The Garima Gospels, p. 134. ²¹ See McKenzie and Watson, The Garima Gospels, p. 139. ²² See McKenzie and Watson, The Garima Gospels, plate 42. ²³ See McKenzie and Watson, The Garima Gospels, plate 14. On the origins and meanings of these two images in the Garima gospelbooks, see McKenzie and Watson, The Garima Gospels, pp. 121 44. ²⁴ See the collected examples in McKenzie and Watson, The Garima Gospels, 142 3, discussed on pp. 139 41. ²⁵ McKenzie and Watson, The Garima Gospels, pp. 121, 130. Cf. Judith McKenzie, The Architecture of Alexandria and Egypt, 300 .. to .. 700 (New Haven, CT: Yale University Press, 2007), pp. 97 8. The connection with the tholoi at Pompeii and Petra was also noted by Nordenfalk, Die spätantiken Kanontafeln, p. 107; Theodor Klauser, ‘Das Ciborium in der älteren christlichen Buchmalerei’, Nachrichten der Akademie der Wissenschaften in Göttingen 7 (1961): p. 195. ²⁶ Nordenfalk, Die spätantiken Kanontafeln, pp. 107 8, who included drawings of several ampullae. See also the photograph of a pilgrim flask reproduced in McKenzie and Watson, The Garima Gospels, p. 134.

Seeing the Salvation of God

241

be fitting for them to conclude where the gospels themselves do—with Christ’s empty tomb.²⁷ The link between the tholos image and the Holy Sepulchre is especially clear in the Georgian Adishi Gospels, which adds tall grills in the space between the columns on either side of the central opening, exactly like depictions of the aedicula built by Constantine, and also includes several inscriptions pertaining to the resurrection within and around the structure.²⁸ Some scholars have argued that such associations might go back to Eusebius himself, in the light of his well-known connection with Constantine and interest in the topography of the holy land.²⁹ In fact, in his Vita Constantini Eusebius himself mentioned that Constantine’s aedicula was decorated with columns, and he also pointed out that he both lectured on the monument in the emperor’s presence and preached a sermon on its symbolism to dedicate the Church of the Holy Sepulchre in 336.³⁰ Nevertheless, other scholars have identified problems in the association of the tholos with the aedicula at the Holy Sepulchre and have instead proposed alternative models, including the ciborium of Eusebius’ cathedral in Caesarea, the Constantinian structure at the Church of the Nativity in Bethlehem, and a symbolic heavenly temple.³¹ ²⁷ Cf. Nordenfalk, Die spätantiken Kanontafeln, p. 108: ‘Als Schlußseite der Kanones, die eine Zusammenfassung der Berichte der Evangelien von dem Leben und der Lehre Christi darstellen, würde jedenfalls eine Erinnerrung an das Endziel des Erdenlebens Christi und den Inbegriff der christlichen Erlösungshoffnung als durchaus sinnvoll erscheinen’. ²⁸ Cf. Nordenfalk, Die spätantiken Kanontafeln, pp. 113 16; Underwood, ‘Fountain of Life’, pp. 93 5; McKenzie and Watson, The Garima Gospels, pp. 134 6. ²⁹ Carol Neuman De Vegvar, ‘Remembering Jerusalem: Architecture and Meaning in Insular Canon Table Arcades’, in Making and Meaning in Insular Art, ed. Rachel Moss (Dublin: Four Courts, 2006), p. 248: ‘one can scarcely envision a reference that would have been closer to the concerns of both Eusebius himself and his imperial patron’. On the possible connection between Eusebius, the Church of the Holy Sepulchre, and the tholos image, see also E. Klemm, ‘Die Kanontafeln der armenischen Handschriften Cod. 697 im Wiener Mechitaristenkloster’, Zeits chrift für Kunstgeschichte 35 (1972): pp. 86 91; Rouzanna Amirkhanian, ‘Les tables de canons arméniennes et le thème iconographique de la Jérusalem céleste’, Revue des Études Arméniennes 31 (2008 2009): pp. 181 232; McKenzie and Watson, The Garima Gospels, p. 138. ³⁰ Eusebius, v. C. 3.34; 4.33; 4.45. ³¹ In the light of the fact that Constantine’s aedicula was not finished until 334, presumably after Eusebius had created the image for the Canon Tables (so Nordenfalk, Die spätantiken Kanontafeln, p. 108 n. 2), Theodor Klauser suggested instead that the tholos was based on the ciborium of Eusebius’ own cathedral in Caesarea (Klauser, ‘Das Ciborium’, pp. 204 5). Klauser’s perceived problem rests upon an overly accurate dating of Eusebius’ archetype and also does not consider that he might have issued more than one model. Wessel, ‘Kanontafeln’, pp. 965 6, argued for a reference to the ciborium over the place of Christ’s birth in Bethlehem on the basis of its closer iconographic similarity. More recently, Gohar Grigoryan, ‘The Roots of Tempietto and Its Symbolism in Armenian Gospels’, Iconographica 13 (2014): pp. 11 24, has again pointed out the supposed chronological problem highlighted by Klauser and argued that instead of representing the aedicula, the image (which he calls a tempietto) ‘represent[s] a symbolic temple of God’ (p. 16). McKenzie and Watson, The Garima Gospels, p. 138, suggested ‘the tholos in Abba Garima I could have been alluding to the source or fountain of the water of life, and the two trees [on either side of the tholos] to the trees on either side of the river of life giving water and at the centre of the garden of Eden. It is possible that allusions to the new Temple and to the Holy Sepulchre are also intended’. Although to my knowledge it has not yet been considered in

242

The Eusebian Canon Tables

Building upon Nordenfalk’s foundational suggestion of an association between the tholos and the aedicula over the Holy Sepulchre, further studies have argued that other aspects of the decorative scheme also evoke the sacred site. For example, Carol Neuman de Vegvar has pointed out that the intercolumniations evident in a variety of insular manuscripts from the seventh century onwards vary between three and four per page, which ‘parallels the experience of a pilgrim moving inward through the arcade of the [Anastasis] Rotunda toward the facade of the Edicule and then departing outward through the Rotunda colonnade’.³² Still more recently Rouzanna Amirkhanian, seemingly unaware of de Vegvar’s study, has made a similar argument about the medieval Armenian examples, contending that in manuscripts such as the Etchmiadzin Gospels the overall decorative scheme is a coherent whole, with the tholos functioning as ‘a continuation or an architectural outcome’ of the sequence.³³ She argued that the adornment employed for the sequence of pages is modeled not just on the circular aedicula over the Holy Sepulchre, but also the ‘longitudinal’ basilica within which it was situated. On this interpretation, the series of Canon Table archways leading up to the final tholos is meant to refer to the nave of the basilica which presented parallel colonnades along which one walked to arrive at the aedicula.³⁴ Thus, both de Vegvar and Amirkhanian argue that the experience of the reader turning the pages of the codex is intended to symbolize the movement of a person through the physical space of the Church of the Holy Sepulchre, whether through the archways (so de Vegvar) or walking alongside them (so Amirkhanian).³⁵ These arguments resonate with the separate analysis of

analyses of the Canon Table tholos, reference should also be made to the sixth century mosaic from the church of the Priest John at Khirbet Mukhayyat in Jordan, which uses an architectural facade to frame a dedicatory inscription and includes several iconographic features that also appear in Canon Table ornamentation (e.g., cocks, peacocks, candles, fruit trees, shell niche, four columns). See the image in Leatherbury, ‘Writing, Reading and Seeing’, p. 579, who argued that it is a depiction of the Jerusalem temple, transported to ‘a paradisiacal, heavenly setting’ (p. 580). Other relevant mosaics are considered at McKenzie and Watson, The Garima Gospels, pp. 125 7. ³² De Vegvar, ‘Remembering Jerusalem’, p. 251. De Vegvar’s thesis depends upon the argument by Richard Krautheimer that early medieval buildings may ‘reference one another without direct copying, and that such referents are often pars pro toto’ (p. 243). Cf. Richard Krautheimer, ‘Introduction to an “Iconography of Medieval Architecture” ’, in Studies in Early Christian, Medieval, and Renaissance Art (Cambridge: 1969). ³³ Amirkhanian, ‘Les tables de canons arméniennes’, p. 190. ³⁴ Amirkhanian, ‘Les tables de canons arméniennes’, pp. 190 1. ³⁵ Although he argued against the notion that the tholos was exclusively a depiction of the aedicula at the Holy Sepulchre, Bandmann reached a similar conclusion about the unity of the overall decorative scheme in the Etchmiadzin Gospels: ‘Damit darf die Beziehbarkeit der Prologseiten des EE auf reale und dargestellte Portalarchitektur gesichert sein. Sie bezeichnen den Eingang zu dem als christlich gekennzeichneten architektonischen Organismus. Mit den porticusartigen Arkaden der Kanonbögen werden wir zum Tempietto am Ende geleitet’ (Bandmann, ‘Beobachtung’, p. 28). See also Grigoryan, ‘The Roots of Tempietto’, p. 17: ‘if we consider the canon tables as the path that the believer passes through, then the tempietto depicted after them, can be the Heavenly Jerusalem’.

Seeing the Salvation of God

243

Bruno Reudenbach, who has observed that, in terms of the experience of reading, the primary feature that distinguishes the codex from the scroll is the fact that the codex segments the unitary writing surface of the scroll into individual pages, which disrupts the continuous flow of text on the page and encourages further sectioning and segmentation of the text. I have already discussed in chapters one, two, and three the way in which Eusebius’ Canon Tables exploit this potential of the codex by encouraging a hypertextual mode of reading. Reudenbach, however, further argues that this segmentation of the writing surface into pages, which must be turned one after another, introduces an element of ‘depth’ (‘Tiefe’) in the reader’s experience of the book, in contrast to the flat surface of the scroll. As a result, the codex has ‘a spatial dimension’ (‘eine raumliche Dimension’) that the scroll lacks.³⁶ Reudenbach proposed that the architectural decoration of the Canon Tables is alluding to this ‘spatial dimension’ and acting as an entryway leading the reader into the ‘book-space’ (‘Buchraum’), and, since this is the preface to a gospelbook, it is not just any ordinary ‘book-space’, but one that is ‘filled with the Word of God, filled with God himself ’.³⁷ Perhaps inspired by the rich theological and liturgical associations of the architectural frames and the concluding tholos, artists added further embellishments to the decorative scheme. Most common are varieties of plants and animals, including a diverse array of birds and occasionally stags, which typically sit atop or alongside the frames.³⁸ The fact that these show up already in the aforementioned two late antique Garima gospelbooks demonstrates that such features appeared early in the tradition, at most two centuries after Eusebius’ exemplar(s).³⁹ Such imagery has been plausibly interpreted as symbolizing a garden paradise, a theme certainly at home in Christian theology.⁴⁰ The earliest tholoi images to appear in the Latin tradition occur in two Carolingian manuscripts—the Godescalc Gospel Lectionary (Paris, Bibliothèque ³⁶ Reudenbach, ‘Der Codex als heiliger Raum’, pp. 59 61. ³⁷ Reudenbach, ‘Der Codex als heiliger Raum’, pp. 61 6. Reudenbach also observed that Eusebius himself used such spatial imagery in his Letter to Carpianus, when he spoke of the section numbering in each gospel καθεξῆς προϊὼν δι’ ὅλου μέχρι τοῦ τέλους τῶν βιβλίων. ³⁸ See, e.g., the analysis of the flora and fauna in the two Ge‘ez gospelbooks in McKenzie and Watson, The Garima Gospels, pp. 94 101, 124 8, 141 4. Such decoration would seem to be exceptional insofar as Byzantine religious art ‘[i]n general . . . avoid[ed] metaphorical imagery drawn from nature’ (Henry Maguire, ‘Art and Text’, in Oxford Handbook of Byzantine Studies, ed. Robin Cormack, John F. Haldon, and Elizabeth Jeffreys (Oxford: Oxford University Press, 2008), p. 726). ³⁹ Theodor Klauser argued that during festivals in late antiquity the roofs of such tholoi, or ciboria, as he calls them, were decorated with plants, flowers, candles, and perhaps even stuffed birds, and that this is the source of these themes as they appear in Canon Tables artwork (Klauser, ‘Das Ciborium’, pp. 195 7). His comparative material to support the argument is, however, fairly limited. Cf. Bandmann, ‘Beobachtung’, p. 19. ⁴⁰ Underwood, ‘Fountain of Life’, pp. 46, 103; Nordenfalk, ‘The Apostolic Canon Tables’, p. 31; Bandmann, ‘Beobachtung’, p. 22; Mathews and Sanjian, Armenian Gospel Iconography, p. 171; Nersessian, Treasures From the Ark, p. 79; McKenzie and Watson, The Garima Gospels, pp. 94, 138.

244

The Eusebian Canon Tables

nationale, MS NAL 1203, fol. 3v) and the Gospels of St Médard de Soissons (Paris, Bibliothèque nationale, MS lat. 8850, fol. 6v; see fig. 38)—and in these manuscripts there is evident a further striking development: a low wall extends around the base of the circular structure, and in at least one of the images water fills the enclosed space.⁴¹ Commonly known as ‘the fountain of life’, this image, as argued in a seminal article by Paul Underwood, no doubt is intended to convey baptismal associations, which is particularly apt for the Godescalc Lectionary given that it was commissioned by Charlemagne in 781–3 to celebrate the baptism of his son Pepin.⁴² In the west, perhaps in the British Isles, another development arose around this time, when artists began including the symbols for each evangelist within the tables, often in place of the names of the individual gospels at the top of each column of numbers, as can be seen in the Book of Kells and some Carolingian manuscripts (see fig. 41).⁴³ Meanwhile in the east, probably in Constantinople, a further innovation took place, with someone adding busts of the twelve apostles to the arches throughout the series of Canon Tables (e.g., the so-called London Canon Tables: British Library, Add. 5111/1, fol. 10–11; see fig. 13 in chapter one), an innovation that Nordenfalk argued was modelled on Constantine’s mausoleum in the imperial capital, which supposedly had twelve columns on which were medallions of the apostles.⁴⁴ A further developmental trajectory may be seen in Syriac gospelbooks such as the Rabbula Gospels and Paris, Bibl. nat. syr. 33, which surround the architectural frames with miniature images from

⁴¹ See reproductions of the two images in McKenzie and Watson, The Garima Gospels, pp. 135, 137. ⁴² Cf. Underwood, ‘Fountain of Life’, pp. 64 6; McKenzie and Watson, The Garima Gospels, pp. 136 8. Underwood argued for the primacy of the ‘fountain of life’ interpretation, supposing that the fons vitae version found in the Gospels of St Médard de Soissons most closely approximates the original model and that the association with the Holy Sepulchre in the version found in the Armenian, Georgian, and Ge‘ez manuscripts is a subsequent development (p. 117). As a result, he referred to all the tholoi, both the western and eastern examples, as the ‘fountain of life’ (p. 107). Judith McKenzie rightly objected that to do so is ‘misleading’ in so far as different associations attach to the Carolingian versions on the one hand and the Armenian, Georgian, and Ge‘ez versions on the other (p. 138). It should be noted that Underwood was unable to take account of the two Garima gospelbooks, which predate all of his eastern examples by at least two centuries. ⁴³ Cf. Carl Nordenfalk, Celtic and Anglo Saxon Painting: Book Illumination in the British Isles, 600 800 (New York: George Braziller, 1977), plate 40. On the origins of this motif, see Netzer, ‘The Origin’, as well as the section on Ailerán above in chapter six. ⁴⁴ Cf. Nordenfalk, ‘The Apostolic Canon Tables’. See the images of the London Canon Tables at McKenzie and Watson, The Garima Gospels, pp. 55, 149. Against Nordenfalk, however, it should be noted that Eusebius’ account of Constantine’s mausoleum is riddled with obscurity, as pointed out by Jaś Elsner, ‘From the Culture of Spolia to the Cult of Relics: The Arch of Constantine and the Genesis of Late Antique Forms’, Papers of the British School at Rome 68 (2000): pp. 157 8.

Seeing the Salvation of God

245

Fig. 41. Canon II in the Gospels of St Médard de Soissons (before 827). Note the evangelists’ heavenly beasts, one above each of the columns, corresponding to the numbers below: Matthew, Mark, and Luke. Paris, Bibliothèque nationale de France, MS lat. 8850, fol. 9r

246

The Eusebian Canon Tables

the life of Christ and depictions of prophets and apostles (see fig. 8 in the introduction).⁴⁵ The Rabbula Gospels also uniquely incorporate images of the evangelists themselves within the Canon Tables, alongside the columns of numbers, as though the authors were in the act of writing the sections of text represented by the string of numbers next to them (see fig. 42).⁴⁶ A final remarkable feature of the Rabbula Gospels is its page presenting portraits of Ammonius and Eusebius (fol. 2r; see fig. 19 in chapter 2), placed opposite the Letter to Carpianus. No other contemporary manuscript has such a dual portrait, though a handful of much later Greek gospelbooks contain opening sets of portraits, variously labelled either Eusebius–Ammonius or Eusebius–Carpianus (see, e.g., fig. 20 in chapter 2).⁴⁷ Giving a fresh assessment of prior scholarship on the artistic adornment of the Eusebian Canon Tables is beyond the scope of this monograph, but it is worth noting that strictly speaking there is no way of knowing with certainty that many of the elements, such as the tholos, go back to Eusebius himself, a point about which prior scholarship has not always expressed sufficient reserve. Moreover, the attempt to narrow down the symbolic import of the tholos to a specific physical structure, whether Constantine’s aedicula at the Holy Sepulchre, the similar one at the Church of the Nativity in Bethlehem, or the ciborium of the Caesarean basilica, is probably misguided.⁴⁸ Indeed, one of the striking aspects of a tholos image is precisely its lack of any unambiguous distinguishing features, aside from the example in the Adishi Gospels, which includes additional elements to clarify the image’s intended referent. Instead, Günter Bandmann was probably closer to the mark when he argued that the tholos is ‘an ideal architectural ensemble’ (‘ein ideales architektonisches Ensemble’), that simultaneously represents the Holy Sepulchre, the fountain of life, and the altar in a church.⁴⁹ Or, one could at least say that at various points over the centuries viewers probably saw all of these associations in ⁴⁵ On this development, see Nordenfalk, Die spätantiken Kanontafeln, pp. 239 54. For an overview of the images in the Rabbula Gospels, see Jules Leroy, Les manuscrits syriaques à peintures conservés dans les bibliothèques d’Europe et d’Orient, Bibliothèque archéologique et historique 77 (Paris: Librairie orientaliste Paul Geuthner, 1964), pp. 139 97, and on syr. 33 in Paris, see pp. 198 206. Although note that, according to Bernabò, Il Tetravangelo di Rabbula; Bernabò, ‘The Miniatures in the Rabbula Gospels’, the famous marginal illustrations of the Rabbula gospels were not original to the manuscript but come from a separate sixth century Syriac copy of the gospels. ⁴⁶ Cf. Reudenbach, ‘Der Codex als heiliger Raum’, p. 63. ⁴⁷ Cf. Hugo Buchthal, ‘Studies in Byzantine Illumination of the Thirteenth Century’, Jahrbuch der Berliner Museen 25 (1983): pp. 33 6, referring to Athos, Dionysiou 4; Parma, Palat. gr. 5; Bodleian, Clarke 10; Bodleian, Codex Ebnerianus; Vani Gospels; the last one being a Georgian copy of the gospels. The Abba Garima III Gospels have, in addition to the expected four evangelist portraits, an unlabelled fifth portrait that very likely represents Eusebius. Cf. McKenzie and Watson, The Garima Gospels, pp. 75, 153 5, plate 2. ⁴⁸ I am grateful to Jaś Elsner for impressing these two points upon me. ⁴⁹ Bandmann, ‘Beobachtung’, pp. 22, 28.

Seeing the Salvation of God

247

Fig. 42. Portraits of Matthew (right) and John (left) alongside Canon VII in the Rabbula Gospels (sixth c.). Florence, Biblioteca Medicea Laurenziana, MS. Plut. 1.56, fol. 9v

248

The Eusebian Canon Tables

different versions of the image, and would probably have been comfortable moving back and forth between them, since these themes were often understood to be theologically associated with one another.⁵⁰ The two points from this survey that are particularly important for the present chapter are 1) the basic assumption that the ornamentation is not merely decorative but bears symbolic significance; and 2) the hypothesis that viewing the decorated pages of a set of Canon Tables was intended as a defined progression through an imagined architectural space leading to some sort of culmination at the end.⁵¹ Both of these conclusions suggested by modern scholarship are corroborated by the Armenian sources to be considered next.

TWO ARMEN IAN I NTERPRETATIONS O F C A N O N T A B L E DE C O R A T I O N

Introduction to Step‘anos of Siwnik‘ and Nerses Šnorhali The two Armenian texts I have in mind are little known and so merit some introduction. The first is attributed to Step‘anos, bishop of Siwnik‘ in northeastern Armenia in the early eighth century.⁵² Step‘anos, the son of a priest, was probably born around 685 and received training in both the Armenian Catholicosate in the capital city of Duin and also at the Monastery of ⁵⁰ For example, Eusebius referred to the Holy Sepulchre as a fountain of living water (l.C. 9.15) and the Patriarch Germanus, in his explanation of the divine liturgy, asserted that the altar of the church is Christ’s tomb, and that the ciborium above the altar represents the place where Christ was crucified, where he was buried, and also where he was raised (hist. 4 5 (Meyendorff, pp. 58 9)). ⁵¹ These two conclusions resonate with an excellent recent article that appeared too late for me to incorporate into my own analysis: Rolf Strom Olsen, ‘The Propylaic Function of the Eusebian Canon Tables in Late Antiquity’, Journal of Early Christian Studies 26 (2018): pp. 403 31. Strom Olsen concluded, ‘framing the canon tables within an elaborate propylaic design that designates passage through the interior space of the Church seems consistent with the need both to honor the significance of the achievement of the apparatus for the text, as well as providing a visually intelligible way of capturing their meaning within or as part of the text itself ’. ⁵² The following overview of Step‘anos’ life and literary activity follows the survey of the evidence in Michael Daniel Findikyan, The Commentary on the Armenian Daily Office By Bishop Step‘anos Siwneci‘ (735): Critical Edition and Translation With Textual and Liturgical Analysis, Orientalia Christiana Analecta 270 (Rome: Pontificio istituto orientale, 2004), pp. 43 57. Brief surveys can also be found in Michael Papazian, ‘Origen’s Commentaries as Sources for Step’anos Siwnec’i’s Commentary on the Gospels’, Le Muséon 117 (2004): pp. 508 11; R. W. Thomson, ‘Is There an Armenian Tradition of Exegesis?’, Studia Patristica 41 (2006): pp. 103 7. At one time it was supposed that there were in fact two authors with this name who had been confused with each other one a fifth century hymnographer and the other the early eighth century figure relevant to the present study but see the rejection of this notion in Findikyan, The Commentary on the Armenian Daily Office, p. 48; Anna Arevshatyan, ‘L’identité de Step‘anos Siwnec‘i’, Revue des Études Arméniennes 30 (2005): pp. 401 10.

Seeing the Salvation of God

249

Siwnik‘, which was reputed to be the ‘head and administrator of all Armenian Vardapets’.⁵³ In this monastery he was ordained as a priest and became a teacher of biblical exegesis, eventually returning to the Catholicosate at Duin. Step‘anos made trips to Athens, Constantinople, and Rome, becoming acquainted in the process with Germanus I, patriarch of Constantinople in 715–30, and an imperial official named David with whom he translated into Armenian a number of Greek patristic works such as the ps-Dionysian corpus,⁵⁴ Gregory of Nyssa’s De opificio hominis, and Nemesius of Emesa’s De natura hominis. Step‘anos also left behind a number of original writings, including homilies; doctrinal letters; commentaries on Ezekiel, Job, and the gospels; hymns; and a commentary on the daily office entitled How Our Prayers are Organized.⁵⁵ His exposition of the gospels has the distinction of being the earliest surviving gospel commentary in Armenian.⁵⁶ Few of these works have received much critical analysis, but recently a convincing argument has been made that Step‘anos’ gospel commentary was heavily reliant on Origen’s commentaries and homilies,⁵⁷ and his Commentary on Job seems to have used Olympiodorus’s commentary on the same book as well as the Greek Physiologus,⁵⁸ supporting the claims of the historical sources that he was well acquainted with Greek literature. Step‘anos’ short exposition of Canon Table decoration survives in a number of manuscripts and a version found in a twelfth-century gospelbook (Freer Art Gallery, MS F1950.3) was published in 1963 by Sirarpie Der Nersessian.⁵⁹ Der Nersessian hesitated to state definitively that the text belonged to Step‘anos but found no reason to doubt the attribution.⁶⁰ Michael Daniel Findikyan, who recently made a thorough philological and liturgical analysis of Step‘anos’ exposition of the daily office, observed that ‘[t]he highly allegorical style, and ⁵³ This is the designation that the thirteenth century historian Step‘anos Orbelean Siwnik‘i says was given to the monastery by the fifth century saints Mesrop and Sahak. Cited from Findikyan, The Commentary on the Armenian Daily Office, p. 44. ‘Vardapet’ is a teaching office in the Armenian church. ⁵⁴ Cf. Robert W. Thomson, The Armenian Version of the Works Attributed to Dionysius the Areopagite, CSCO 488, Scriptores Armeniaci 17 (Louvain: Peeters, 1987). ⁵⁵ On the latter text, see now the edition and translation of Findikyan, The Commentary on the Armenian Daily Office. ⁵⁶ Thomson, ‘Is There an Armenian Tradition’, p. 101. See the recent translation of this commentary in Michael B. Papazian, Step‘anos Siwnec‘i: Commentary on the Four Evangelists (New York: SIS Publications, 2014). ⁵⁷ Papazian, ‘Origen’s Commentaries’. ⁵⁸ Michael Papazian, ‘The Commentary on Job of Step‘anos of Siwnik‘ and its Sources’, Le Muséon 130 (2017): pp. 343 63. ⁵⁹ Der Nersessian, Armenian Manuscripts in the Freer Gallery of Art, pp. 103 4. It is this edition that will be used here. For a description of the manuscript, see pp. 7 17. I have not been able to examine any of the extant manuscripts of Step‘anos’ treatise myself. Der Nersessian does not mention any title for the text in the manuscript, so for the sake of convenience I will call it Commentary on the Xorank‘. On the term xorank‘, see the next section of this chapter. ⁶⁰ Der Nersessian, Armenian Manuscripts in the Freer Gallery of Art, pp. 16 17. The attribu tion to Step‘anos is also accepted in Mathews and Sanjian, Armenian Gospel Iconography, p. 170.

250

The Eusebian Canon Tables

the vocabulary of [the interpretation of the Canon Tables] correspond quite well with the interpretation of the liturgical rituals and texts’ in the exposition of the daily office.⁶¹ In fact, given that the gospelbook was a central element in the Armenian liturgy,⁶² it is possible to view the exposition of Canon Table artwork as part of Step‘anos’ larger project of providing a symbolic interpretation of the rituals of the Armenian church, including not only the exposition of the daily office but also an Explanation of the Rite of the Foundation of a Church.⁶³ There is, therefore, no reason to doubt, and contrariwise, good reason to suppose that the explanation of Canon Table symbolism that has come down under the name of Step‘anos indeed goes back to the bishop and exegete of the early eighth century. It is striking to observe that both the earliest surviving exposition of the gospels and the earliest surviving explanation of Canon Table decoration come from the same figure. We do not know what sources prior to Step‘anos have not survived, but, based on the evidence we have, it seems that from the outset of the Armenian exegetical tradition the Canon Tables were seen as so integral a component of the fourfold gospel that an exposition of the gospel text also required an explanation of the gospel paratext. Nevertheless, something unfortunate appears to have happened in the transmission of Step‘anos’ work, because the style changes abruptly after the discussion of the fifth Canon to very brief summary remarks about the final five Canons. Thomas F. Mathews and Avedis K. Sanjian are probably correct in surmising that the latter half of the text was lost at some point and a later scribe tried to supply what was missing with these few short remarks.⁶⁴ The second commentary is attributed to Nerses Šnorhali (i.e., ‘the Graceful’), who was born in 1102 and served as Catholicos of the Armenian Church

⁶¹ Findikyan, The Commentary on the Armenian Daily Office, p. 54. More specifically, both the commentary on the daily office and the exposition of the Canon Tables comment upon the symbolic meaning attached to specific colours, and both draw attention to the way in which the four elements are symbolically represented. See pray. or. (SS L) VII.3 (Findikyan, text on pp. 99 100; trans. on p. 143). ⁶² See pray. or. (SS L) VII.6, where Step‘anos comments, ‘At the elevation of the gospel, with the eyes of the soul we see the Son of God seated on a throne high and lifted up’ (Findikyan, text on p. 101; trans. on p. 145). The defence of images in the Armenian tradition written in the early seventh century by Vrt‘anes K‘ert‘oł mentions the practice of prostrating before the gospelbook in the liturgy. Vrt‘anes describes a codex painted with gold and silver on purple parchment, bound with ivory covers (yał; text in Tourian, pp. 61 2, ln. 319 35; French translation in Der Nersessian, p. 65). Presumably the manuscript he had in mind would have been equipped with Canon Tables and other illustrations, though he does not explicitly refer to any artwork. On Vrt‘anes’ treatise, see also Thomas F. Mathews, ‘Vrt‘anēs K‘ert‘oł and the Early Theology of Images’, Revue des Études Arméniennes 31 (2008 9): pp. 101 26. ⁶³ On the latter, see Findikyan, The Commentary on the Armenian Daily Office, pp. 54 5. It is surely not a coincidence that the Patriarch Germanus, whom Step‘anos met in Constantinople, also wrote an influential commentary on the divine liturgy at around the same time. ⁶⁴ Mathews and Sanjian, Armenian Gospel Iconography, p. 207 n. 6.

Seeing the Salvation of God

251

from 1166 until 1173, succeeding his brother Gregory III in that office.⁶⁵ As Catholicos, he engaged in ecumenical dialogue with the Byzantine church, and is even said to have reached an understanding with Byzantine representatives in which the two sides acknowledged each other as orthodox. However, he died before a council of Armenian bishops could meet to ratify the agreement. Nerses left behind numerous treatises and poems, including a great number of hymns, and for this reason he is said to have been for the Armenians what Ephrem was for the Syriac church and Romanos was for the Byzantines.⁶⁶ Among his surviving works is an exposition of the Canon Tables, which in fact forms the preface to his Commentary on the Gospel of Matthew.⁶⁷ Writing nearly half a millennium after Step‘anos, Nerses presents the ideas put forth in his commentary as those of the ‘fathers and brothers’ (§17), implying that by this point there was in the Armenian tradition an established approach to the interpretation of Canon Table artwork that stretched back centuries, a supposition confirmed by the parallels between his exposition and that of the much earlier bishop of Siwnik‘.⁶⁸ In fact, the expositions by Step‘anos and Nerses are merely the two most accessible examples of a larger corpus of commentaries on Canon Table artwork, which seems to be a uniquely Armenian genre. A collection of thirteen such texts was published in 1995 by Vigen Ghazaryan including a modern Armenian translation on facing pages,⁶⁹ and the present chapter makes no claim to be a comprehensive treatment of these sources. Rather, in keeping with the aim of this monograph, the question at hand is how

⁶⁵ For a brief summary of Nerses’ life, see Isaac Kéchichian, Nersès Šnorhali. Jésus Fils Unique du Père, SC 203 (Paris: Les Éditions du Cerf, 1973), pp. 13 18. ⁶⁶ Kéchichian, Jésus Fils Unique du Père, p. 15. ⁶⁷ Text published in Nerses Šnorhali and Yovhannes Erznkac‘i, Meknut‘iwn Surb Awetaranin or ěst Matt‘ēosi (Constantinople, 1825), pp. 5 12. This is the edition used here, and I will cite both the section number of the commentary and the page and column of the edition. Both the treatises of Step‘anos and of Nerses were translated by James R. Russell and included as ‘Appendix D’ in Mathews and Sanjian, Armenian Gospel Iconography, pp. 206 11. Here I follow Russell’s translation, with some modifications. ⁶⁸ So also Mathews and Sanjian, Armenian Gospel Iconography, p. 171. pace Underwood, ‘Fountain of Life’, p. 79, who, seemingly unaware of the treatise of Step‘anos, asserted that ‘there is a complete absence of extant eastern sources from which [Nerses] may have derived some of his ideas regarding the iconography of the Canon Tables’. ⁶⁹ Ghazaryan, Commentaries of Canon Tables. Ghazaryan provided an English summary of his analysis on pp. 180 9, which focused primarily on the aesthetic theory that can be derived from these texts, particularly with reference to their ‘doctrine of colour’. An expanded and revised version was published by Ghazaryan as ՄԵԿՆՈՒԹԻՒՆՔ ԽՈՐԱՆԱՅ = The Commen taries of Canon Tables (Holy Etchmiadzin, 2004), with an English summary on pp. 430 40. I am grateful to Michael Stone for drawing Ghazaryan’s work to my attention and to Fr Shahe Ananyan for providing me with a copy of the 2004 publication. See also Der Nersessian, Armenian Manuscripts in the Freer Gallery of Art, pp. 100 1, who mentioned the rhymed adaptations composed by Mik‘ayel of Sebastia and Step‘annos Dzik‘ from the seventeenth century.

252

The Eusebian Canon Tables

the artwork surrounding Canon Tables could function as a paratext to the fourfold gospel, and the two Armenian treatises examined below provide at least one answer to this question. The degree to which they can be taken as representative of the wider corpus to which they belong is a question that the present study does not address, but it is hoped that this chapter might make better known this fascinating genre of literature and so inspire further studies on the topic.

The Function of Architecture and Ornamentation according to Step‘anos and Nerses First, a word must be said about the terminology that these two authors use to refer to their subject matter, for it reveals that the aspect of the paratextual system they took to be most significant was the architectural frames housing the numeric tables. In both the Latin and Syriac traditions, the ten tables of Eusebius were referred to using transliterated forms of the Greek word κανών, and both Step‘anos and Nerses on occasion use a similar Armenian transliteration.⁷⁰ However, the word that they regularly use when they are discussing the decorated pages is խորան (transliteration: xoran), a term that unmistakably refers to the architectural frames. Hence, James R. Russell’s English translation of Step‘anos and Nerses’ treatises is misleading in rendering xoran throughout as ‘Canon Tables’ because these two authors are not referring to the numeric matrices, the κανόνες, but rather to the pillars and arches around them. Because խորան was something of a technical term in Armenian theology, with an especially rich and diverse range of meaning, I will here simply transliterate it as xoran (plural: xorank‘). The polyvalent possibilities of the term խորան were exploited to great effect in the Armenian tradition.⁷¹ In its most mundane sense, it can mean ‘pavilion’ or ‘tent’, and so a խորանաբնակ is a ‘nomad’ and a խորանագործ a ‘tentmaker’.⁷² However, the word was also used in a variety of other ways. In one of the earliest dogmatic texts written in Armenian, the fifth-century Teaching of Saint Gregory, heaven is said to be ‘domed’ (խորանարդ), a usage that probably is related to the fact that խորան was also used in the Armenian translation of

⁷⁰ Step‘anos uses the word կանովն three times (comm. xor. 1; 4; 6; Der Nersessian, pp. 103 4), and Nerses does once, spelled կանօն (comm. Mt., prol. 5; p. 6b). ⁷¹ Cf. R. W. Thomson, ‘Architectural Symbolism in Classical Armenian Literature’, Journal of Theological Studies 30 (1979): p. 104: ‘Khoran . . . is one of the key terms used in Armenian architectural symbolism’. The following survey of the term is indebted to Thomson’s article. See also Drost Abgaryan, ‘The Reception of Eusebius of Caesarea (ca. 264 339) in Armenia’, pp. 226 7. ⁷² See Bedrossian, New Dictionary, s.v. խորան; խորանաբնակ; խորանագործ.

Seeing the Salvation of God

253

the LXX in passages such as Isaiah 40:22 and Psalms 103:3.⁷³ Nerses Šnorhali himself, in another work, describes heaven as a ‘hemisphere’ that is like a խորան, ‘self-moving and unsupported’.⁷⁴ Moreover, in keeping with the usage of σκήνη in the LXX, Nerses, in his exposition of the Canon Tables, similarly uses խորան to refer to the tabernacle in which Israel met with God in the wilderness.⁷⁵ Finally, as highlighted by R. W. Thomson, the term was also ‘widely used in descriptions of the symbolism of a church building’, to refer sometimes to the central dome in a church, sometimes to the ciborium above the altar resting on four pillars.⁷⁶ The Armenian interest in the term may derive in part from the Armenian translation of Philo’s corpus, for in his Questions and Answers on Exodus he provided an allegorical interpretation of Moses’ tabernacle, and the Armenian translator of the treatise referred to the structure as a խորան.⁷⁷ Interestingly, in this treatise Philo also describes the Holy of Holies at one end of the tabernacle as having four columns and a hanging curtain, which is reminiscent of the tholos image discussed above.⁷⁸ Thus, the term խորան became a sort of nodal point in Armenian theology, drawing together a wide range of themes and texts that were thought to be symbolically linked together. Clearly the theological significance of խորան is not due solely to the architectural ornamentation that adorned Canon Tables, but when Step‘anos and Nerses focus on the architectural frames as the basis of their interpretation, they are no doubt tapping into this wider associative field. In fact, it is precisely this theological polyvalence that is the most important assumption in our two texts, since their authors allude to all of the aforementioned themes and Nerses even justifies his hermeneutical approach to Canon Table ornamentation by providing a fanciful etymology for the term: ‘And they are called xorank‘ because they are all mystical’ (եւ խորանք ասին քանզի խորհրդականք են ամենեքեան).⁷⁹ His argument rests on the similarity between the words խորան and խորհրդական (‘mystical’), and his point seems to be that the very fact that these structures take the form of arches supported by pillars implies that they are susceptible to

⁷³ Thomson, ‘Architectural Symbolism’, p. 104. Cf. Teaching of Saint Gregory §259 (Thomson, p. 41). ⁷⁴ Citation from Thomson, ‘Architectural Symbolism’, p. 105. ⁷⁵ Nerses, comm. Mt., prol. 4 (p. 6b). ⁷⁶ Thomson, ‘Architectural Symbolism’, pp. 106 13, citing examples in Agathangelos, John of Odzun, Stephen of Tarōn, and Kirakos, among others. ⁷⁷ Philo, qu. Ex. 2.83 (Aucher, pp. 524 5; LCL 401, pp. 132 3). On the cosmological interpretation of the tabernacle offered by Philo and Josephus, see Craig R. Koester, The Dwelling of God: The Tabernacle in the Old Testament, Intertestamental Jewish Literature, and the New Testament, The Catholic Biblical Quarterly Monograph Series 22 (Washington, DC: Catholic Biblical Association of America, 1989), pp. 59 67. ⁷⁸ Philo, qu. Ex. 2.93 (Aucher, p. 529; LCL 401, p. 141). For a discussion of this passage, see Grigoryan, ‘The Roots of Tempietto’, pp. 18 19. ⁷⁹ Nerses, comm. Mt., prol. 6 (p. 7a).

254

The Eusebian Canon Tables

the kind of symbolic interpretation to which he subjects them. In other words, for these two authors, those aspects of the Canon Tables that might strike one as purely ornamental, and so non-functional, are in fact the key to discerning the proper use of the Eusebian paratext. The basic approach in the expositions of Step‘anos and Nerses is the same. Both regard the pages of a gospelbook containing decorated Canon Tables as symbolic, and so seek to decode for the reader the meaning of the various elements of the artistic scheme, progressing through the ornamented pages from the first xoran all the way through to the tenth and providing an interpretation of each page in sequence. Once again, although they reference ten xorank‘, they do not have in mind the ten ‘Canon Tables’, as I have called them throughout this study. In fact, because the ten Canon Tables vary considerably in length, in the standard Armenian version of Eusebius’ paratext some of the shorter tables are presented together on a single page and so are housed within a single xorank‘. In other words, one xoran does not correspond with one table. Nevertheless, the overall number of xorank‘ does match the number of tables, and this correspondence is not accidental. Step‘anos and Nerses seem to have based their expositions on actual copies of Canon Tables they had before them,⁸⁰ and the number of xorank‘ they allude to is supported by the manuscript tradition. As first recognized by Nordenfalk, Armenian manuscripts with Canon Tables demonstrate a tendency to retain a total of ten pages for the entire prefatory sequence, a feature that he plausibly supposed reflected the Eusebian archetype. So, for example, in the Etchmiadzin gospels, the recto and verso of the first folio contain the Epistle to Carpianus, housed in two architectural frames; the ten numeric tables then follow, distributed under seven xorank‘ over the next seven pages; and finally the sequence concludes with the tholos image on the tenth page.⁸¹ When a later tradition emerged in ⁸⁰ Mathews and Sanjian, Armenian Gospel Iconography, p. 172: ‘It is clear from the specificity of their references that both Step‘anos Siwnec‘i and Nersēs Šnorhali had before their eyes one (or more) actual sets of canon tables; it is also clear from the immense variety of surviving sets that no sequence of symbols was ever regarded a fixed and canonical’. Cf. Der Nersessian, Armenian Manuscripts in the Freer Gallery of Art, p. 17. On Canon Tables in Armenian manuscripts, see also Klemm, ‘Die Kanontafeln’; Helen C. Evans, ‘Canon Tables as an Indication of Teacher Pupil Relationships in the Career of T’oros Roslin’, in Medieval Armenian Culture, ed. Thomas J. Samuelian and Michael E. Stone, University of Pennsylvania Armenian Texts and Studies (Chico, CA: Scholars Press, 1984); Mathews and Sanjian, Armenian Gospel Iconography, pp. 166 76; Kouymjian, ‘Armenian Manuscript Illumination’; Nersessian, Treasures From the Ark, pp. 79 82. A number of manuscripts containing Canon Tables are included in the catalogue accompanying the recent special exhibit at the Metropolitan Museum of Art in New York: Helen C. Evans, ed, Armenia: Art, Religion, and Trade in the Middle Ages (New York: The Metropolitan Museum of Art; distributed by Yale University Press, 2018). ⁸¹ Nordenfalk, Die spätantiken Kanontafeln, pp. 65 72, 102 3. Dickran Kouymjian argued that the Canon Tables in the Etchmiadzin Gospels are in fact ‘[i]n every respect . . . anomalous and, therefore, very hard to accept as a prototype’, but nevertheless acknowledged that ‘what was important for the Armenian scribes or artists responsible for the Eusebian apparatus was the number ten’ (Kouymjian, ‘Armenian Manuscript Illumination’, p. 1037).

Seeing the Salvation of God

255

Armenian Cilicia that dropped the final tholos, the preceding numeric tables were then redistributed over eight pages rather than seven to retain a total of ten pages for the entire sequence, as can be seen, for example, in the Gladzor Gospels.⁸² Because they refer to ten xorank‘, Step‘anos and Nerses must have been working with versions of the paratext that followed the standard Armenian practice of presenting the sequence across ten pages. Given his earlier date, Step‘anos was presumably using a model that was laid out according to the older pattern, with a tholos on the final page, but because the latter half of his treatise is lost, we do not have any comment from him about the concluding image. Nerses, on the other hand, was almost certainly working from a manuscript that included the later Cilician version without the tholos. This is suggested not only by his date but also because he points out that the tenth xoran stands facing ‘opposite’ (հանդէպ) the ninth.⁸³ The older Armenian version, found in the Etchmiadzin Gospels, begins on a recto and so ended on a verso, in which case the tenth page of the sequence stands alone, opposite whatever text or image follows the sequence of Canon Table pages. However, with the Cilician version a decision was made to begin the series on a verso, such that the open codex always presented two pages of decoration, first xorank‘ one and two, then three and four, and so on up to nine and ten.⁸⁴ As a result, when Nerses talks about xoran number ten, he is referring not to the tholos image, but rather to the same sort of architectural frame that is used throughout the series. While modern viewers can easily recognize the aesthetic appeal of the rich ornamentation of a medieval Armenian gospelbook, for these two authors such visual stimuli were communicative of divine truth. Step‘anos announces this theme at the outset of his exposition: Worthy of great admiration is the mighty magnificence of the words beautifully and usefully adorned, for with lofty speech these varicoloured houses of the ten canons, adorned with pictures, with multi storied canopies, with different colours, and with paints of varying hues, announce the prophecy of the four evangelists, describing properly the participation of the celestial body.⁸⁵

Here Step‘anos initially mentions the ‘mighty magnificence of the words’ on the page before him, presumably referring thereby to the textual content of the Canon Tables, or perhaps to the gospel text itself, but he then asserts that it is the ‘houses of the ten canons’ (տասնեցունց կանովնացն . . . տունք) that ‘announce the prophecy of the four evangelists’. The ‘houses’ must be a

⁸² ⁸³ ⁸⁴ ⁸⁵

Mathews and Sanjian, Armenian Gospel Iconography, p. 167. Nerses, comm. Mt., prol. 15 (p. 10b). Mathews and Sanjian, Armenian Gospel Iconography, p. 167. Step‘anos, comm. xor. 1 (Der Nersessian, p. 103).

256

The Eusebian Canon Tables

reference to the architectural frames, since they are described as having ‘multistoried canopies’, the word translated here as ‘canopies’ (խորանայարկ) being cognate to xoran. Thus, although Step‘anos makes a passing allusion to the text in the codex before him, it is the xorank‘ that he says are speaking to the viewer. The architectural frames, with their rich decoration, have a message to proclaim, and the content of their discourse is revealed in the final phrase: ‘the participation of the celestial body’ (երկնային մարմնոյն մետաղփսն), surely an allusion to the Incarnation.⁸⁶ The message of the xorank‘ is, therefore, the redemption brought by Christ as it was enacted, and continues to be effective, in human history. Nerses holds to the same basic notion, though he provides it with a more elaborate theoretical basis. Everything in the world, he asserts, can be divided into two categories: ‘important things and pleasures’ (ի կարեւորն եւ ի վայելմունս). Into the former category fall such mundane realities as air, earth, light, bread, and water—in other words, items that are necessary for existence. In the latter category are unnecessary but delightful things such as wine, fruit, spices, fine cuisine, colours, and pleasing sounds—everything that humans experience as ‘luxuries of the senses’ (զգայութեանց . . . հեշտալիք). These latter items are not strictly speaking necessary, but to the properly trained person they have ‘great utility’ (մեծապէս օգուտ) in so far as they become the means by which one can progress from the merely mundane to higher realities: ‘by this manifest colour, taste, smell, hearing and the rest we ascend to the spiritual, and to the rational enjoyment of the good tidings of God’ (երեւելի գունովս, եւ համովս, եւ հոտով, եւ լրով, եւ հանգստեամբ, ամբառնամք ի հոգեւորն եւ յիմանալին վայելչուի յաւէտսն ա՟յ).⁸⁷ Nerses explains that it was for this reason that God richly decorated the tabernacle and the temple for Israel, ‘giving the lover of material things understanding of the heavenly by beguiling (him) with earthly treasures’. Aware of the same principle, the ‘compilers and founders of the gospel’ (ցանկիչք եւ հիմնադիրք աւետարանիս) similarly ‘illustrated (it) with luxurious herbs and multicoloured flowers and various inventions, as a foretelling of earthly and heavenly things and a service to the virtues, and through these the text before us gives us to understanding also what the spiritual pleasures and imperishable beauties are’.⁸⁸ At the end of his exposition Nerses returns to this theme, once more stating that the ‘first fathers’ did not decorate the pages of the Canon Tables ‘in vain or without meaning’ but rather as ‘cleansing baths of the sight and hearing for those approaching the soaring peaks of God’.⁸⁹ Thus, for both Step‘anos and Nerses, the decoration adorning the Canon Tables is not merely incidental, nor is it intended for mere enjoyment. Rather, the pleasure that these pages produce in ⁸⁶ As recognized by Mathews and Sanjian, Armenian Gospel Iconography, p. 206. ⁸⁷ Nerses, comm. Mt., prol. 3 (p. 6a). ⁸⁸ Nerses, comm. Mt., prol. 4 (p. 6b). ⁸⁹ Nerses, comm. Mt., prol. 17 (p. 12a b).

Seeing the Salvation of God

257

their viewers is the means by which one should contemplate transcendental realities, and this pleasure-induced contemplation is the paratextual process the reader must undergo prior to meeting the sacred text. Thomas F. Mathews and Avedis K. Sanjian, who have written the most extensive treatment of these two treatises, have argued that they present a uniquely Armenian tradition in several respects, one of which is the way in which Nerses articulates an aesthetic that includes ‘the frank acceptance of the sensuous as something good in itself and therefore worthy of serious attention of the educated or initiate’.⁹⁰ However, the basic thesis of both our authors may be found in a wide range of late antique and medieval Christian sources. To cite but one example particularly relevant to the present discussion, ps-Dionysius argues that for humans, unlike incorporeal beings, ‘it is by way of the perceptible images that we are uplifted as far as we can be to the contemplation of what is divine’.⁹¹ It is specifically the divine liturgy that psDionysius has in mind, and he further states that its ‘sacred symbols are actually the perceptible tokens of the conceptual things. They show the way to them and lead to them, and the conceptual things are the source of the understanding underlying the perceptible manifestations of hierarchy’.⁹² When we recall that Step‘anos himself translated the ps-Dionysian corpus into Armenian, we should then hardly be surprised to find that his understanding of artwork is in keeping with the Areopagite. What would seem to make the treatises of Step‘anos and Nerses unique is not their understanding of the purpose of beautiful and otherwise pleasurable things, but rather the fact that they apply this understanding to the interpretation of Canon Tables ornamentation, something that is not done in any other texts that have so far come to light. It is implicit in what has been said so far, but should be stated explicitly, that Nerses thinks Canon Table decoration (and presumably the manner of interpreting it that he offers) goes back to Eusebius himself, and that it in fact came to Eusebius via a divine revelation. The term that I, following Russell, have translated two paragraphs above as ‘compilers’ (ցանկիչք) of the gospel comes from the verb ցանկեմ, which means ‘to surround with hedges, to hedge or wall in’, but can also mean ‘to divide into chapter and verse’.⁹³ It is, therefore, a reference to Eusebius and Ammonius as those who had not only ‘divided’ the gospel into sections but also ‘illustrated’ it. Early in his exposition Nerses makes this even clearer, explaining that a certain Carpianus had asked Eusebius to ⁹⁰ Mathews and Sanjian, Armenian Gospel Iconography, p. 172. ⁹¹ Ps Dionysius, e.h. 1.2 (Heil and Ritter, p. 65; Luibheid and Rorem, p. 197). ἡμεῖς δὲ αἰσθηταῖς εἰκόσιν ἐπὶ τὰς θείας ὡς δυνατὸν ἀναγόμεθα θεωρίας. ⁹² Ps Dionysius, e.h. 2.3.2 (Heil and Ritter, p. 74; Luibheid and Rorem, p. 205). τὰ μὲν αἰσθητῶς ἱερὰ τῶν νοητῶν ἀπεικονίσματα καὶ ἐπ’ αὐτὰ χειραγωγία καὶ ὁδός, τὰ δὲ νοητὰ τῶν κατ’ αἴσθησιν ἱεραρχικῶν ἀρχὴ καὶ ἐπιστήμη. ⁹³ Bedrossian, New Dictionary, s.v. ցանկեմ.

258

The Eusebian Canon Tables

confirm for us unerringly the correspondence of the gospels: where the four assent, and where the three, and in what place the two, and where they speak alone and individually. And list the number of their verses, lest they be stolen by heretics, and also so that we may with facility separate each without confusion.⁹⁴

Nerses explains that in response Eusebius ‘wrote in ten Canons (կանօնս) the reliable knowledge of them’, but then further explains that ‘by a mystery he saw (տեսեալ) ten, which is a holy number and a gift of God’.⁹⁵ The contrast in Eusebius’ two activities is telling. Nerses acknowledges that Eusebius ‘wrote’ the ten Canons, fulfilling Carpianus’ request, but then shifts to a visionary mode of reception when he says Eusebius ‘saw’ the mystical number ten, implicitly elevating Canon Table image over Canon Table text. As proof of the fittingness of the number ten, he adduces, among other things, the decalogue, Aristotle’s ten categories, the ten sayings of the Paternoster, and ten clauses of the Nicene Creed.⁹⁶ Given the fact that Eusebius seems to have chosen for his paratext to include precisely ten tables, most likely for the symbolic significance of this number (see chapter two), Nerses’ interpretation is in keeping with his original design. The decalogue is the most important item in Nerses’ list,⁹⁷ for he returns to it at the end of his exposition as a part of a complex metaphor in which he presents the reader of the gospel as standing at the foot of the ‘soaring peaks of God’, at the base of the mountain which is the ‘preface to the gospel’ (ի ստորին նախադրութեան աւետարանիս). The preface, or base of the mountain, is of course the Canon Tables. The reader, then, is about to ascend the mountain of God ‘in the manner of Moses, into the mists of unknown multitudes, into the depths of this book of the gospel’ (ի խորս գրոց աւետարանիս). Yet, although Nerses presents the act of reading the fourfold gospel as akin to Moses’ ascent, he implies that Eusebius has already made this journey, bringing back with him the Canon Tables. God, he says, ‘has assigned (նշանակեաց) this gospel to us in a figure (տպաւորութեամբ) as ten xorank‘, instead of the ten commandments which are on the divine tablets’.⁹⁸ Eusebius, then, has received the ten xorank‘ from God and included them at the front of the codex as the threshold across which the reader passes to enter

⁹⁴ Nerses, comm. Mt., prol. 5. ⁹⁵ Nerses, comm. Mt., prol. 5 (pp. 6b 7a). ⁹⁶ Nerses, comm. Mt., prol. 5. On Armenian number symbolism in general, see R. W. Thomson, ‘Number Symbolism and Patristic Exegesis in Some Early Armenian Writers’, Handes Amsorya 90 (1976): pp. 117 38. ⁹⁷ As noted in the last chapter, Sedulius Scottus similarly compared the number of Canon Tables with the decalogue. ⁹⁸ Nerses, comm. Mt., prol. 17 (p. 12b). I have here departed from Russell’s translation. He translated the noun տպաւորութեամբ (in the instrumental case) as ‘with the impress’ but the root տպաւոր can mean ‘form, figure’ and its derivatives ‘symbolically, mystically, anagogically’, akin to the Greek word τύπος and its derivatives such as τυπικῶς. See Bedrossian, New Dictionary, s.v. տպաւոր.

Seeing the Salvation of God

259

into the mystical depths of the sacred text. Yet the metaphor extends one step further, for Nerses also calls on his readers to follow Eusebius by ascending the mountain carrying ‘the tablets of [their] hearts’, so that God might ‘by inspiration (ազդողութեամբ) inscribe [upon them] the power of this gospel which stands before us’.⁹⁹ Thus, as readers of the gospelbook follow in Eusebius’ footsteps, contemplating the mysterious depths of the ten beautiful images, the gospel revealed by God ‘in the figure [of the ten xorank‘]’ becomes the gospel written within the human person ‘by inspiration’.

Interpretations of Specific Motifs in the Decorative Scheme With this framework in place, Step‘anos and Nerses proceed through the illustrated pages explaining the symbolic significance of specific details in the decorative scheme. As they do so it becomes apparent that for each page there is an overarching theme that coordinates the various motifs contained therein. These overarching themes progress in a certain sequence, which is clearer in Nerses, who early in his exposition gives a succinct overview of what each of the pages in the series represents:¹⁰⁰ Xoran 1 Xoran 2 Xoran 3 Xoran 4 Xoran 5 Xoran 6 Xoran 7 Xoran 8 Xoran 9 Xoran 10

The Divine Being Middle Angelic Priesthood Last Angelic Priesthood Garden of Eden Noah’s Ark Abraham’s Sacrifice of Isaac The Outer Court of Moses’ Tabernacle The Holy of Holies in Moses’ Tabernacle Church ?

In this initial listing, Nerses seems to have made a mistake, because he makes no mention of what the tenth xoran represents.¹⁰¹ His later exposition basically follows this same outline, though the ninth xoran is there said to ⁹⁹ Nerses, comm. Mt., prol. 17 (p. 12b). I have again departed from Russell’s translation. He renders ազդոզութեամբ (a noun in the instrumental case) as ‘in efficacious (letters)’. However, the term is singular rather than plural. According to Bedrossian, ազդոզութիւն is equivalent to ազդեցութիւն, and the latter can mean ‘notice, warning, admonition; perception, instinct; inspiration; suggestion, instigation: efficacy, force, power; emphasis, energy’. I have gone with ‘inspiration’ because it seems to indicate best the contrast Nerses is making between writing in a codex and writing in the ‘tablets of the heart’. ¹⁰⁰ Nerses, comm. Mt., prol. 5 6. ¹⁰¹ Mathews and Sanjian, Armenian Gospel Iconography, p. 173, similarly listed Nerses’ identification of each of the ten but did not seem to notice the discrepancy in his account.

260

The Eusebian Canon Tables

represent Solomon’s temple (§14), while the tenth doubly serves to signify the church (§15) as well as the heavenly tabernacle that will appear at the eschaton (§16). Nerses’ inconsistency here is somewhat puzzling. Clearly his initial explanation of only nine xorank‘ is a mistake, and if we added in either Solomon’s temple or the eschatological tabernacle to the first list, we would have a complete set of ten explanations, one for each of the decorated pages. On the other hand, his later exposition would make the most sense as an explanation for a set of not ten but eleven pages, since the final image serves to represent both the church and the eschatological tabernacle. This inconsistency may be due to the fact that he was operating with competing traditions and was not entirely successful at harmonizing them, or could have resulted merely from some corruption in the manuscript tradition. Nevertheless, in both his initial listing and his fuller explanation of the images, it is clear enough that Nerses regards these xorank‘ as progressing through the important stages in the history of salvation, leading up to and including the present moment in the church, ‘which contains within itself the mystery of all [the others]’.¹⁰² In other words, according to Nerses, the xorank‘ with their embellishment are structured as an ordered progression, such that the reader turning the pages is led from one stage of redemptive history to the next, culminating in the final page, which recapitulates the entire preceding sequence. Step‘anos and Nerses creatively find ways to align the symbolic significance of seemingly all the motifs on each page with its overarching theme. So, for example, with respect to the first xoran, they both explain that the colour purple (presumably referring to the colour of the marble columns) represents the throne of God that exists outside of the four elements (see fig. 43).¹⁰³ Similarly, the black that they observe in the second xoran denotes the incomprehensibility of God (see fig. 44).¹⁰⁴ Both authors also observe a progression in the usage of colour in the xorank‘. Step‘anos first notes the colour red in the second xoran, and connects this with Melchizedek’s wine, a foreshadowing of Christ, while Nerses first mentions red in relation to the fifth xoran, pointing out that Noah’s ark foretold ‘the mystery of Christ’s blood’.¹⁰⁵ Though they first mention red in different xorank‘, they make the same point further on in their exposition, namely, that as one progresses through the painted pages, the colour red

¹⁰² Nerses, comm. Mt., prol. 6. ¹⁰³ Step‘anos, comm. xor. 2 (‘the existent establishment of the thrones of the outer elements’); Nerses, comm. Mt., prol. 7 (‘the fixedness of the existent throne of God outside the four elements’). ¹⁰⁴ Step‘anos, comm. xor. 3 (‘the black colour inside, which demonstrates true existence, too difficult for the seers to see’); Nerses, comm. Mt., prol. 8 (‘black [is symbolic of] the incompre hensibility of God, which is hidden’). ¹⁰⁵ Step‘anos, comm. xor. 3; Nerses, comm. Mt., prol. 10.

Seeing the Salvation of God

261

Fig. 43. First xoran in the Etchmiadzin Gospels (989). Erevan, Matenadaran, MS 2374, fol. 1r

comes increasingly to dominate the palette, signifying that one is drawing nearer to the death of Christ (see fig. 45).¹⁰⁶ Similarly, the various birds that perch on the xorank‘ have distinct meanings. For both authors, doves, not surprisingly, denote those who have received the Holy Spirit, with Nerses in xoran seven highlighting figures ¹⁰⁶ Step‘anos, comm. xor. 6 (‘the red becomes brighter . . . because the cross has come near’); Nerses, comm. Mt., prol. 14 (‘[in the ninth xoran] the black and blue (have) waned and the red has waxed brilliantly. For the advent of the Emmanuel has come close’); Nerses, comm. Mt., prol. 15 (‘And since the tenth xoran is most ornamented and resplendent and brilliant, with rosy red paint, blue is entirely exhausted, for the ancient darkness of sin and ignorance, of lamentation and sadness, is past, and all has become new, bedaubed with the blood of Christ’).

262

The Eusebian Canon Tables

Fig. 44. Fourth xoran in the Etchmiadzin Gospels (989). Erevan, Matenadaran, MS 2374, fol. 2v

such as Bezalel and Ohaliab, who built Israel’s tabernacle, along with Moses, Joshua, and Israel’s elders (see fig. 46).¹⁰⁷ In addition, our two authors agree that the partridges standing atop the xorank‘ (number five for Step‘anos and number eight for Nerses) symbolize the foreign women in Jesus’ ancestry, namely Tamar, Rahab, and Ruth (see fig. 47).¹⁰⁸ Nerses adds that the cock is particularly suited to the ninth xoran, since this is just before the ‘morning of righteousness’, whose dawn is represented by the tenth xoran (see fig. 48), while the herons ¹⁰⁷ Step‘anos, comm. xor. 5; Nerses, comm. Mt., prol. 12. ¹⁰⁸ Step‘anos, comm. xor. 6; Nerses, comm. Mt., prol. 13.

Seeing the Salvation of God

263

Fig. 45. Seventh xoran in the Etchmiadzin Gospels (989). Erevan, Matenadaran, MS 2374, fol.4r

adorning the tenth xoran are a fitting representation of the apostles who were fishermen (see fig. 49).¹⁰⁹ Nevertheless, their interpretations do on occasion diverge, nowhere more sharply than when Step‘anos says the peacocks in the first xoran refer to followers of the Mosaic Law who have ‘splendid bodily adornment’ but no relation to the church, while Nerses says that the ‘goldfeathered and gilt-tailed peacocks’ in the second and third xorank‘ symbolize the ‘pure, fine and unblemished nature’ of the angelic hosts (see fig. 50).¹¹⁰

¹⁰⁹ Nerses, comm. Mt., prol. 14; 15. ¹¹⁰ Step‘anos, comm. xor. 2; Nerses, comm. Mt., prol. 8.

264

The Eusebian Canon Tables

Fig. 46. A pair of doves in the Etchmiadzin Gospels (989). Erevan, Matenadaran, MS 2374, fol. 6v

Fig. 47. Partridges perched atop the fourth xoran in the Etchmiadzin Gospels (989). Erevan, Matenadaran, MS 2374, fol. 2v

Finally, Step‘anos and Nerses make much of the various plants that grow above and around the arcades. Both authors suggest that the pomegranate, appearing in xoran five (for Step‘anos) and nine (for Nerses), denotes the relation of the new dispensation to the old, since this fruit has a bitter peel concealing a sweet fruit within (see fig. 51).¹¹¹ On the other hand, the date palm in xoran four, according to Step‘anos, demonstrates that the truth has ‘sprouted up’ from the earth, while for Nerses

¹¹¹ Step‘anos, comm. xor. 6; Nerses, comm. Mt., prol. 14.

Seeing the Salvation of God

265

Fig. 48. Cocks on top of xoran eight in the Gladzor Gospels (c.1300). Gladzor Gospels, Library Special Collections, Charles E. Young Research Library, UCLA, Armenian MS 1, p. 17

Fig. 49. Herons perched atop the eighth xoran in the Etchmiadzin Gospels (989). Erevan, Matenadaran, MS 2374, fol. 4v

the tree’s height in xorank‘ two and three depicts the lofty grandeur of the angelic hosts and its fruit reveals ‘the sweetest pronouncement of blessing’ (see fig. 52).¹¹² This is only a representative sampling of the imagery discussed by these authors, but it suffices to demonstrate the way in which they regard the details ¹¹² Step‘anos, comm. xor. 5; Nerses, comm. Mt., prol. 8. In §14, in his consideration of the ninth xoran, Nerses repeats the same interpretation of the date palm given by Step‘anos (truth sprouting up from the earth), and also connects the tree with the Messiah who came from the root of David.

266

The Eusebian Canon Tables

Fig. 50. Peacocks above the first xoran in the Etchmiadzin Gospels (989). Erevan, Matenadaran, MS 2374, fol. 1r

Fig. 51. Pomegranates growing above the third xoran in the Etchmiadzin Gospels (989). Erevan, Matenadaran, MS 2374, fol. 2r

of the decorative scheme—as redolent with symbolic import, ‘mystically’ representing the most important stages of the Christian view of history in an ordered progression. Mathews and Sanjian reached a similar conclusion, noting that [i]n the Armenian approach to the canon tables, two kinds of themes predom inate: soteriological and ecclesiological. Though the pious viewer is given free rein to enlarge on the subjects according to his own reflection on Scripture, the commentators suggest that the thread that holds the decoration together is the unfolding of God’s plan of salvation in the church.¹¹³

¹¹³ Mathews and Sanjian, Armenian Gospel Iconography, p. 183.

Seeing the Salvation of God

267

Fig. 52. Date palm growing alongside xoran ten in the Gladzor Gospels (c.1300). Gladzor Gospels, Library Special Collections, Charles E. Young Research Library, UCLA, Armenian MS 1, p. 21

The only thing to be added to this statement is that the history recounted by the ten xorank‘, centred of course on the incarnation, extends to include the Canon Tables themselves, since they were contained within the gospelbook that played a central role in the performance of the liturgy in the church, which is signified by the tenth xoran. Moreover, this liturgical performance was of course carried out beneath the real architectural xoran that stood above

268

The Eusebian Canon Tables

the altar of the church upon which the gospelbook rested.¹¹⁴ It therefore seems plausible to suppose that the way Step‘anos and Nerses understood the built xoran within the church influenced their approach to the xorank‘ painted on the pages of the book beneath it. In fact, given that most people would have encountered the text of the gospel through readings in the liturgy, the entire liturgical performance functioned as a kind of paratext in terms of framing the participant’s experience of the text. It is, I suggest, this dense network of theological themes expressed in the liturgy that these authors are suggesting is also present in the prefatory artwork contained within the book itself, blurring the distinction between paratextual performance and paratextual images within the codex. The fact that Step‘anos and Nerses present, not a random assortment of symbolic references, but instead an orderly, logical progression through the pages, supports the proposal offered in modern scholarship that the overall decorative scheme is modelled on the experience of a person moving linearly through space, either through or alongside the archways, with the turn of each page bringing the reader or viewer to a new stage in the journey. It is this latter point that we need to explore further in the remainder of this chapter, to see more clearly how this paratextual process of meditative contemplation worked and for what purpose it was intended.

IMAGE AND SYMBOL: CANON TABLE DECORATION AS A S YSTEM OF LOCI AND IMAGINES Because no prior texts of this genre (i.e., commentaries on Canon Table artwork) have yet come to light, and because the contexts from which these two works emerged have received little scholarly attention, it is difficult to identify the sources informing the expositions of Step‘anos and Nerses, apart from the likelihood that Nerses was drawing upon Step‘anos. Hence, the following section is not intended as an argument that these two medieval authors were directly dependent upon certain classical texts from a millennium earlier, though, as I will explain more fully further on, I do not think it impossible that these sources are part of a single, broad tradition. Instead, my more limited aim in this section is to shed light on what Step‘anos and Nerses envision by adducing comparative material from another body of literature that presents a number of similarities to what we observe in our ¹¹⁴ On the liturgical use of books in the late antique period, see especially Lowden, ‘The Word Made Visible’, pp. 13 47, who argued that surviving book covers suggest that high quality codices were most likely propped open so that both exterior covers could be viewed simultan eously, as a kind of diptych.

Seeing the Salvation of God

269

two Armenian treatises. If the analogical relationship between these sources is compelling—and naturally the reader will have to decide if it is or not—then the comparative material might help us in uncovering some of the tacit assumptions in the two texts that are our main focus. The body of literature I have in mind is the classical rhetorical tradition, and the specific aspect of that tradition that is relevant here is the way in which rhetoricians discussed the purpose and operation of memory.¹¹⁵ Greek rhetoricians were the first to theorize memory, classing it as the fourth of the five parts of their craft,¹¹⁶ though that Greek tradition survives today in three Latin treatises from the first centuries  and  which drew upon now lost Greek sources:¹¹⁷ the work of unknown authorship known as Rhetorica ad Herennium, Cicero’s De oratore, and Quintilian’s Institutio oratoria. The goal of the advice offered by these three treatises was that an orator be able to memorize a speech so that it could be delivered fluidly, and to accomplish this end they all three recommend a mnemonic technique that relied on a combination of ‘places’ (loci) and ‘images’ (imagines/effigies). The loci are like background scenes into which one mentally places imagines that represent the ideas to be remembered. These authors allow for a variety of loci, though they have certain requirements for the sorts of backgrounds that are most effective. So, for example, the author of Rhetorica ad Herennium advises, ‘By places [locos] I mean such scenes as are naturally or artificially set off on a small scale, complete and conspicuous, so that we can grasp and embrace them easily by the natural memory—for example, a house, an

¹¹⁵ It is beyond my focus here, but it might also be fruitful to explore the descriptions of Step‘anos and Nerses as instances of rhetorical ekphrasis. The literature on this topic is immense. A helpful place to begin is Ruth Webb, Ekphrasis, Imagination and Persuasion in Ancient Rhetorical Theory and Practice (New York: Routledge, 2016), and Michael Squire, ‘Ecphrasis: Visual and Verbal Interactions in Ancient Greek and Latin Literature’, Oxford Handbooks Online, http://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780199935390.001.0001/oxfordhb 9780199935390 e 58 (accessed 18 January 2018), and on Byzantine ekphrasis in particular, see Liz James and Ruth Webb, ‘ “To Understand Ultimate Things and Enter Secret Places”: Ekphrasis and Art in Byzantium’, Art History 14 (1991): pp. 1 17; Ruth Webb, ‘The Aesthetics of Sacred Space: Narrative, Metaphor, and Motion in Ekphraseis of Church Buildings’, Dumbarton Oaks Papers 53 (1999): pp. 59 74; Maguire, ‘Art and Text’, pp. 721 4. ¹¹⁶ Kennedy, A New History, p. 6. For succinct summaries of this tradition highlighting a number of points relevant to the present discussion, see Michael Squire, ‘The Ordo of Rhetoric and the Rhetoric of Order’, in Art and Rhetoric in Roman Culture, ed. Jaś Elsner and Michel Meyer (Cambridge: Cambridge University Press, 2014), pp. 405 16; Jaś Elsner and Michael Squire, ‘Sight and Memory: The Visual Art of Roman Mnemonics’, in Sight and the Ancient Senses, ed. Michael J. Squire (London: Routledge, 2016), pp. 182 90. For a lengthier treatment see Small, Wax Tablets, pp. 81 137, and for a classic exposition, see Frances A. Yates, The Art of Memory, Pimlico ed. (London: The Bodley Head, 1992), pp. 17 41. ¹¹⁷ Elsner and Squire, ‘Sight and Memory’, p. 186: ‘There can be no doubting that these three Latin authors understood their methods of visual memory in relation to much earlier Greek paradigms. Revealingly, they also trace the origins back to the deep past of late Archaic Greece’.

270

The Eusebian Canon Tables

intercolumnar space, a recess, an arch, or the like’.¹¹⁸ Cicero similarly says that the loci ‘must be clear and defined and at moderate intervals apart’ (illustribus, explicatis, modicis intervallis).¹¹⁹ In his presentation of this concept, Quintilian focuses primarily on the example of a house with many rooms into which discrete items can be placed, but he also says that the same can be done with ‘public buildings, a long road, a town perambulation, or pictures’ (in operibus publicis et in itinere longo et urbium ambitu et picturis).¹²⁰ Based on the advice of these three authors, the architectural frames around the Eusebian tables are well suited to serve as memorial loci, since each is a discrete and wellarticulated space that can be grasped in a single viewing. Moreover, in view of the fact that the Canon Table pages are painted images of columns supporting an arch, it is especially striking that among the examples of suitable loci offered by the first author are an ‘intercolumnar space’ and an ‘arch’, while Quintilian mentions ‘pictures’. Next, the imagines are items that represent specific aspects of the matter one is attempting to remember, which are placed against the backdrop of the loci, with the result that a single locus should contain several imagines positioned in fixed locations. By scanning a single locus one can see at a glance a given set of items one has stored in the memory, and by progressing through a series of loci a number of such sets can be recalled. Our rhetoricians also have recommendations to offer regarding the most effective kind of imagines. The Rhetorica ad Herennium defines an imago as ‘a figure, mark, or portrait of the object we wish to remember’,¹²¹ and observes that not all imagines are equally effective. Rather, because ‘ordinary things easily slip from the memory while the striking and novel stay longer in the mind’, it is advisable to choose imagines that are ‘as striking as possible, . . . assigning to them exceptional beauty or singular ugliness’.¹²² Similarly Cicero asserts that imagines should be ‘effective and sharply outlined and distinctive, with the capacity of encountering and speedily penetrating the mind’,¹²³ a line that is quoted verbatim by Quintilian.¹²⁴ Viewed in the light of such advice, the various decorative elements of the Canon Tables, such as the colour scheme and the diverse flora and fauna, seem well suited to serve as imagines, since, although abstract, they are strikingly beautiful images and are positioned against the backdrop of the loci of the architectural frames.

¹¹⁸ rhet. Her. 3.16.29 (LCL 403, pp. 208 9). Locos appellamus eos qui breviter, perfecte, insignite aut natura aut manu sunt absoluti, ut eos facile naturali memori conprehendere et amplecti queamus: ut aedes, intercolumnium, angulum, fornicem, et alia quae his similia sunt. ¹¹⁹ Cicero, de or. 2.358 (LCL 348, pp. 470 1). ¹²⁰ Quintilian, inst. 11.2.21 (LCL 494, pp. 68 9). ¹²¹ rhet. Her. 3.16.29 (LCL 403, pp. 208 9). ¹²² rhet. Her. 3.22.36 7 (LCL 403, pp. 218 21). ¹²³ Cicero, de or. 2.358 (LCL 348, pp. 470 1). ¹²⁴ Quintilian, inst. 11.2.22 (LCL 494, pp. 68 9).

Seeing the Salvation of God

271

The more philosophically inclined Cicero provides an explanation for why this technique works, one that is reminiscent of the purpose Nerses gives for pleasurable things. According to the Roman orator, ‘the most complete pictures are formed in our minds of the things that have been conveyed to them and imprinted on them by the senses, but . . . the keenest of all our senses is the sense of sight’. As a result, words and abstract ideas ‘can be most easily retained in the mind if they are also conveyed to our minds by the mediation of the eyes’.¹²⁵ Quintilian makes a closely related point, observing that upon returning to a place after being away for a while, our minds are reminded of what we previously did there, who we were with, ‘and sometimes even unspoken thoughts’ that occurred to us previously.¹²⁶ Both these authors, therefore, provide evidence to support the claim that ‘within a Roman cultural imaginary . . . the very act of remembering was deemed inseparable from that of seeing’.¹²⁷ This, I suggest, is the same basic insight expressed by Nerses, though he of course presents it in a Christian garb—God has provided the pleasant things in the world, notably those things that most engage our senses, in order to lead us to transcendent, immaterial realities. Therefore sensible things like ‘colour, taste, smell, hearing’ and so on are imminently useful to the one properly trained in using these to achieve ‘the rational enjoyment (of) the good tidings of God’.¹²⁸ The teleology of Nerses’ point is certainly grander— delight of God—while Cicero is simply concerned here with remembering a speech, but the two authors agree that the human mind functions well when it joins abstract ideas with corporeal stimuli. There are two final ways in which the discussions of memory in these rhetorical treatises present analogies to what we have observed in the treatments of Step‘anos or Nerses. The first is in the emphasis on order that is intrinsic to the rhetorical use of memory. This is an idea that goes as far back as Aristotle, who noted that ‘things which have a certain order are easily recalled’.¹²⁹ Similarly, the author of Ad Herennium observes, ‘I likewise think it obligatory to have these loci in an order’,¹³⁰ and advises the wouldbe orator to place a distinctive mark in every fifth locus in order to preserve the sequence of the backgrounds.¹³¹ Cicero, too, points out that ‘the best aid to clearness of memory consists in orderly arrangement’, and that as a result

¹²⁵ Cicero, de or. 2.357 (LCL 348, pp. 470 1). ea maxime animis effingi nostris quae essent a sensu tradita atque impressa; acerrimum autem ex omnibus nostris sensibus esse sensum videndi; quare facillime animo teneri posse ea quae perciperentur auribus aut cogitatione si etiam commendatione oculorum animis traderentur. ¹²⁶ Quintilian, inst. 11.2.18 (LCL 494, pp. 66 7). ¹²⁷ Elsner and Squire, ‘Sight and Memory’, p. 181. ¹²⁸ Nerses, comm. Mt., prol. 3. ¹²⁹ Aristotle, de Mem. 452a. ἔστιν εὐμνημόνευτα ὅσα τάξιν τινὰ ἔχει. ¹³⁰ rhet. Her. 3.17.30 (LCL 403, pp. 208 9). Item putamus oportere ex ordine hos locos habere. ¹³¹ rhet. Her. 3.18.31 (LCL 403, pp. 210 11).

272

The Eusebian Canon Tables

‘the order of the loci will preserve the order of the facts [to be remembered]’.¹³² Quintilian advises that in order to memorize a longer speech it is necessary to divide it into sections with ‘well-defined stopping points’, so that the ‘sequence of the parts’ may be maintained ‘by the repeated recall of the order’.¹³³ The advantage of maintaining such an order, according to Ad Herennium, is that it forms an unbroken chain, allowing one to move backwards and forwards as needed in the sequence from whatever point one is presently recalling.¹³⁴ Here again we note a similarity with the treatment of the Canon Tables’ decoration by Step‘anos and Nerses. They too, particularly Nerses, maintain a distinct order in their explanation of the decorated pages, one that follows the unfolding of redemptive history. Another feature of the expositions of Step‘anos or Nerses that may be explained by the rhetorical tradition is the divergence in their identification of certain of the elements and the overall open-ended nature of their symbolic interpretation. In the light of the fact that no two sets of Canon Tables have identical decorative schemes, Mathews and Sanjian observed that ‘To the modern viewer this variability implies an arbitrariness that contravenes any serious symbolic intention’.¹³⁵ Similarly, Mathews and Sanjian drew attention to several differences in the expositions of the Canon Tables offered by these two authors, and the fact that even for a single author the same motif can have various symbolic interpretations at different points in the sequence. They attempted to explain such ‘flexibility’ as being due to the peculiarity of this distinctly Armenian ‘genre of literature’, one which is not ‘prescriptive’ like western painters’ manuals but instead merely instructs the reader in the right way to approach these images: ‘As instruments for contemplation, the Armenian canon tables were designed to be suggestive rather than explicit, with open-ended significance’.¹³⁶ Mathews and Sanjian were correct to highlight the flexibility of both Canon Table artwork and also the explanations of the decoration in these two treatises, but the expositions of Step‘anos or Nerses are in fact not as idiosyncratic as they claimed, for the rhetorical tradition as well emphasized that no single mnemonic scheme is suitable for everyone. The author of Ad Herennium criticized Greek manuals that he was aware of (now lost to us), which provided long lists of imagines for various words, since such aids remove the initiative from the industrious student who would otherwise have to come up with his own personal set, and, more importantly, because what is most memorable for one student will not necessarily be so for another. As a result, he counselled, ‘Everybody, therefore, should in equipping himself

¹³² ¹³³ ¹³⁴ ¹³⁵ ¹³⁶

Cicero, de or. 2.353 4 (LCL 348, pp. 466 7). Quintilian, inst. 11.2.28 (LCL 494, pp. 72 3). rhet. Her. 3.18.30 (LCL 403, pp. 210 11). Mathews and Sanjian, Armenian Gospel Iconography, p. 169. Mathews and Sanjian, Armenian Gospel Iconography, pp. 171 2.

Seeing the Salvation of God

273

with images suit his own convenience’, with the role of the teacher being simply to offer a few representative examples so that the student can learn the right approach to the technique.¹³⁷ The same principle might also explain why painters felt free to experiment with a variety of decorative schemes when creating the Canon Table pages, and why Step‘anos and Nerses are not dogmatic in their interpretations, but seem willing to countenance a high degree of variability. Of course, these three Latin sources are seemingly talking about purely mental images rather than actual paintings or physical structures. However, classicists have argued that this rhetorical recognition of the mnemonic power of visual stimuli was materialized in a number of ways in the Roman world.¹³⁸ So, for example, Agnès Rouveret has suggested that the first-century  Tabulae Iliacae, miniature marble tablets presenting images of Homeric scenes in addition to complex word puzzles, were intended as an elaborate mnemonic tool, associating specific episodes with topographical features, keyed to the letters of the alphabet.¹³⁹ In addition, Bettina Bergmann has suggested that the House of the Tragic Poet at Pompeii was a sort of ‘Roman memory theatre’ in which ‘sustained contemplation of the arrangements [of the wall paintings] exercised the educated viewer’s memory by unlocking a variety of associations and inviting a sequence of reasoned conclusions’.¹⁴⁰ More recently Jaś Elsner and Michael Squire have extended this line of analysis of Pompeian wall paintings by arguing that the various domestic spaces of a house could be regarded as individual loci containing ‘interrelated schematic imagines’, and that by progressing through the rooms in a certain ordo a viewer could use the entire house as a grand locus containing a variety of mnemonic associations that could be explored.¹⁴¹ In other words, the arrangement of the paintings was designed to ‘provide the viewer with a visual matrix, and one that had the potential to structure the viewer’s social experience of the house’.¹⁴² Hence, the principles of the mnemonic system of loci and imagines did not remain solely with the elite world of practising rhetors but were also used as a theoretical basis for forming objects, images, and physical spaces

¹³⁷ rhet. Her. 3.23.38 9 (LCL 403, pp. 220 3). ¹³⁸ Cf. Squire, ‘The Ordo of Rhetoric’, p. 412. ¹³⁹ Agnès Rouveret, ‘Les Tables iliaques et l’art de la mémoire’, Bulletin de la Societé Nationale des Antiquaires de France (1988): pp. 166 76. On Rouveret’s thesis see Michael Squire, The Iliad in a Nutshell: Visualizing Epic on the Tabulae Iliacae (Oxford: Oxford University Press, 2011), pp. 71 2. On these tablets, see also Mary Carruthers, The Craft of Thought: Meditation, Rhetoric, and the Making of Images, 400 1200, Cambridge Studies in Medieval Literature 34 (Cambridge: Cambridge University Press, 1998), pp. 201 3. ¹⁴⁰ Bettina Bergmann, ‘The Roman House as Memory Theater: The House of the Tragic Poet in Pompeii’, The Art Bulletin 76 (1994), p. 255. ¹⁴¹ Elsner and Squire, ‘Sight and Memory’, pp. 190 201. ¹⁴² Elsner and Squire, ‘Sight and Memory’, p. 198.

274

The Eusebian Canon Tables

in the ancient world.¹⁴³ Moreover, the number of examples of the materialization of mnemotechnical principles could be greatly multiplied if we considered the Latin Middle Ages, whether in architecture, painting, or book illustration.¹⁴⁴ This materialization can be understood as but a natural extension of the insights into human cognition expressed in the three Latin rhetorical treatises. If the memory works best when using mental images of corporeal objects to stand in for abstract concepts, why not use actual corporeal objects to this end?¹⁴⁵ The principles of human cognition that made the system of loci and imagines effective were not confined to Latin rhetoricians of the imperial period. Rather, similar insights can on occasion be found among Byzantine authors, who stood much closer to our Armenian commentators. The locus classicus in this respect is the famous homily of Patriarch Photius, delivered at the unveiling of the Theotokos icon in Hagia Sophia: Has a man lent his ear to a story? Has his intelligence visualized and drawn to itself what he has heard? Then, after judging it with sober attention, he deposits it in his memory. No less indeed much greater is the power of sight. For surely, [sight], through the outpouring [i.e., from the eye] and emanation [i.e., from the object] of the optical rays, as if somehow contacting and regarding the form of the visible object, sends it to the intellect, letting it be conveyed from there to the memory for the concentration of unfailing knowledge. Has the mind seen? Has it apprehended? Has it imagined? Then, it has effortlessly transmitted the images to the memory.¹⁴⁶

¹⁴³ For further applications of the art of memory in the ancient world, see Small, Wax Tablets, pp. 224 39. ¹⁴⁴ See Carruthers, The Craft of Thought; Carruthers, The Book of Memory. ¹⁴⁵ Cf. Yates, The Art of Memory, pp. 91 2: ‘For when people were being taught to practise the formation of images for remembering, it is difficult to suppose that such inner images might not sometimes have found their way into outer expression. Or, conversely, when the “things” which they were to remember through inner images were of the same kind as the “things” which Christian didactic art taught through images, that the places and images of that art might themselves have been reflected in memory, and so have become “artificial memory” ’. ¹⁴⁶ Photius, hom. 17.5 (Laourdas, p. 171). Ἔκλινέ τις τὸ οὖς εἰς διήγημα; εἵλκυσε ϕανταζομένη τὸ ἀκουσθὲν ἡ διάνοια; νηϕούσῃ μελέτῃ τὸ κριθὲν τῇ μνήμῃ ἐναπέθετο. Οὐδὲν τούτων ἔλαττον, εἰ μὴ καὶ πολὺ μᾶλλον, κρατεῖ τὰ τῆς ὄψεως· καὶ γὰρ καὶ αὐτή γε δήπου τῇ προχύσει καὶ ἀπορροῇ τῶν ὀπτικῶν ἀκτίνων τὸ ὁρατὸν οἱονεί πως ἐπαϕωμένη καὶ περιέπουσα τὸ εἶδος τοῦ ὁραθέντος τῷ ἡγεμονικῷ παραπέμπεται, ἐκεῖθεν διαπορθμευθῆναι διδοῦσα τῇ μνήμῃ πρὸς ἐπιστήμης ἀπλανεστάτης συνάθροισιν. Εἶδεν ὁ νοῦς, ἀντελάβετο, ἐϕαντάσθη, τοὺς τύπους ἀκόπως ἐν τῇ μνήμῃ παρεπέμψατο. The portion of the above passage about hearing follows the translation of Mango, The Homilies of Photius, p. 294, and the portion about sight follows the translation of Roland Betancourt, ‘Why Sight is Not Touch: Reconsidering the Tactility of Vision in Byzantium’, Dumbarton Oaks Papers 70 (2016): pp. 1 23, at 12 21, who made important modifications to Mango’s translation. On this passage, cf. Leslie Brubaker, ‘Pictures Are Good to Think With: Looking At Byzantium’, in L’écriture de la mémoire: la littérarité de l’historiographie, ed. Paolo Odorico, Panagiotis A. Agapitos, and Martin Hinterberger, Dossiers byzantins 6 (Paris: Centre d’études byzantines, néo helléniques et sud est européennes, École des hautes études en sciences sociales, 2006), p. 226; Amy Papalexandrou, ‘The Memory Culture of Byzantium’, in A Companion to Byzantium, ed. Liz James (Malden, MA: Wiley Blackwell, 2010), pp. 112 13.

Seeing the Salvation of God

275

Similarly, the earliest surviving theological defence of images in any language, an Armenian treatise written by Vrt‘anes K‘ert‘oł in the early seventh century, asserts that ‘We know the invisible by that which is visible, and the inks and paintings are the memory of God and his servants’ (յայտնեօք զաներեւոյթսն ճանաչեմք, եւ դեղք եւ նկարք յիշողութիւն է Աստուծոյ [եւ] ծառայից նորա).¹⁴⁷ Thus, for both Photius and Vrt‘anes, images have a unique effectiveness when it comes to imprinting something upon the memory, and in this respect they are in agreement with Cicero and the rhetorical tradition that he represents. Of course they had not read Cicero, but were drawing upon earlier Christian authors such as Basil of Caesarea and John of Damascus, who had also stressed the primacy of sight, perhaps owing to the influence of the same broad rhetorical tradition.¹⁴⁸ Step‘anos and Nerses were probably aware of traditions similar to those found in Vrt‘anes and later Photius, which emphasized viewing sacred images as means of perceiving the immaterial,¹⁴⁹ and, although neither of them specifically mentions ‘memory’ in their treatments of Canon Table decoration, the comments from Vrt‘anes and Photius demonstrate that the connection between vision and memory was well known in the later centuries of Byzantium and Armenia, indeed that it was central to the widely held understanding of the function of religious images.¹⁵⁰ As such it was probably assumed, even if unstated, by these two Armenian authors.

¹⁴⁷ Vrt‘anes K‘ert‘oł, yał. (text in Tourian, p. 63, ln. 429 31; French translation in Der Nersessian, p. 69). ¹⁴⁸ On the primacy of sight in Byzantium, see Leslie Brubaker, Vision and Meaning in Ninth Century Byzantium: Image as Exegesis in the Homilies of Gregory of Nazianzus, Cambridge Studies in Palaeography and Codicology 6 (Cambridge: Cambridge University Press, 1999), pp. 48 9; Liz James, ‘Art and Lies: Text, Image and Imagination in the Medieval World’, in Icon and Word: The Power of Images in Byzantium, ed. Antony Eastmond and Liz James (Burlington, VT: Ashgate, 2003), pp. 59 72; Liz James, ‘Senses and Sensibility in Byzantium’, Art History 27 (2004): p. 529; Brubaker, ‘Pictures Are Good to Think With’, pp. 222 4. On theories of sight in Byzantium, see most recently Betancourt, ‘Why Sight is Not Touch’, pp. 1 23. This point is severely underestimated in Mathews and Sanjian, Armenian Gospel Iconography, p. 172, who implied that the Armenian tradition is the only Christian tradition that does not ‘denigrate the pleasurable or the sensuous’. ¹⁴⁹ The central argument of James and Webb, ‘To Understand Ultimate Things’, p. 12, is that this was also the primary purpose of ekphrasis in Byzantium: ‘Ekphrasis thus made present not the actual picture, which could be seen, but the spiritual reality behind it’. ¹⁵⁰ A treatment of memory in Byzantium akin to Yates, The Art of Memory or Carruthers, The Book of Memory has yet to be written, but see the following studies: James, ‘Art and Lies’; Amy Papalexandrou, ‘Memory Tattered and Torn: Spolia in the Heartland of Byzantine Hellenism’, in Archaeologies of Memory, ed. Ruth M. Van Dyke and Susan E. Alcock (Malden, MA: Blackwell, 2003), pp. 56 80; Papalexandrou, ‘The Memory Culture of Byzantium’, pp. 108 22; Michael Grünbart, ‘Zur Memorialkultur im byzantinischen Mittelalter’, in Byzantine Religious Culture: Studies in Honor of Alice Mary Talbot, ed. Denis Sullivan, Elizabeth Fisher, and Stratis Papaioannou (Leiden: Brill, 2012), pp. 373 94; Charles Barber, ‘Writing on the Body: Memory, Desire, and the Holy in Iconoclasm’, in Desire and Denial in Byzantium, ed. Liz James (New York: Routledge, 2016), pp. 111 20.

276

The Eusebian Canon Tables

However, what sets Step‘anos and Nerses apart from the broader Byzantine and Armenian reflection upon images is that they have developed an elaborate scheme of symbolic representation, using not just one image but a series of pictures to convey a complex, multifaceted theological message. Moreover, given that the xorank‘ remain the dominant element in all of the ten pages, it is striking that their interpretations rest upon the variation of minor details such as colours and flora and fauna, a feature that is paralleled in classical sources influenced by mnemotechnical principles.¹⁵¹ Furthermore, no other text, so far as I am aware, in Byzantine or Armenian literature relies so extensively on abstract symbolic representation. That is to say, Step‘anos and Nerses highlight not merely motifs with obvious Christian resonance, such as crosses or xorank‘, but also various birds (cocks, doves, partridges, herons, ducks, and peacocks), and plants (date palms, olive trees, lilies, and pomegranates), only some of which would otherwise have a commonly known theological significance.¹⁵² These idiosyncratic aspects of their interpretations may, I suggest, be explained as deriving from, or at least analogous to, the rhetorical programme of loci and imagines. Mary Carruthers has argued persuasively that ‘the history of medieval rhetoric’ cannot be written ‘simply by examining the transmission lines of ancient texts’.¹⁵³ She proposed that to uncover the influence of classical rhetoric on the Latin Middle Ages it is necessary to consider also the orthopraxis developed in monasticism ‘for the invention of meditation and the composition of prayers’, which represented an adaptation

¹⁵¹ Cf. Elsner and Squire, ‘Sight and Memory’, p. 193: ‘From a Roman rhetorical perspective, then, each visual detail within the picture could subsequently serve to open up narrative vistas to provide the viewer with prompts to turn the image into verbal interpretation or response. For the viewer accustomed to using paintings as mnemonic tools, relatively minor differences in attributes could serve to connote major differences of subject matter’. The examples they adduced are Pompeian paintings in which a ‘single . . . bodily schema’ is used repeatedly, but takes on entirely different associations through the presence of specific details in each image (p. 192). ¹⁵² The development of this approach to Canon Table artwork in Armenia may be related to the thesis of John Onians, ‘Abstraction and Imagination in Late Antiquity’, Art History 3 (1980): pp. 1 24, who argued that during late antiquity and the early medieval period artwork shifted away from naturalism or realism towards more abstract representation, coinciding with a heightening of the imaginative, or mimetic, capacity of the viewer to see more in the artwork. He proposes that late antique rhetoric, specifically ekphrasis, played a central role in this gradual shift over the centuries, which resulted in art that ‘concentrates . . . on the making of images which are capable of both attracting and holding the attention of the spectator and of commu nicating mysteries of meaning beyond the physical and material level’ (p. 1). This argument is particularly well suited to an example like the Gladzor Gospels, in which the architectural frames surrounding the tables and the decoration within the arcades, though certainly aesthetically pleasing, are very abstract and complex, hardly resembling any structure in the real world, which represents a shift away from those in earlier Canon Tables that are more realistic. ¹⁵³ Mary Carruthers, ‘Late Antique Rhetoric, Early Monasticism, and the Revival of School Rhetoric’, in Latin Grammar and Rhetoric: From Classical Theory to Medieval Practice, ed. Carol Lanham (New York: Continuum, 2002), p. 252.

Seeing the Salvation of God

277

of classical rhetorical techniques, though rarely in the form of technical treatises.¹⁵⁴ It is plausible that, even if our two authors had not read a rhetorical handbook describing the use of loci and imagines, a similar mode of transmission might have mediated these insights to them.¹⁵⁵ Alternatively, this peculiar approach to the decoration might have arisen independently, as a result of extended reflection on the cognitive principles by which the human mind works, which would be as applicable for Latin rhetors of the first century  as it would for Armenian monks seven centuries later: Step‘anos and Nerses recognized that sensuous objects can serve as hooks upon which to hang abstract concepts, and that the more delightful the objects the greater efficacy they possesses; they, like our three Latin authors, were aware that a series of backgrounds that could be taken in at a single glance, containing several discrete items in each, was the best visual and so mental organization of the sensuous objects standing in for incorporeal realities; they understood that depictions of architectural structures could function extremely well as these backgrounds and that the transition from one to another could be envisioned as a movement through space; finally, they realized that these backgrounds functioned best when put in some kind of logical order. If this analogy holds, it suggests that, although Step‘anos and Nerses do not mention memory explicitly in their expositions, they intended for their symbolic interpretations of Canon Table artwork to serve as a mnemotechnical program.¹⁵⁶ In fact, it may be that Nerses alludes to precisely this point at the conclusion of his treatise. I pointed out earlier that he ends his exposition with the metaphor of the reader, standing at the base of the mountain of God, about to ascend so that the gospel typologically depicted in the xorank‘ might be inscribed upon the tablet of his heart.¹⁵⁷ The inscribing of something upon one’s inner self (whether mind or heart) was not merely a pious image drawn ¹⁵⁴ Carruthers, The Craft of Thought, pp. 1 3. ¹⁵⁵ Though it should be noted that Greek rhetorical treatises continued to be passed down in Byzantium and at least some were translated into Armenian. Mathews and Sanjian, Armenian Gospel Iconography, pp. 200 1, noted the existence of three thirteenth and fourteenth century manuscripts executed at the monastery of Gladzor, which contain an Armenian version of Aphthonius’ Progymnasmata, attributed to Movses Xorenac‘i (Erevan, Mat. 1681 and Erevan, Mat. 6897, which are nos. 21 and 32 in the list of Mathews and Sanjian; the location of no. 57 in their list is now unknown). Aphthonius’ treatise became ‘the standard [rhetorical] handbook in the Byzantine period’ (Mary Whitby, ‘Rhetorical Questions’, in A Companion to Byzantium, ed. Liz James (Malden, MA: Wiley Blackwell, 2010), p. 242), so it is not surprising to find it also in use in medieval Armenia. Cf. Robert W. Thomson, ‘The Formation of the Armenian Literary Tradition’, in East of Byzantium: Syria and Armenia in the Formative Period, ed. Nina G. Garsoïan, Thomas F. Mathews, and Robert W. Thomson (Washington, DC: Dumbarton Oaks Center for Byzantine Studies, 1982), pp. 145 6. ¹⁵⁶ Heidi C. Gearhart has similarly proposed that the variation in decoration in the twelfth century Gospels of Helmarshausen (J. Paul Getty Museum, Los Angeles, MS Ludwig II 3) might be for the purpose of ‘help[ing] viewers remember which stories fell into which canon’ (Gearhart, ‘From Divine Word to Human Hand’, pp. 442 3). ¹⁵⁷ Nerses, comm. Mt., prol. 17.

278

The Eusebian Canon Tables

from the New Testament (cf. 2 Cor 3:2–3; Heb 10:16) but also a metaphor widely used to explain the function of memory in human cognition from Plato well into the Middle Ages—perhaps even the most widely used image for this purpose.¹⁵⁸ To take but one relevant example, at the beginning of the famous passage in which he proclaims the superiority of apprehension by sight over apprehension by words, Photius explains, ‘Just as discourse comes through hearing, so through sight a form is engraved upon the tablets of the soul, sketching knowledge concordant with piety for those who have not had the preconception of wicked doctrines imprinted upon them’.¹⁵⁹ It is this ‘engraving upon the tablets of the soul’ that Photius a few sentences later refers to as ‘depositing [what has been seen] in the memory’. The use of this metaphor in a similar discussion of visuality in Nerses’ exposition suggests that here, at the conclusion of his treatise, he is revealing the purpose of the preceding symbolic interpretation of Canon Table artwork, namely, that the reader should memorize the decorative scheme of the ten pages in order to be able to recall the transcendental realities they symbolically depict.

IMAGE AND TEXT: ARMENIAN CANON TABLES AS BILDEINSÄTZE This argument can be pushed one step further, drawing in more comparanda that offer a suggestion as to what the purpose of this mnemotechnical exercise might have been for Step‘anos and Nerses. The argument from the last section could have been referring to a set of decorated Canon Tables that existed as a detached set of five folios, standing alone as a symbolic representation of salvation history. Does it make any difference that this set of ten illustrated pages was actually the preface to a corpus of texts? I want to argue that it does, and that, in fact, the images relate to the text in a specific way.¹⁶⁰ Again here I follow the ground-breaking work on the Latin Middle Ages of Mary Carruthers, who has drawn attention to the way in which medieval treatises in the West often begin with a description of an image or painting, a tendency that has classical antecedents. Representative examples include Cicero’s De natura ¹⁵⁸ Cf. Small, Wax Tablets, pp. 131 6; Carruthers, The Book of Memory, pp. 18 37. ¹⁵⁹ Photius, hom. 17.5 (Laourdas, p. 170). ὥσπερ γὰρ ὁ λόγος ὁ δι’ ἀκοῆς, οὕτω δι’ ὄψεως ἡ μορϕὴ τοῖς τῆς ψυχῆς ἐγχαράσσεται πίναξιν, ὁμόϕωνον τῆς εὐσεβείας, οἷς πονηρῶν δογμάτων ἡ πρόληψις οὐκ ἐνέστικται, τὴν μάθησιν διαγράϕουσα. Cf. Betancourt, ‘Why Sight is Not Touch’, pp. 17, 20. ¹⁶⁰ On the relation of image to text in antiquity, see the stimulating study of Michael Squire, Image and Text in Graeco Roman Antiquity (Cambridge: Cambridge University Press, 2009). For Byzantium, see the collected studies in Liz James, ed., Art and Text in Byzantine Culture (Cambridge: Cambridge University Press, 2007).

Seeing the Salvation of God

279

deorum, several of Lucian’s works, such as Slander and On Salaried Posts, ps-Cebes’ Pinax, Martianus Capella’s treatise on the liberal arts, and Boethius’ Consolation of Philosophy. In the surviving rhetorical treatises from antiquity no name is given to this technique, so it is usually referred to using the modern term Bildeinsatz, coined in the early twentieth century by O. Schissel.¹⁶¹ These vivid descriptions of images are clearly related to the technique of ekphrasis that was explicitly theorized by rhetoricians and has been much studied in recent decades. However, according to Carruthers, Bildeinsätze differ from ekphrasis in two respects: [A Bildeinsatz] is at the beginning of a work (or of a major division or change of subject within a long work), a trope of the introduction whereas ekphrasis can occur at any point. And because of its location, it acts as the elementary founda tion, the dispositio of what follows. Introductory rhetorical pictures serve as orienting pages and summaries of the matters which are developed within the work. . . . [They are located at the beginning of a work so that] a reader can hold the picture in mind as a way of recognizing the major themes of what follows . . . ¹⁶²

Two later medieval examples cited by Carruthers show the perpetuation of this technique into the Middle Ages, during which time they came to be known as picturae.¹⁶³ In the Anglo-Saxon poem Beowulf, King Hrothgar prepares to give a speech in praise of Beowulf by first gazing upon the images of the flood on his sword-hilt, which Carruthers interpreted to mean that the pictures on the sword were ‘the inventional, ordering instrument with which he composes’. The second example comes from Peter of Celle, an almost exact contemporary of Nerses. In his treatise titled ‘On Conscience’, composed around 1170, Peter prefaced the main section with an extended description of a banquet scene, rich in sensuous detail, which provides, in symbolic form, a summary of the subsequent text. Other portions of his treatise also use the method, with paragraph-size sections being introduced by smaller pictures that serve the same purpose. Writing this rhetorical technique into the narrative itself, at one point Peter describes the figure of Charity painting a picture ¹⁶¹ O. Schissel, ‘Die Technik des Bildeinsatzes’, Philologus 72 (1913): pp. 83 114. See further Eva Keuls, ‘Rhetoric and Visual Aids in Greece and Rome’, in Communication Arts in the Ancient World, ed. E. Havelock and J. P. Herschbell (New York: Hastings House, 1978), pp. 121 34, who examined the literary sources for Schissel’s Bildeinstätze and concluded, ‘the evocation of a visual image, usually a painting, as a pretext to introduce a discourse or narrative, had a continuous history in the rhetorical tradition and . . . profoundly influenced the creative literature of the Greco Roman age’ (pp. 128 9). The phenomenon was also acknowledged in passing in Jaś Elsner, ‘Introduction’, in Art and Rhetoric in Roman Culture, ed. Jaś Elsner and Michel Meyer (Cambridge: Cambridge University Press, 2014), pp. 27 8. ¹⁶² Carruthers, The Craft of Thought, p. 199. ¹⁶³ Carruthers, The Craft of Thought, pp. 196 220.

280

The Eusebian Canon Tables

of herself, in which important Christian themes are symbolized by her various body parts (e.g., head for the Trinity, eyes for the Passion, ears for the gospels, etc.) along with the colours used (e.g., scarlet for blood, black for death, green for resurrection, etc.).¹⁶⁴ Carruthers argued persuasively that Peter intended these picturae to function as a mnemonic system that provided an orienting summary for the various sections of his treatise. Of course, the examples cited in the last paragraph from antiquity into the Middle Ages are seemingly descriptions of imagined paintings and images rather than representations in visual media. In fact, most often Bildeinsätze probably existed solely in the minds of their authors and readers. Finding introductory images in books that functioned in this manner is surprisingly difficult with respect to antiquity and late antiquity.¹⁶⁵ To some degree this is no doubt due to the chance survival of material evidence, but it is striking that evidence of images in bookrolls in general is extremely thin.¹⁶⁶ However, the same cannot be said with respect to the Middle Ages, in which illustrated codices were common, and Carruthers has convincingly argued that such images in the West often functioned as mnemotechnical devices.¹⁶⁷ ¹⁶⁴ Carruthers, The Craft of Thought, pp. 205 9. ¹⁶⁵ Keuls, ‘Rhetoric and Visual Aids’, 129: ‘That any such standard pictorial aids were used in rhetorical practice, either for the perpetuation of the technique of “Bildeinsatz” or for any other descriptive composition, is not indicated by the evidence presently available’. ¹⁶⁶ The treatment of this issue that dominated much of twentieth century scholarship was that of Kurt Weitzmann, Ancient Book Illumination, Martin Classical Lectures 16 (Cambridge, MA: Harvard University Press, 1959); Kurt Weitzmann, Illustrations in Roll and Codex: A Study of the Origin and Method of Text Illustration, Studies in Manuscript Illumination 2 (Princeton, NJ: Princeton University Press, 1970), who argued that although there are few surviving illustrated rolls, the existence and significance of those that once existed could be inferred from their influence on a wide variety of other media. For a critique of the philological method Weitzmann adapted for the study of images, see Leslie Brubaker, ‘Critical Approaches to Art History’, in Oxford Handbook of Byzantine Studies, ed. Robin Cormack, John F. Haldon, and Elizabeth Jeffreys (Oxford: Oxford University Press, 2008), pp. 61 2, and for a critique of Weitzmann’s understanding of images as mere ‘illustrations’ of an accompanying text, see Squire, Image and Text, pp. 122 31. Weitzmann’s view is also critiqued in Netz, ‘Authorial Presence’, pp. 232 41, in the light of broader cultural assumptions about what texts and books are thought to be. For a recent survey of the material and literary evidence for illustrated texts from antiquity, see Jocelyn Penny Small, The Parallel Worlds of Classical Art and Text (Cambridge: Cambridge University Press, 2003), pp. 118 54, who concluded, ‘[i]llustrated rolls with pictures integrated with text begin in the second half of the first century B.C., are rare in the first century A.D., and increase slowly in popularity thereafter’ (pp. 140 1). For an evaluation of Weitzmann’s thesis as it pertains to the Tabulae Iliacae, see Squire, The Iliad in a Nutshell, pp. 129 39, who pointed out that Weitzmann’s argument is one from silence, and moreover that ‘scholarship on surviving codices has come to realize that the placing of images beside narrative texts in manuscripts was a specific invention of the (very) late antique world’ (p. 139). For a wide ranging analysis of graphic images as an information technology in the Roman world, see Riggsby, Mosaics of Knowledge, chapter 5. ¹⁶⁷ Cf. Carruthers, The Book of Memory, pp. 274 337. Vandendorpe therefore was right to emphasize the rise of the visual in medieval books, but wrong to assume that images were mainly ‘to make reading attractive to a population that was largely illiterate’ (Vandendorpe, From Papyrus to Hypertext, pp. 97 98).

Seeing the Salvation of God

281

Whatever the sources may have been for the uniquely Armenian mode of commenting on Canon Table decoration, the similarities between it and the tradition of Bildeinsätze are striking. The Armenian explanations of Canon Table artwork function along the same lines as the pictura in Peter of Celle’s treatise, with the exception that they are descriptions of actual painted images, and so would seem to be an instance of Bildeinsätze that existed not just in the mental realm, but also in material form. Despite this difference, in the descriptions of Step‘anos and Nerses, Canon Table decoration related to the subsequent text of the gospels in the same way that Peter of Celle’s pictura related to the remainder of his treatise. That is to say, the symbolic explanations of these two Armenian authors provide a summary or outline intended to orient the reader to the major themes of the text that follows. This might at first seem counter-intuitive, in so far as none of the gospels presents in linear fashion the sequence of stages of redemptive history that Step‘anos and Nerses see in the ten xorank‘. This, however, is precisely what makes such a Bildeinsatz useful. The expositions of Step‘anos and Nerses present in a logical sequence a summary of the important concepts that occur in a more haphazard narrative sequence in the gospels that follow. It is a thus a process undertaken to order the textual content of the fourfold gospel, the same paratextual effect produced by the cross-referencing numeric matrices devised by Eusebius. Carruthers’ work on memory in the Latin Middle Ages can offer us one final insight relevant to the treatises of Step‘anos and Nerses. When the author of Rhetorica ad Herennium called memoria the ‘treasury of inventions’ (thesaurum inventorum),¹⁶⁸ he highlighted the creative goal towards which one was expected to use memory, and the emphasizing of this fact is one of the most significant contributions of Carruthers’ work. She has argued at length that both Roman rhetoricians as well as later monastic practitioners of the art conceive of memory not only as ‘rote’, the ability to reproduce something (whether a text, a formula, a list of items, an incident) but as the matrix of a reminiscing cogitation, shuffling and collating ‘things’ stored in a random access memory scheme, or set of schemes a memory architecture and a library built up during one’s lifetime with the express intention that it be used inventively. Medieval memoria was a universal thinking machine, machina memorialis . . .¹⁶⁹

Because memory was always assumed to have this creative telos, ‘the proof of a good memory lies not in the simple retention even of large amounts of material; rather, it is the ability to move it about instantly, directly, and securely that is admired’.¹⁷⁰ For a practising orator, this was essential because

¹⁶⁸ rhet. Her. 3.16.28 (LCL 403, pp. 204 5). ¹⁶⁹ Carruthers, The Craft of Thought, p. 4. See also Carruthers’ distinction between mneme and mimesis in Carruthers, The Craft of Thought, p. 3; Carruthers, ‘Late Antique Rhetoric’, p. 247. ¹⁷⁰ Carruthers, The Book of Memory, pp. 21 2.

282

The Eusebian Canon Tables

one had to be prepared to respond extemporaneously to the exigencies of a debate. This might require reshuffling the original ordo one had intended, and the ability to make that judgement in the moment required a well-trained memory, able to move nimbly between the various aspects of a topic or theme by following the links forged in one’s memory bank.¹⁷¹ In other words, the purpose of mnemotechniques in antiquity and the Middle Ages was to furnish the mind with the raw materials necessary for creative composition. This understanding of the purpose of mnemotechniques is particularly fitting for the treatises of Step‘anos and Nerses. What they offer to their reader is a method of using these decorated pages as a scheme for recalling the outline of salvation history, so that the mind of the viewer would be well furnished with the raw materials necessary for fruitfully engaging with the text of the fourfold gospel. This kind of paratextual meditation upon the xorank‘ would be a benefit to any reader of the gospels, but would be especially relevant for those who were preparing homilies or commentaries on the biblical text, activities that we know these two authors undertook. Nerses himself signals this fact, when he begins his discussion of the Canon Tables by providing a synopsis of salvation history, and then concludes, ‘This is the history of the salvation of God’s glory, which this book of good tidings encompasses, which is before us for interpretation’.¹⁷² Following this statement he goes on to explain the symbolism of the ten xorank‘ as the stages of redemptive history which are ‘encompassed’ by the fourfold gospel. In other words, the story told by the ten xorank‘ is identical to the story told by the four gospels. Yet because a given passage of the gospels might pertain to the history of salvation at any of the stages represented by the ten xorank‘, the reader or interpreter who had undergone the preparatory meditative exercise would be equipped to move backwards and forwards as needed from that point to provide the exposition of the passage required for the exegete’s specific context. My reading of Step‘anos and Nerses is in no way intended to detract from the mystical depths of meaning that they find in the artwork of these sumptuous pages, as though the deeply religious sentiment they express is mere window dressing for what was in fact nothing more than an exercise designed ¹⁷¹ Cf. Elsner and Squire, ‘Sight and Memory’, pp. 202 3: ‘Yet what matters in each case is less the visual prompts themselves than what the orator imaginatively does with them: the sights summoned up in the memory are designed to provide rhetorical speech with an ordo that can then be retraced at personal whim’. See also Squire, ‘The Ordo of Rhetoric’, p. 413: ‘Objects like the Tabulae Iliacae facilitate a process very much akin to the one advocated by Quintilian. If, like Agnès Rouveret, we understand the composition of the Tabula Capitolina . . . as a sort of material counterpart to the imaginary mnemonic spaces recommended by Roman rhetoricians, its system of imagines and loci at once preserves a narrative structure and allows the speaker to depart from it and in whichever way he chooses. Far from simply preserving “the” order of Homer, the tablet operates within a rhetorical culture where orators were trained in using visual prompts to reorder their speech according to the needs of any given scenario’. ¹⁷² Nerses, comm. Mt., prol. 2 (pp. 5b 6a).

Seeing the Salvation of God

283

to produce rote recall.¹⁷³ Rather, what I hope to have clarified is the mechanics by which this meditative exercise operated and the specific purpose towards which it was aimed. The obvious next stage in pursuing this line of investigation would be to examine their gospel commentaries to see if their actual expositions of the text draw upon their meditative mnemotechniques on the Canon Tables, but that is a task that is beyond the scope of the present project.

CO NCLUSION At one level the symbolic interpretation offered by these two Armenian authors is but an extension of the numeric symbolism intrinsic to Eusebius’ paratextual system.¹⁷⁴ If the Canon Tables are the gospel reduced to numeric form, then these ten tables too would logically express the same message as the gospel itself, but whereas the gospel communicates this message in the form of four narratives of the life of Jesus, the ornamentation of the Canon Tables does so in radically different form, with abstract artistic embellishments whose theological significance can be decoded only by the one initiated into these mysteries. In this way, the Canon Tables continue to function for Step‘anos and Nerses as Eusebius’ paratext did elsewhere—as an entryway into the fourfold gospel that conditioned the reader to engage in a certain way with the sacred text that followed. However, for these two authors the paratext accomplishes this same goal without the user making any reference whatsoever to the actual numeric content of the ten tables, since it relies on the more abstract symbolism of the artistic ornamentation. In other words, what is unique in these Armenian treatises, in comparison with the other sources considered in this study, is the way they prioritize visuality as the medium best suited to provide the paratextual experience required to take up the text of the fourfold gospel. Moreover, because the mode of symbolic communication used here is more open-ended and undefined than the comparatively simple numeric symbolism of the ten tables, the ‘text’ to which the paratext relates can be broadened to ¹⁷³ Cf. Dickran Kouymjian, ‘Armenian Iconography: A New Approach’, Journal of the Society for Armenian Studies 6 (1992 1993): p. 214: ‘The very act of making a codex beautiful either through illumination, or through perfect calligraphy, or a fine binding, was considered a pious act, a prayer to God, independent of its other uses. A clear example of this concept of almost abstract beauty is articulated in the commentaries on the canon tables . . . ’ While I do not dispute Kouymjian’s claim that the making of a beautiful codex was itself regarded as a ‘pious act’, I do not think that this point is diminished by also considering the utility that that object had for its users. In fact, as I have tried to argue above, the commentaries of Step‘anos and Nerses encourage us to ask this question. ¹⁷⁴ As recognized also by Mathews and Sanjian, Armenian Gospel Iconography, p. 166.

284

The Eusebian Canon Tables

include not just the literal text of the gospels but the wider imaginative universe in which the gospelbook is situated. As a result, for Step‘anos and Nerses, Eusebius’ paratext not only represents the order of the fourfold gospel but is also understood as reflecting the divinely ordered plan of the entire sweep of redemptive history, from the origins of creation in the ineffable depths of the divine being to the consummation of history in the heavenly eschaton. Here the attempt to use Eusebius’ paratext to perceive the divinely inspired order of the sacred text and its harmony with the divinely ordained cosmological order reaches its most ambitious extent.

Conclusion I began this book by situating the Canon Tables with respect to two bodies of scholarly literature, first, paratextuality and second, information visualization. In concluding the volume, I want to revisit these two domains in reverse order and to add a comment on a third area—the fourfold gospel itself. First, then, information visualization, which is but one example of the larger category of information technologies, the subject of a forthcoming monograph by classicist Andrew M. Riggsby. Readers will recall that in the first chapter I leaned heavily on Riggsby’s survey of tables in the Roman world, emphasizing the scarcity of these devices and the difficulty of imagining their manifold applications in a culture that is not already saturated with them. Against such a backdrop, the Eusebian apparatus appears stunningly original and creative for two reasons. First, the use of a tabular matrix as a structuring system and navigation device was entirely without precedent for a corpus of literature and should be regarded as the distant ancestor of the systems of cross-referencing, citation, and indexing that are common today.¹ Secondly, the fact that Eusebius seemingly came up with his paratext by adapting the format of an astronomical table is one of the rare instances of the crossover of an information technology from one domain to another in antiquity, which again highlights the ingenuity of its creator. Indeed, as I emphasized in chapter two, following the work of Anthony Grafton and Meredith Williams, when it came to exploiting the potential of the codex as a tool for ordering and displaying information, Eusebius had no peers in antiquity. Nevertheless, Eusebius’ experiments in composition and design were not entirely sui generis, but rather were, in more general terms, indicative of the broader intellectual climate of the period. In concluding his survey of the representation of information in the Roman imperial period, Riggsby suggested that the two centuries following Diocletian witnessed ‘an information ¹ On the later, medieval background to the ‘scholarly apparatus’ typical of the modern book, see M. B. Parkes, ‘The Influence of the Concepts of Ordinatio and Compilatio on the Develop ment of the Book’, in Medieval Learning and Literature: Essays Presented to Richard William Hunt, ed. J. J. G. Alexander and M. T. Gibson (Oxford: Clarendon Press, 1976), pp. 115 41.

The Eusebian Canon Tables: Ordering Textual Knowledge in Late Antiquity. Matthew R. Crawford, Oxford University Press (2019). © Oxford University Press. DOI: 10.1093/oso/9780198802600.003.0009

286

The Eusebian Canon Tables

technology revolution’ in comparison with the knowledge schemes that preceded it, which were the focus of his monograph. He proposed that this revolution was fostered by, among other things, ‘a more centralized state as catalyst for information technology developments, a high-cultural emphasis on conservation and curation, [and] Christianity as a vector for transmitting previously idiosyncratic information devices to a broad audience’.² All three of these factors would seem to be in play with respect to Eusebius’ Canon Tables, since the popularity of his paratext probably owed something to Constantinian endorsement;³ his apparatus represented the curation of a textual corpus through the issuing of a sophisticated new edition; and his inclusion of this information device in one of the most sacred objects central to Christian ritual and practice ensured that it would be transmitted nearly everywhere the Empire’s new faith spread. The latter is in fact one of the most remarkable features of the reception history of the Eusebian apparatus. The second part of this book demonstrated the use that later readers made of Eusebius’ novel technology, in places far removed from one another geographically and chronologically. Moreover, three of the cultural traditions examined (Syriac, Hiberno-Latin, and Armenian) scarcely even had a written literary tradition prior to their influence by Christianity, which suggests that the emphasis upon texts inherent to Christian thought and practice played a particular role in the spread of this specific information device. The second factor highlighted by Riggsby has also been noted by other scholars as central to the intellectual milieu of late antiquity. For example, Scott F. Johnson has written,

² Riggsby, Mosaics of Knowledge, ‘Conclusion’. ³ Carl Nordenfalk long ago associated the Canon Tables with Constantine’s request, at some point in the 330s, for ‘fifty copies of the divine scriptures’ (σωμάτια . . . τῶν θείων δηλαδὴ γραφῶν) for churches in the imperial capital (Eusebius, VC 4.36.2 (GCS 7.1, p. 134); cf. Nordenfalk, Die spätantiken Kanontafeln, p. 50). This request, of course, is unclear regarding what kind of books the emperor desires, whether gospel codices or whole bible pandects, but in either case, there can be no doubt that at least some of the manuscripts contained the four gospels. It is unlikely that Eusebius would have drawn up the Canon Tables for the first time for these copies, since their creation would have been a complex undertaking that would have hampered the speed with which the scriptorium could comply with the imperial request. However, this surely was an ideal occasion for the propagation of his apparatus and may help to explain why it became so popular. On Constantine’s request, see Grafton and Williams, Christianity and the Transformation, pp. 215 21, who tentatively follow the earlier argument of T.C. Skeat that Codex Sinaiticus and Codex Vaticanus originated in Caesarea in relation to Constantine’s request (T. C. Skeat, ‘The Codex Sinaiticus, the Codex Vaticanus and Constantine’, Journal of Theological Studies 50 (1999): pp. 583 625). This view, however, has not convinced many New Testament scholars. For a recent discussion, see D. C. Parker, Codex Sinaiticus: The Story of the World’s Oldest Bible (London: The British Library, 2010), pp. 19 22. Sinaiticus has the marginal notations dividing the text of the gospels into the appropriate sections and Canons, but the tables themselves are now lost, or perhaps were never present. Vaticanus contains no trace of the apparatus. Skeat’s suggestion (p. 615) that Eusebius abandoned his idea of including Canon Tables out of fear of enraging Constantine with further delay is more than a little speculative.

Conclusion

287

commentary, compilation, and repackaging are all creative endeavors in their own right, and arguably part of the defining literary aesthetic of Late Antiquity. This is a period of consolidation and reorientation: knowledge from the Greco Roman civilizations was queried, repackaged, and disseminated; classical litera ture was copied, commented upon, and imitated; Roman law was collected, rearranged, and declared authoritative. . . . The organization of knowledge was one of the required principles of the literary and cartographic art of the period.⁴

Similarly, Marco Formisano has proposed that the literary aesthetic of late antiquity was marked by genres [in which] content is subordinate to form, whereby this form consists precisely in an already given content: a literary tradition that provides an already existing and codified language. Thus the constitution of a text depends on its approaching literature in a deeply analytic fashion. It is just this ability to read a work analytically and decode it be it Virgil, the Bible, or ‘scientific’ texts that presents a characteristic of Late Antiquity: the texture of the tradition is used as the template for new creations that would have been inconceivable without that form . . . One of the primary characteristics of Late Antiquity is certainly to be found in this fragmentation [evident in the practice of writing analytic commen tary on texts].⁵

The statements of Johnson and Formisano are apt descriptions of Eusebius’ achievement. The fourfold gospel, which had existed as a concept since at least the late second century and as a book form since the mid-third century, provided Eusebius with the content for his creation, and by presenting it with a paratext that enabled a finely nuanced analysis of this existing textual corpus he consolidated and repackaged it, ensuring its preservation and dissemination. Moreover, Eusebius’ treatment of this corpus as a whole that could be broken down into its constituent parts, which could then be related to other units of text similarly sectioned off from the larger whole, illustrates the same sort of ‘fragmentation’ or ‘decoding’ approach to literature that one sees in other late antique literary forms such as the cento.⁶ In fact, given that the Canon Tables were transmitted so widely and over so many centuries, they must be regarded as one of the most successful late antique attempts at curating knowledge, analogous to other better-known achievements such as Justinian’s treatment of the Roman legal tradition two centuries later.⁷ ⁴ Scott Fitzgerald Johnson, Literary Territories: Cartographical Thinking in Late Antiquity (Oxford: Oxford University Press, 2016), pp. 56 7, 59. ⁵ Marco Formisano, ‘Towards an Aesthetic Paradigm of Late Antiquity’, Antiquité Tardive 15 (2007): p. 283. ⁶ Cf. Jaś Elsner, ‘Late Antique Art: The Problem of the Concept and the Cumulative Aesthetic’, in Approaching Late Antiquity: The Transformation From Early to Late Empire, ed. Simon Swain and Mark Edwards (Oxford: Oxford University Press, 2004), pp. 271 309. ⁷ There are in fact suggestive similarities between Eusebius’ Canon Tables and Justinian’s Digest. Justinian prohibited the writing of any commentarial material within copies of his Digest,

288

The Eusebian Canon Tables

In the passage just cited, Johnson tied this curatorial aesthetic to an impulse to organize knowledge, which other scholars have also placed at the centre of late ancient epistemology. C. M. Chin and Moulie Vidas have suggested that knowing in late antiquity was conceptualized as ‘the ability to see the correct order that structures and underlies the objects of study’, which implies the corollary that seeking knowledge is ‘an act that transforms separate and even conflicting elements into coherent wholes’.⁸ This attempt to discern unity and order in the midst of multiplicity and diversity was surely the ultimate goal of the ‘conserving and curating’ tendency identified by Riggsby and Johnson as a common late antique intellectual endeavour, and here again the Canon Tables exemplify this nexus of themes. Even if one assumes, as no doubt Eusebius did, that the canon of the fourfold gospel is internally consistent as a result of being divinely given, the experience of reading it seems haphazard and disorderly, since the four narratives relate to one another in such complex ways: sometimes repeating material verbatim; often relocating even those passages that agree so closely; sometimes presenting the same material though with different wording or details; and of course each of the gospels has material unique to itself. Eusebius’ paratext showed the relation of each of the parts of this fourfold corpus to one another, with the result that his edition seemed to reveal the ordered unity that underlay its chaotic appearance. On then to paratextuality. In chapter one I cited Laura Jansen’s comment that paratexts are ‘sites of reception where readers or viewers are prompted to (re)negotiate trajectories of plotting meaning or (re)consider their own construction as audiences’.⁹ The moments of reception in chapters four to seven of this study provide several case studies illustrating this phenomenon. The Eusebian Canons were the path that guided Augustine to a greater awareness of the complexity of the interrelations amongst the gospels, which then prompted him to formulate the first general theory of gospel composition in the history of Christianity.¹⁰ The Eusebian apparatus apparently inspired someone in the Syriac tradition to improve upon Eusebius’ design by more and commanded that instructions to readers could be given ‘only through indexes and exactness of titles, which are called paratitla’ (per indices tantummodo et titulorum subtilitatem quae paratitla nuncupantur; CI 1.17.1.12 (Krüger, p. 108). These paratitla included marginal cross references keyed to other parts of the legal code, which functioned similarly to Eusebius’ paratext. Cf. Herbert F. Jolowicz and Barry Nicholas, Historical Introduction to the Study of Roman Law, 3rd ed. (Cambridge: Cambridge University Press, 1972), p. 481. ⁸ Catherine M. Chin and Moulie Vidas, ‘Introduction: Knowing’, in Late Ancient Knowing: Explorations in Intellectual History, ed. Catherine M. Chin and Moulie Vidas (Oakland, CA: University of California Press, 2015), pp. 5 6. ⁹ Jansen, ‘Introduction’, p. 7. ¹⁰ As pointed out at the conclusion to chapter four, earlier authors had passed on traditions about the sources that individual evangelists had used, but no one had formulated a theory about the process of composition in a formal and abstract manner, as did Augustine in De consensu evangelistarum.

Conclusion

289

finely nuancing what counts as a ‘similar’ passage. The same kind of reasoning, again prompted by the Canon Tables, is evident among Hiberno-Latin authors who reflected upon the various ways that texts can relate to one another, to history, and to a larger symbolic world. Finally, and perhaps most surprising of all, the decoration around the Canon Tables motivated Armenian authors to consider the wider theological context within which both text and reader were situated. Rather than merely being a useful tool during the act of reading, for these figures the Eusebian paratext indicated a process of instruction that one had to undergo prior to even reaching the sacred text. For each of these case studies, the Eusebian paratext did not dictate to the reader what to make of the text to which it related (recall the argument in chapter three that it is underdetermined), but rather enabled readers to give expression to their own assumptions about the text at hand and indeed to explore those assumptions in comparison to the actual text. Thus, for example, Augustine began his research assuming the harmony of the four gospels, and then, as he analysed the textual material made apparent by the Canon Tables, devised strategies to confirm this theological judgement. The same assumption of textual concord was shared by all the figures considered in this study, and its scope was often extended well beyond the text itself. Given the late ancient understanding of the cosmos in linguistic terms,¹¹ the harmony of a text thought to be divinely revelatory tapped into the wider cosmic sympathy characteristic of late ancient knowing.¹² In chapter two I suggested, somewhat speculatively, that Eusebius might have envisioned himself as following in the footsteps of the Creator by marking out an undifferentiated text with all kinds of numbers, resulting in a finished product that displayed the beauty of its Maker and provided its users with material for contemplation. Even if Eusebius had no such pretensions, later readers of his paratext, from Victor of Capua to Sedulius Scottus, to Dionysius bar Salibi, to Nerses Šnorhali, thought that the number of tables was not accidental but rather somehow necessary to express the concordance between the harmony of the text and the harmony of the cosmos.¹³ For Sedulius and Nerses, the number of Canon Tables was also fitting in the light of the Ten Commandments, implying that the Eusebian paratext gave expression to the sympathy not only between text and cosmos,

¹¹ Jeremy Schott, ‘Language’, in Late Ancient Knowing: Explorations in Intellectual History, ed. Catherine M. Chin and Moulie Vidas (Oakland, CA: University of California Press, 2015), pp. 58 79. ¹² Cf. Maud W. Gleason, ‘Afterword’, in Late Ancient Knowing: Explorations in Intellectual History, ed. Catherine M. Chin and Moulie Vidas (Oakland, CA: University of California Press, 2015), p. 289: ‘As the ancient world becomes late ancient, the microcosm anchors itself more firmly in the macrocosm’. ¹³ On cosmic sympathies in late antiquity, see further Catherine M. Chin, ‘Cosmos’, in Late Ancient Knowing: Explorations in Intellectual History, ed. Catherine M. Chin and Moulie Vidas (Oakland, CA: University of California Press, 2015), pp. 99 116.

290

The Eusebian Canon Tables

but also between the old and new dispensations within the text itself. This concern to find concordance within the sacred text was also evident in Ailerán’s poetic riddles, but found its fullest development in the treatments of Step‘anos and Nerses, who devised an elaborate scheme whereby the decoration could be interpreted as manifesting the continuous history of divine action in human affairs, from the eternal, ineffable divine essence to the present moment of the church’s celebrated liturgy. Thus, the Canon Tables could be used to express readers’ assumptions about the harmony existing between the various parts of a text, between text and world, between text and history, and between text and ritual performance. Yet, we miss a crucial part of the history I have traced if we think simply in terms of ‘text’. One of the developments apparent in the reception history of the Eusebian paratext is the increasing significance of the book as something more than merely the carrier of a text. A similar point has been suggested by Claudia Rapp, who asserted, ‘The holy book and the holy man are the most powerful icons for our interpretation of the culture and mentality of late antiquity’,¹⁴ a claim that is supported by the lavishness with which Canon Tables in a variety of traditions were artistically embellished. The work of classicists on Roman and Greek book cultures suggests that this mentality marks a transformation of the classical heritage. In chapter one I noted Riggsby’s hypothesis that one reason paratexts are rare in Latin literature is because ‘[e]ven in their most literary moments, Romans preferred imagining texts (at least potentially) as speech acts’.¹⁵ Reviel Netz, in the context of a study of diagrams in ancient Greek mathematical treatises, has made a similar point, specifically in contrast to the later period: it is immediately obvious why all interpretation is eschewed on the ancient papyrus. This is because interpretation, for the ancients, is understood to reside in the actual moment of performance and the roll for this reason provides no interpretation, not even punctuation and word boundaries (let alone contrasts in word size and color). This is a matter for the voice to provide and the roll merely provides the barest minimum of information required for the voice to do its part . . . The underlying point is that this is not about the papyrus. We are not using the piece of written document as a tool for thinking about Troy and Corinth, or as an object of pleasure on its own right. Rather, this is a tool meant to evoke a more central cultural phenomenon, that of the performance. It acts on the one hand to recall an originating, canonical moment of perform ance, the one constitutive of the authorial presence of a Homer, a Euripides; it acts on the other hand as a tool for the re enactment of such canonical moments of performance in the salons of the Greco Roman elite. This is the opposite of the medieval book, where it was all about the parchment, a ritual tool of devotion or

¹⁴ Rapp, ‘Aspects of Scriptural Holiness in Late Antiquity’, p. 222. ¹⁵ Riggsby, Mosaics of Knowledge, see the section ‘A Brief Orientation’.

Conclusion

291

of erudition calling for the reader to relate to it the actual physical presence of a written text as the focus of one’s cultural identity.¹⁶

Perhaps Netz has somewhat overdrawn the distinction; exceptions could most likely be proposed. Nevertheless, it seems undeniable that even with respect to the figures considered in this study one can perceive a shift in the understanding of the book, revealed in the diverse attitudes towards the Eusebian paratext. Augustine did not comment on the Canon Tables as such, but tacitly used them in writing De consensu, perhaps because he still shared the mindset of the classical world, in which a text was an imagined performance, as seen in the fact that he envisioned the evangelists composing their works in the style of ancient rhetors using the technique of memoria rerum. However, by the time we reach the Armenian authors covered in chapter seven, the book itself has taken a central place in that most central of cultural activities of the medieval world, the divine liturgy, in which the sacred gospels were solemnly processed in and placed on the altar to denote the entrance of Christ into the world and his presence within the church. Here the book is not merely the carrier of a text either representing a past performance or enabling a future one. Rather, the book as a material object is invested with meaning, deriving from the social role it performs, and, accordingly, the commentaries from these Armenian authors focus not just upon the text of the book but also upon paratextual features that can be regarded as representing in miniature, visual form the ritual activities within which the book is situated. Ailerán stands at the midpoint of this transition in that he still, like Eusebius and Jerome, imagined passages of gospel text as speech acts, but also explored the symbolic meaning of the paratext, rather than simply seeing it as a utilitarian tool for finding parallel moments of vocalization. If such a shift is accepted, then it seems that the Canon Tables must have both benefited from it and contributed to it in at least two ways. One of the primary pieces of evidence that Netz highlighted to make this contrast is the almost complete absence of images in bookrolls in ancient times, in sharp distinction to the medieval period, which saw such elaborate and sophisticated attempts at conveying meaning in artistic form within the codex. If you assume that the book is fundamentally a vehicle for an oral performance, then images seem unnecessary and are perhaps even detrimental, since they interrupt the imagined performance. However, if the book is regarded as not just the necessary equipment for the performance but an object that itself bears significance, then it seems natural to endow it with artwork that expresses the meaning that the book has in a wider symbolic universe. ¹⁶ Netz, ‘Authorial Presence’, pp. 238 9, emphasis original. On the social role of reading and books in the imperial period, see especially William A. Johnson, ‘Toward a Sociology of Reading in Classical Antiquity’, The American Journal of Philology 121 (2000): pp. 593 627, and Johnson, Readers and Reading Culture.

292

The Eusebian Canon Tables

Given that the Canon Tables are some of our earliest examples of images in books, and became more elaborate as time progressed, they illustrate well this increased sense of the book as bearer of meaning. A second way in which the Canon Tables are tied to the transformation of the book during the period in question is the way in which they act as a distraction from the imagined oral performance and thereby distance the text further from the spoken voice. A classically minded reader, such as Augustine, could have linearly read a single gospel and understood it easily in accord with the rhetorical assumptions of his day, that is, as a transcript of a past performance. However, to read a gospel text equipped with Eusebius’ paratext running as a constant companion in the margin is to experience a persistent nagging reminder that this text does not just relate to an oral performance but also relates to other written texts. And if one follows the paratextual prompting by stopping to consider how a passage at hand relates to one or more other passages, then not only has the performance been put on pause but the text itself has now become the object of scrutiny. The process of abstraction at work here is not fundamentally different from the way in which writing itself removes language from oral performance, but the Canon Tables intensify it by pressing the decontextualization still further, and it seems plausible that such decontextualization contributed to the greater focus upon the book itself that one sees in the medieval period.¹⁷ Finally, then, what does this study imply about the fourfold gospel? Imagining cosmic sympathies reflected in textual form strikes most readers today as an absurd mixing of unrelated content, akin to thinking it might be possible to deduce a theory of morality from the fact that the square of the hypotenuse of a right triangle is equal to the sum of the squares of its two sides. The fact that such a notion seems so out of place in our mental landscape suggests that deep intellectual shifts have taken place between Eusebius’ day and our own. Nevertheless, this distance between the fourth century and the twenty-first only throws into sharper relief the success of Eusebius’ paratext. One of the overarching arguments of Grafton and Williams’ book is that ‘orthodox Christianity’ created a ‘fundamental problem’ for itself ‘when it decided to incorporate the Jewish Scriptures within the nascent Christian biblical canon’, since this meant that ‘the new church founded itself on a tradition that it could never fully incorporate or reduce to sameness’.¹⁸ This ‘problem’ is, in their telling, the irritant that gave rise to the novel attempts at incorporating ‘barbarian’ sources into Greek history that one sees in the literary careers of Origen and Eusebius. The same is analogously true with ¹⁷ On the decontextualization fostered by writing and related information technologies, see Jack Goody, The Domestication of the Savage Mind, Themes in the Social Sciences (Cambridge: Cambridge University Press, 1977). ¹⁸ Grafton and Williams, Christianity and the Transformation, p. 235.

Conclusion

293

respect to the fourfold gospel canon. It too cannot be reduced to sameness, if by that one means a single, univocal voice. Rather, the decision to canonize four parallel texts is a choice for an irreducible polyphony that sometimes sounds as a harmony and sometimes as dissonance, but never as four voices singing exactly the same note. Yet it would be misleading to call this a ‘problem’, since that term implies either the possibility of resolution or the persistence of a less than ideal state of affairs, when neither of these hold true for the fourfold gospel. Instead of being problematic, the irresolvable tension inherent in the fourfold nature of the church’s gospel canon can also be viewed as divinely given and theologically generative. The crucial insight here was already expressed long ago by Augustine, who realized that the formal diversity of the church’s canon is intended as a reminder that the res about which it speaks cannot be contained within or reduced to these texts, but rather ever exceeds them, owing to the divine source from which it comes. In other words, it is not finally the words themselves that matter, as though they were a kind of ‘deified sound’, but rather the person and event about which the texts speak. Nevertheless, it is also true that this res, Jesus Christ, is to be found within this canon of texts, rather than in some abstraction either prior to or deduced from them, with the result that the texts are not merely a temporary medium along the way to attaining that res in all its fullness. Rather, the texts themselves in all their ‘diversity of agreement’ must ever be the focus of attention for the one who seeks to understand their subject matter.¹⁹ It is precisely for this reason that the Canon Tables continue to be useful even today as a paratext for the fourfold gospel, in so far as they resist a separation of the form and the content of this corpus and instead encourage readers ‘to turn with all the more attentiveness, accuracy, and love to the texts as such’,²⁰ since this is where Christ is made known, in words that are true, even if never fully adequate to their subject matter.

¹⁹ Augustine, cons. ev. 2.66.128 (CSEL 43, pp. 230 1; trans. Paffenroth, p. 238). ²⁰ Karl Barth, Church Dogmatics, trans. G. T. Thomson, G. W. Bromiley, and Harold Knight (London: T&T Clark, 2009), I/2, pp. 492 5.

APPENDIX 1

A Translation of Eusebius’ Letter to Carpianus The translation below is based on the Greek text printed in NA²⁸, pages 89* 90*.¹ Eusebius, to Carpianus, beloved brother in the Lord. Greetings. Ammonius the Alexandrian, having naturally put in a lot of hard work and study, has left behind for us the Diatessaron Gospel, in which he placed alongside the [Gospel] according to Matthew the concordant sections from the other evangelists. The unavoidable result was that the continuous thread of the other three was des troyed, as far as a text for reading is concerned. So in order to preserve the body and sequence of the other [gospels] throughout and yet still be able to know the particular passages in each evangelist where they were moved by love for the truth to speak about the same things, I took my starting points from the labour of the aforementioned man, and, with the application of a different method, I have designed ten canons for you, which are subjoined [to this letter]. The first of these contains the numbers in which the four have spoken about similar matters Matthew, Mark, Luke, and John; the second [contains the numbers] in which the three [have spoken about similar matters] Matthew, Mark, and Luke; the third [contains the numbers] in which the three [have spoken about similar matters] Matthew, Luke, and John; the fourth [contains the numbers] in which the three [have spoken about similar matters] Matthew, Mark, and John; the fifth [contains the numbers] in which the two [have spoken about similar matters] Matthew and Luke; the sixth [contains the numbers] in which the two [have spoken about similar matters] Matthew and Mark; the seventh [contains the numbers] in which the two [have spoken about similar matters] Matthew and John; the eighth [contains the numbers] in which the two [have spoken about similar matters] Luke and Mark; the ninth [contains the numbers] in which the two [have spoken about similar matters] Luke and John; the tenth [contains the numbers] in which each of them has recorded certain matters on his own. ¹ There are three previously published English translations of the entire letter. See Oliver, ‘The Epistle of Eusebius to Carpianus’, pp. 144 5; Barnes, Constantine and Eusebius, pp. 121 2; McKenzie and Watson, The Garima Gospels, pp. 221 7 (the last one includes a synoptic presentation of the translations of the Greek letter alongside the Ge‘ez version). Portions of the letter are also translated in Dungan, A History of the Synoptic Problem, pp. 109 10; Grafton and Williams, Christianity and the Transformation, p. 198; Watson, The Fourfold Gospel, p. 117.

296

Appendix 1

Now this is the subject matter (ὑπόθεσις) of the subjoined canons and here is a straightforward explanation (διήγησις) of them.² In the margin alongside each of the four gospels a sequence of section numbers is prefixed, beginning with one and carrying on successively through two and three, and so on in series throughout the entirety of the books until the end. And below each number there is a notation in red which indicates in which of the ten canons the present number occurs. So if the notation is a ‘1’ [αʹ], then it is in the first [canon]; if it is a ‘2’ [βʹ], then it is in the second [canon]; and so on for all ten. So if, on opening any of the four gospels you like, you want to study whatever chapter (κεφαλαίῳ) interests you that is, to know who has said similar things and to find out the respective passages in each [gospel] where they felt moved [to discuss] the same things then note the number prefixed to the section you’re concentrating on and search for it within the canon suggested by the red notation, and immediately, based on the titles in the heading of the canon, you will know how many and which [of the evangelists] have spoken about the matter that you are looking for. And by studying also the numbers of the other gospels that are lying adjacent in the canon you are holding, and looking for them within the respective passages of each gospel, you will find them saying similar things.

² On the sense of ὑπόθεσις and διήγησις, see Grant, Eusebius as Church Historian, p. 23; Wallraff, ‘The Canon Tables of the Psalms’, pp. 9 10. As pointed out by Wallraff, copies of the Canon Tables in some Greek manuscripts, including most notably the Rossano Gospels, are prefaced by a related title (ὑπόθεσις κανόνος τῆς τῶν εὐαγγελιστῶν συμφωνίας).

APPENDIX 2

Eusebian Parallels in Augustine’s De consensu evangelistarum What follows is not simply a list of the passages discussed by Augustine in De consensu. In other words, it is not a comprehensive biblical reference index. Rather, I have included only those passages that Augustine discusses together as some sort of parallel, along with the explicit comments he makes along the way about passages unique to a given evangelist. Hence, this table is not as comprehensive as it could be, but by following this method only those places where there is the clearest evidence of dependence on Eusebius’ Canons are included.

Section of De consensu

Eusebian Canon

Eusebian Section Nos.

Comments

2.1.2 2.4.13

Canon III

Mt §1; Lk §14; Jn §1, 3, 5

Augustine discusses only Matthew and Luke, omitting the Johannine parallels, so I have counted this parallel only once.

2.5.14 2.5.14 2.5.15 2.5.15 2.6.19

Canon V Canon XMt Canon XLk Canon XMt Canon III

Mt §3; Lk §2 Mt §4 Lk §3 Mt §4, 6 Mt §7; Lk §6; Jn §2, 25

2.12.25

Canon I

2.12.25 2.12.26 2.12.26

Canon VI Canon V Canon I

2.13.30 2.14.31 2.15.32

Canon XMt Canon I

2.16.33

Canon II

Augustine omits the Johannine parallels, so I have counted this only once. Mt §8; Mk §2; Lk §7; Jn §10 Augustine omits the Markan parallel Mt §9; Mk §3 Mt §10; Lk §8 Mt §11; Mk §4; Lk §10; Augustine includes only Jn §6, 12, 14, 28 one of the Johannine parallels (§6), so I have counted this parallel only once in the totals. Mt §13 Mt §14; Mk §5; Lk §13; Jn §15 Mt §15; Mk §6; Lk §15 (continued )

298

Appendix 2

Continued Section of De consensu

Eusebian Canon

Eusebian Section Nos.

Comments

2.16.33 2.16.33 2.17.34

Canon V Canon VI Canon IV

Mt §16; Lk §16 Mt §17; Mk §7 Mt §18; Mk §8; Jn §26

2.17.34 2.17.34

Canon XJn Canon I

Jn §16, 18 Mt §166; Mk §82; Lk §94; Jn §17, 74

2.17.35 2.17.35 2.17.35

Canon VI Canon VI Canon II

2.17.35

Canon VIII

Mt §20; Mk §9 Mt §22; Mk §11 Mt §62; Mk §13; Lk §4, 24 Augustine mentions only the Matthean and Markan passages. Mk §14; Lk §25

2.17.35 2.17.37 2.17.39 2.18.42

Canon II Canon II Canon VII Canon IV

Mt §67; Mk §15; Lk §26 Mt §21; Mk §10; Lk §32 Mt §19; Jn §19 Mt §18; Mk §8; Jn §26

Augustine presents this as parallel to Luke 4:14, unlike Eusebius, though Augustine does note that Luke omits here the specific mention of John’s being thrown into prison, which seems to be the basis of the Eusebian parallel. Augustine also does not cite the Johannine passage, though it is possible it influenced his chronological interpretation of the unique Johannine material that occurs before this point in the gospel. The basis of this Eusebian parallel is the confession of Jesus as the Christ. Augustine, however, is concerned with the problematic chronology of Peter’s name or renaming, which occurs in Mt §167 and Jn §17. Eusebius had placed Mt §167 in Canon XMt, but Augustine’s attention to this passage and its conflict with Jn §17 could have arisen from the Canon I parallel that Eusebius created. He does not mention Jn §74 at all, so I have counted this parallel only once.

Appendix 2

299

2.19.43 2.19.44 2.19.44 2.19.44 2.20.48

Canon II Canon V Canon V Canon III

Mt §63; Mk §18; Lk §33 Mt §25; Lk §46 Mt §61; Lk §64 Mt §64; Lk §65; Jn §37

2.21.51 2.22.53 2.23.54 2.23.54 2.24.55 2.24.56 2.25.57 2.25.58

Canon II Canon V Canon XLk Canon II Canon I

Mt §67; Mk §15; Lk §26 Mt §68; Lk §105 Lk §106 Mt §69; Mk §47; Lk §83 Mt §70; Mk §20; Lk §37; Jn §38

2.26.59 2.27.60 2.27.61

Canon II Canon II

Mt §71; Mk §21; Lk §38 Mt §72; Mk §22; Lk §39, 186

2.27.61 2.27.63 2.27.64 2.28.68 2.29.69 2.29.69 2.30.70

Canon II Canon II Canon XMt Canon II Canon II

2.30.70

Canon II

2.30.71 2.30.75

Canon II

2.31.78 2.32.79 2.33.80 2.33.80 2.33.80 2.34.81 2.35.82

Canon V Canon V Canon V Canon XMt Canon XLk Canon II Canon II

2.36.83 2.37.84 2.38.85 2.38.85 2.38.85 2.38.85

Canon XMt Canon V Canon II Canon XMk Canon II Canon XMt

2.39.86 2.39.86 2.39.86

Canon V Canon XLk Canon V

Mt §73; Mk §23; Lk §40 Mt §74; Mk §49; Lk §85 Mt §75 Mt §205; Mk §116; Lk §224 Mt §76; Mk §52; Lk §169 Augustine only mentions the Matthean and Markan passages. Mt §80; Mk §30; Lk §44 Augustine does not mention the Markan passage. Mt §82; Mk §53; Lk §87, Augustine is only 110 concerned with the first Lukan passage. Mt §102; Lk §69 Mt §108; Lk §115 Mt §110; Lk §118 Mt §113 Lk §117 Mt §114; Mk §24; Lk §41 Mt §116; Mk §25; Lk §42, Augustine only notes the 165, 177 first Lukan parallel. Mt §118 Mt §119; Lk §126 Mt §121; Mk §32; Lk §127 Mk §31 Mt §122; Mk §33; Lk §129 Mt §124, 126; Augustine alludes to ‘other things’ in the section from Mt 12:23 37 that are omitted by both Mark and Luke, and so unique to Matthew. Presumably he is referring to these two Canon XMt passages. Mt §127; Lk §128 Lk §131 Mt §128; Lk §132

Augustine does not discuss the Johannine parallel.

Augustine omits the Johannine parallel, perhaps assuming it is a different event. Augustine is only concerned with the first Lukan parallel.

(continued )

300

Appendix 2

Continued Section of De consensu

Eusebian Canon

Eusebian Section Nos.

2.39.86 2.40.87 2.42.89 2.42.90

Canon V Canon II Canon I

2.42.89

Canon I

Mt §129; Lk §130 Mt §130; Mk §35; Lk §82 Mt §141; Mk §50; Lk §19; Jn §59 Mt §142; Mk §51; Lk §21; Jn §35

2.43.91 2.44.92 2.44.92 2.45.93 2.45.93; 2.46.95 2.46.98 2.45.93 2.45.94 2.47.99 2.47.100 2.47.99 2.48.102

Canon II Canon VI Canon II Canon XMk Canon I

2.49.103 2.49.103 2.50.104 2.50.104 2.51.106 2.52.107 2.53.108 2.53.108

Canon VI Canon VI Canon VI Canon XMk Canon VI Canon VI Canon XMk Canon I

2.54.110

Canon II

2.54.110 2.55.111 2.55.111

Canon VI Canon II Canon XMt

Canon VIII Canon XJn Canon IV Canon XMt Canon II

Mt §143; Mk §57; Lk §90 Mt §145; Mk §60 Mt §144; Mk §59; Lk §12 Mk §62 Mt §147; Mk §64; Lk §93; Jn §49 Mk §61; Lk §91 Jn §33, 39 Mt §150; Mk §67; Jn §51 Mt §151 Mt §153; Mk §69; Lk §36

Mt §157; Mk §72 Mt §159; Mk §73 Mt §160; Mk §76 Mk §74 Mt §163; Mk §78 Mt §165; Mk §80 Mk §81 Mt §166; Mk §82; Lk §94; Jn §17, 74 Mt §168; Mk §83; Lk §95, 206 Mt §169; Mk §84 Mt §170; Mk §85; Lk §96 Mt §171

Comments

Augustine seems to have in view both the previous parallel in Canon I and this parallel, because he refers to a place in John similar to Mt §141 2, in which words are recorded that were either spoken ‘by the Lord or to him’. In Jn §35 Jesus himself speaks and in Jn §59, words are spoken to Jesus. In neither passage is it ambiguous who the speaking subject is, so Augustine’s statement is best taken as a brief allusion to both passages together.

Augustine mentions only the Matthean and Markan passages.

Augustine does not note the Johannine parallels. Augustine mentions only the first Lukan parallel. Augustine suggests that this verse has the same sense as Mark §86 (= 8:38).

Appendix 2 2.56.112 2.56.114 2.57.115 2.58.116 2.59.117 2.60.118 2.61.119

Canon II Canon VI Canon II Canon II Canon XMt Canon II

2.61.119 2.61.119

Canon VI Canon XMk

2.61.119

Canon II

2.61.119

Canon V

2.61.119 2.62.120 2.62.122 2.63.123 2.63.123

Canon VII Canon VI Canon II Canon II

301

Mt §172; Mk §87; Lk §98 Mt §173; Mk §89 Mt §174; Mk §91; Lk §99 Mt §176; Mk §93; Lk §101 Mt §177 Mt §179; Mk §99; Lk §197 Augustine does not cite these passages, but he alludes to them when he says that within Mt 18:1 35 there are some parts of the discourse that are the same in Matthew and Mark, even in the same order. He is probably referring to Mt §179 180 and Mk §99 100. He does not mention the Lukan passage. Mt §180; Mk §100 Mk §101 Augustine does not explicitly cite this passage but refers to some things that Mark has which are not in Matthew’s version of this discourse. This is the only uniquely Markan passage in this section of the gospel, so it is probably what he had in mind. Mt §178; Mk §95; Lk §102, Augustine notes only the 217 first Lukan parallel. Mt §182; Lk §187, 189 Although it is not explicit which Lukan passages Augustine has in mind in this paragraph, he refers to ‘other things’ said elsewhere in Luke, out of sequence, which are similar to the discourse of Matthew 18:1 35. There are only three Canon V parallels in this passage, so it must be one of them. The other two possible parallels are Mt §183; Lk §198 and Mt §187; Lk §199. Mt §185; Jn §215 Mt §189; Mk §103 Mt §192; Mk §106; Lk §216 Mt §193; Mk §107; Lk Augustine is only §121, 218 concerned with the second Lukan parallel. See 2.73.141 in this table for further discussion. (continued )

302

Appendix 2

Continued Section of De consensu

Eusebian Canon

Eusebian Section Nos.

Comments

2.63.123 2.64.124 2.64.124 2.64.124

Canon XMt Canon II Canon VI Canon II

2.65.125 2.65.126

Canon II

2.66.127 2.66.127 2.66.128 2.67.129

Canon XLk Canon II Canon VII Canon I

2.68.130 2.68.131 2.68.131 2.69.132 2.70.133 2.70.133 2.71.139

Canon VI Canon XMk Canon II Canon XMt Canon II Canon V

2.71.139 2.71.139

Canon XMt Canon I

2.72.140 2.73.141

Canon II Canon VI

Mt §200 Mt §201; Mk §112; Lk §222 Mt §202; Mk §113 Mt §203; Mk §114; Lk §270 Augustine notes the Lukan parallel, even though it occurs in a different narrative sequence, and so concludes that Jesus said this on multiple occasions. Mt §205; Mk §116; Lk §224 Augustine notes all three parallel passages but concludes that the Lukan version refers to a separate, but similar, event. Lk §225, 227 Mt §206; Mk §117; Lk §232 Mt §207; Jn §101 Mt §211; Mk §121; Lk Augustine notes the §238; Jn §21 parallels in all four gospels but concludes that John refers to a separate incident, so Jesus must have cleansed the temple twice. Mt §214; Mk §120 Mk §123 Mt §217; Mk §127; Lk §240 Mt §218 Mt §219; Mk §128; Lk §241 Mt §221; Lk §181 Augustine acknowledges that the two passages are similar, but concludes that they are different stories. Mt §222 Mt §220; Mk §122, 129; Augustine is only Lk §239, 242, 261; Jn §77, concerned with the 85, 88 parallels Mt §220, Mk §129, Lk §242. Mt §223; Mk §130; Lk §243 Mt §224; Mk §131 Augustine notes another possible parallel that Eusebius does not include: Lk §121 (= 10:25 28). Eusebius had included this Lukan passage as a parallel to the story of the rich young ruler, discussed by Augustine at 2.63.123. Augustine omitted discussing the passage at that point, but brings it up

Appendix 2

303 here. Although he raises the possibility that Lk §121 could be the same incident as the two passages recorded in Canon VI (Mt §224; Mk §131), in the end he decides it is a different event.

2.74.143 2.75.144

Canon II Canon V

2.75.145

Canon V

2.76.146 2.76.146

Canon VIII Canon II

2.76.146

Canon II

2.77.147 2.77.148 2.77.149 2.77.150

Canon II Canon VI Canon VI Canon II

2.77.151

Canon VI

2.77.151 2.77.151

Canon XLk Canon V

Mt §225; Mk §134; Lk §245 Mt §228; Lk §139 There are multiple parallels between Matthew and Luke in the two large passages that Augustine has in view in this paragraph, though he doesn’t go through them in detail. Hence I have included just one possible parallel. He decides that they are in fact two separate events, even though they are similar. Mt §241; Lk §175 Augustine notes these parallel passages even though, as he acknowledges, they occur in different narrative sequences. Mk §136; Lk §247 Mt §229; Mk §135; Lk Augustine is only §137, 246 interested in the second Lukan passage, which occurs in the same sequence as the Matthean and Markan parallels. Mt §242; Mk §137; Lk Augustine is only §237, 248 concerned with the second Lukan passage. Mt §243; Mk §138; Lk §249 Mt §246; Mk §140 Mt §247; Mk §142 Mt §248; Mk §143; Lk Augustine only notes the §209, 253 second Lukan parallel, which occurs in the same narrative sequence as the Matthean and Markan passages. Mt §250; Mk §145 Augustine signals that he is aware of many more agreements between Matthew and Mark in this long discourse, but does not discuss them because they pose no problems. Lk §259 Mt §265; Lk §157 In this paragraph discussing Matthew’s eschatological discourse, Augustine mentions that (continued )

304

Appendix 2

Continued Section of De consensu

Eusebian Canon

Eusebian Section Nos.

Comments Luke includes similar material in a separate discourse, even preserving Matthew’s order. The only passage that he could have in mind is Luke 12:37 47, which has three sections parallel to Matthew 24 (see the next two rows in the chart as well).

2.77.151 2.77.151 2.78.152

Canon V Canon V Canon I

2.78.153 2.79.156

Canon I

2.80.157 3.1.2

Canon II Canon I

3.1.2

Canon I

3.1.2 3.1.3

Canon VI Canon II

3.1.3 3.1.4 3.2.5 3.2.6

Canon XJn Canon IX Canon I

3.3.9 3.3.9 3.3.9 3.4.10

Canon VI Canon VI Canon VI Canon I

3.4.10 3.4.11

Canon VI Canon IV

3.4.12

Canon XLk

Mt §266; Lk §155, 157 Mt §267; Lk §158 Mt §274; Mk §156; Lk §260; Jn §20, 48, 96

Augustine mentions all three of the Johannine parallels. Mt §276; Mk §158; Lk §74; Augustine acknowledges the Jn §98 Lukan parallel, which occurs much earlier in his gospel, and argues that the same Mary anointed Jesus twice, so that there are two events being referred to. Mt §278; Mk §160; Lk §263 Mt §284; Mk §165; Lk Augustine alludes to John §266; Jn §55, 63, 65, 67 in a general way, so it is not clear which specific parallel he has in mind. Mt §280; Mk §162; Lk Augustine does not §269; Jn §122 mention the Lukan parallel. Mt §282; Mk §164 Mt §285; Mk §166; Lk §265, 267 Jn §123 Lk §262; Jn §113, 124 Mt §289; Mk §170; Lk Augustine notes all four §275; Jn §126 parallel passages, but argues that they refer to three different events. Mt §286; Mk §167 Mt §288; Mk §169 Mt §290; Mk §171 Mt §291; Mk §172; Lk §279; Jn §156 Mt §292; Mk §173 Mt §299; Mk §180; Jn §103 Augustine does not mention the Johannine parallel. Lk §283

Appendix 2 3.4.13

Canon I

3.5.16

Canon I

3.5.17 3.5.18 3.6.20 3.6.20

Canon XMt Canon VI Canon VI Canon I

3.6.20

Canon I

3.6.23 3.6.25

Canon I

3.6.26 3.7.27 3.7.28 3.7.31 3.8.34 3.8.34 3.8.35

Canon II Canon II Canon XMt Canon XLk Canon I

3.8.34

Canon IV

3.8.34 3.8.34

Canon XLk Canon II

3.8.35 3.9.36

Canon XJn Canon IV

3.9.36 3.10.37

Canon VI Canon I

3.11.38

Canon I

3.11.38

Canon IV

3.12.39

Canon I

305

Mt §295; Mk §176; Lk §282; Jn §42, 57; Augustine discusses only the Synoptic passages and makes no mention of the Johannine ones. Mt §302; Mk §183; Lk §287; Jn §160 Mt §303 Mt §305; Mk §185 Mt §309; Mk §190 Mt §310; Mk §191; Lk Augustine mentions only §297; Jn §69 the Matthean and Markan parallels. Mt §313; Mk §194; Lk Augustine does not §294; Jn §172 mention the Johannine parallel. Mt §315; Mk §196; Lk §292; Jn §175 Mt §316; Mk §197; Lk §293 Mt §317; Mk §198; Lk §295 Mt §319 Lk §301 Mt §320; Mk §200; Lk §302; Jn §178, 180 Mt §321; Mk §201; Jn §192 Augustine ignores the Johannine parallel. Lk §304, 306 Mt §308; Mk §189; Lk §305 Augustine does not explicitly discuss the Matthean and Markan passages, but simply notes that in the unique Lukan passage of Lk 23:4 12 there are some things that occur in the other evangelists. This parallel, flagged by Eusebius, is the only passage in this range that fits this description. Jn §179, 181 Mt §329; Mk §207; Jn §185, 187 Mt §330; Mk §208 Mt §331; Mk §209; Lk §315; Jn §197 Mt §332; Mk §210; Lk §318; Jn §197 Mt §333; Mk §211; Jn §203 Augustine does not discuss the Johannine parallel. See 3.17.54 in this table. Mt §334; Mk §212; Lk §321; Jn §201 (continued )

306

Appendix 2

Continued Section of De consensu

Eusebian Canon

Eusebian Section Nos.

Comments

3.13.46 3.14.51

Canon XMt Canon I

3.15.52 3.15.52 3.16.53 3.17.54 3.17.54 3.17.54

Canon VI Canon II Canon II Canon II Canon VI Canon II

3.18.55

Canon I

3.19.56 3.20.57 3.20.57 3.21.58 3.21.58 3.22.59

Canon II Canon XMt Canon II Canon VI Canon XLk Canon I

3.22.59 3.23.60

Canon XJn Canon I

3.23.60 3.24.61 3.24.62 3.24.64

Canon XJn Canon VI Canon XMt Canon II

3.24.65

Canon I

Mt §324 Mt §336; Mk §215; Lk §317, 319; Jn §198 Mt §337; Mk §217 Mt §338; Mk §218; Lk §322 Mt §339; Mk §219; Lk §325 Mt §340; Mk §220; Lk §327 Mt §341; Mk §221 Mt §342; Mk §222; Lk §323 In addition to the Synoptic passages, Augustine also here discusses Jn §203, which Eusebius had listed in Canon IV as parallel to Mt §333; Mk §211 (see section 3.11.38 for Augustine’s discussion of the latter parallel). This is one of the few genuine mistakes in Eusebian parallels. Augustine is right that Jn §203 is parallel to these passages, so this should be a Canon I parallel, not Canon II. The Peshitta version of the Canon Tables corrects the error. Mt §343; Mk §223; Lk Augustine acknowledges §329; Jn §204 the Johannine parallel but treats it as a distinct saying. Mt §344; Mk §224; Lk §328 Mt §345 Mt §346; Mk §225; Lk §330 Mt §347; Mk §226 Lk §331 Mt §348; Mk §227; Lk §332; Jn §206 Jn §205 Mt §349; Mk §228; Lk §333; Jn §208 Jn §207 Mt §350; Mk §229 Mt §351 Mt §354; Mk §233; Lk §338 Augustine does not mention the Lukan passage. Mt §352; Mk §231; Lk In 3.24.65 Augustine is §336; Jn §209, 211 only concerned with the first Johannine passage, though he discusses Jn §211 at 3.24.68.

Appendix 2

307

3.24.67 3.25.74 3.25.74 4.1.2 4.2.3 4.2.3

Canon II Canon IX Canon XJn Canon VIII Canon VIII Canon VIII

Mt §353; Mk §232; Lk §337 Lk §340; Jn §213, 217 Jn §214 Mk §12; Lk §23 Mk §14; Lk §25 Mk §16; Lk §27 Augustine does not explicitly discuss this parallel, but he notes that in the section of Mark 1:22 39 there are several things unique to Mark and Luke. The use of the plural suggests that he has in mind more than just the Canon VIII parallel of Mk §14 and Lk §25, so he probably is thinking of this parallel also, though he does not discuss it in detail since it presents no problems.

4.2.3

Canon VIII

Mk §17; Lk §28

4.3.4 4.3.4

Canon VIII Canon VIII

Mk §28; Lk §27 Mk §48; Lk §84

4.4.5

Canon VIII

Mk §61; Lk §91

Augustine stops his selection of text to be considered in this paragraph at the end of this Canon VIII passage, implying that he is aware of the parallel. Augustine does not discuss this parallel, presumably because it presents no contradictions. However, the only explanation for why he has chosen to treat Mark 3:13 5:20 as a section is that Mark 5:20 marks the end of a Canon VIII parallel, which Augustine is treating in this section. However, after stopping at this point, he seems to have forgotten his purpose and reverts to an earlier discussion about a possible contradiction regarding Peter’s naming. Then he moves on to the next passage without ever treating the Canon VIII parallel. (continued )

308

Appendix 2

Continued Section of De consensu

Eusebian Canon

Eusebian Section Nos.

Comments

4.4.5

Canon VIII

Mk §56; Lk §89

Augustine does not explicitly discuss this parallel, but it seems that he is aware of it. In this paragraph he treats the section from Mk 5:21 to Mk 6:30, stopping at 6:30 because this presents a Canon VIII parallel, which this section of book four is devoted to. However, after noting the parallel at Mk 6:30, he briefly comments that ‘the rest has already been discussed’ in this range of verses from 5:21 to 6:30. Since he has already stated that he is dealing only with parallels between Mark and Luke, there would be no reason for him to give this explanation unless he was aware that he was skipping over a Canon VIII parallel without comment. Indeed, although he had not explicitly dealt with this parallel (Mk §56; Lk §89), he did give an extensive treatment of the missionary discourse from which these passages come (see De consensu 2.30.70 7).

4.4.5 4.5.6 4.6.7 4.6.7 4.6.7 4.6.7 4.6.7 4.7.8 4.9.10

Canon VIII Canon VIII Canon XMk Canon VI Canon VI Canon II Canon II Canon VIII Canon IX

Mk §75; Lk §100 Mk §97; Lk §103 Mk §101 Mt §100; Mk §98 Mt §180; Mk §100 Mt §179; Mk §99; Lk §197 Mt §31; Mk §102; Lk §185 Mk §136; Lk §247 Lk §30; Jn §219, 222

Appendix 2

309

Summary Table Number of Eusebian Parallels Canon I Canon II Canon III Canon IV Canon V Canon VI Canon VII Canon VIII Canon IX Canon XMt Canon XMk Canon XLk Canon XJn

74 111 22 25 82 47 7 13 21 62 19 72 96

Number of Parallels Noted by Augustine

Percentage

35 57 3 7 22 33 3 10 6 19 6 12 10

47% 51% 14% 28% 27% 70% 43% 77% 29% 31% 32% 17% 10%

APPENDIX 3

The Gospel Synopsis in Codex Climaci Rescriptus and its Possible Connection to Ammonius’ Diatessaron-Gospel In chapter two of this book I suggested that no trace of the gospel synopsis produced by Ammonius of Alexandria has survived to the present day, and in making this claim I was merely following the scholarly consensus on this question. There is, however, a little known exception to this consensus which warrants discussion here as an appen dix to the present volume owing to its possible relation to Eusebius’ Canon Tables. As suggested by its name, the manuscript known as Codex Climaci Rescriptus (CCR) is a palimpsest whose overtext is a Syriac translation of two works by the Byzantine ascetic John Climacus, head of St Catherine’s Monastery at Mt Sinai in the sixth century. The final manuscript that has come down to us is a product of the ninth century, but, in order to create the palimpsest, the ninth century scribe reused portions of ten previous manuscripts, now known as CCR1 10. The manuscript is well known in some specialist circles owing to the fact that six of its constituent parts contain the largest surviving corpus of Christian Palestinian Aramaic. However, the second most sizeable portion of the manuscript, now known as CCR5 and running to twenty four folios,¹ has as an undertext portions of the gospels in Greek, which has received very little scholarly attention despite the fact that it must count as one of the most unusual witnesses to the text of the gospels to have come down to us. A portion of this Greek undertext was published in 1909 by Agnes Smith Lewis, who was involved in the purchase of the manuscript and its relocation to Cambridge.² In 1956 Ian Moir published a fuller edition, after having applied ultraviolet light to the palimpsest to recover much more of the Greek undertext. Although nothing more has been published on CCR5 since Moir’s edition, a research team in Cambridge will soon be producing a new edition with commentary and analysis based on a re imaging of the manuscript using the latest technology. The present discussion was undertaken with some awareness of the preliminary findings of this research project, and although the following analysis depends solely on Moir’s study, so far nothing I have seen of the new research would invalidate the conclusions presented here. Of course, further discoveries made possible by new imaging of the manuscript could change the picture considerably. The reason for the inclusion of a discussion of CCR5 here is that a portion of the text of the Greek gospels in CCR5 presents parallel passages from the gospels collected together in synoptic like fashion, leading an anonymous reviewer of Lewis’ 1909 publication to comment that ‘it does not at all agree with what we have of Tatian’s ¹ CCR5 consists of fol. 65 72 and 81 96 of the current manuscript. ² Agnes Smith Lewis, Codex Climaci Rescriptus, Horae Semiticae VIII (Cambridge: The University Press, 1909), pp. xxvii xxx.

Appendix 3

311

harmony (Ciasca, Fuldensis, etc.), and the question arises whether it represents that hardly less famous ancient harmony made by Ammonius in the third century, on the basis of Matthew’.³ Moir was cautious in his judgements, admitting his puzzlement over several features of this manuscript. He did, however, point out that the text in CCR5 could not be a ‘gospel harmony’ but might represent ‘the rough drafts of some worker who planned to produce an early equivalent of Tischendorf ’s or Huck’s synopses. That such works were common we know from the examples of Ammonius and Eusebius’.⁴ The possible relation of this work to either Ammonius’ Diatessaron Gospel or Eusebius’ Canon Tables is a question that the Cambridge team is currently investigating. The goal of this appendix is to highlight some of the issues that must be taken into account in forming any such hypothesis. The undertext contained in CCR5, which Moir dated to 650 700 CE,⁵ falls into three discontinuous sections: Matthew 2:12 13:46, John 6:53 21:2, and Matthew 26:75 27:40 with parallels. The manuscript presents multiple unsolved riddles (e.g., the passages that for no apparent reason are missing from otherwise sequential sections of narrative text) and here I will focus only on the two features pertinent to the present monograph. First, all three sections of text are demarcated with an enumeration system that is completely unique, without parallel in any other surviving manuscript.⁶ Like the Eusebian apparatus, the system in CCR5 numbers each gospel separately, starting with 1 at the beginning of each narrative and continuing sequen tially to the end of each text. The most striking feature of this enumeration is the fine degree of granularity that it exhibits. The numbered sections vary in length but on average are shorter than Eusebian sections, with the result that the total number of sections in each gospel is greater even than the revised Canon Tables version found in Peshitta gospelbooks. For Matthew, the earliest number recorded is §11, correspond ing to a section consisting of Matthew 2:13, and the highest number recorded is §568, corresponding to a section beginning at Matthew 27:37.⁷ As discussed in previous chapters, Eusebius’ Greek enumeration system for the gospel of Matthew ran to §355 and the revised Peshitta version went as far as §426, so without even reaching the end of Matthew the system in CCR5 has already far surpassed the totals in the two versions of Eusebius’ apparatus. Similarly, the earliest portion of John included in CCR5 is numbered §163 (= John 6:63) and the last portion of text is §490 (= John 21:1), well surpassing the totals in the two versions of the Canon Tables (§232 and §271). The highest number for Mark in CCR5 is §396 (= Mk 15:26) and for Luke it is §560 (= Lk 23:38), again well above the total Eusebian figures for those gospels (§233 and §342 respectively). This increase in the number of sections in each gospel is explained by the different rationale for the partitioning in comparison with Eusebius’ system. Whereas Euse bius started a new numbered section whenever the text changed its relation to one or more of the other gospels, the method of the system in CCR5 seems to have been to divide the text for almost any reason conceivable, including change of speaker ³ Anonymous, American Journal of Theology 14 (1910): p. 325. ⁴ Ian Moir, Codex Climaci Rescriptus Graecus, Texts and Studies 2 (Cambridge: Cambridge University Press, 1956), p. 19. ⁵ On the dating, see Moir, Codex Climaci Rescriptus Graecus, p. 11. ⁶ Moir, Codex Climaci Rescriptus Graecus, p. 20. ⁷ For these figures and those that follow, see the table in Moir, Codex Climaci Rescriptus Graecus, pp. 101 4.

312

Appendix 3

(Mt §127 9 = Mt 9:27 8a, 28b, 28c [Moir, p. 50]), change of subject of the main verb (Mk §383 5 = Mk 15:10, 11, 12 [Moir, pp. 71 2]), citation of Old Testament material (Mt §12 = Mt 2:17 18 [Moir, p. 29]), as well as intuitive narrative breaks like the start of a new parable (Mt §214 16 = Mt 13:36 43, 44, 45 [Moir, pp. 35 6]). As one would expect, some of these breaks coincide with the section breaks created by Eusebius (and perhaps by implication those breaks previously made by Ammonius), but others cut across the Eusebian sections. Some overlap in the two systems of demar cation would be expected simply based on chance, so the convergences that occa sionally appear in the start and end of sections cannot be taken as evidence of a link to either Ammonius or Eusebius. It is not impossible that the creator of this enumeration system was aware of Eusebius’ Canon Tables (indeed, it would be surprising if a scribe of the seventh or eighth century was not), but it does not in any necessary way depend upon Eusebius’ prior creation and the differing rationales for the two systems of demarcation point in the opposite direction. The only admittedly speculative thing that could be said is that a system such as this could have served as a precursor to a cross referencing tool like the Canon Tables. If one had Eusebius’ goal of finding all gospel parallels and had nothing more than a completely undifferentiated text to work with, one possible initial step would be to divide the text in every conceivable place and then recombine the sections as necessary as one identified the parallels that exist across other gospels. This hypothesis is impossible to confirm and surely other explanations for the paratextual features of this manuscript could be offered. Whatever the case, the numbered demarcation system in CCR5 is remarkable for its idiosyncrasy and raises the question of what other numbering systems might once have existed but have since perished. Hence, while it cannot be excluded that this is a lone preserved fragment of Ammonius’ Diatessaron Gospel, neither is there evidence to suggest that it is. Nevertheless, even if there is no direct connection to Ammonius or Eusebius, the fact that a seventh or eighth century scribe copied the work testifies to a broader interest in the issues of demarcation and paratextuality in the Byzantine period. The second relevant feature of CCR5 is just as striking. As noted above, the third section of text comprises Matthew 26:75 27:40 with parallels. It is the latter that once again is utterly idiosyncratic in comparison with the rest of the manuscript tradition of the Greek gospels. The text of CCR5 is written in two columns throughout, as is typical for Greek gospel manuscripts. However, for this section of text from the Passion narrative of Matthew the columns do not simply present a continuous running text of the first gospel, but instead intersperse Matthew’s text with the parallels from one or more of the other three gospels, producing something akin to a modern gospel synopsis. The most significant difference between a modern synopsis and the mis en page of this section of text in CCR5 is that the parallels in CCR5 are read down a single column rather than across the columns. So, for example, on fol. 95r (Moir, p. 69) one first reads Matthew 27:35 6, and immediately below the text of Matthew one finds Mark 15:24, followed by Luke 23:34 and next John 19:23 24a, which carries over into the first six lines at the top of the next column. Immediately following the Johannine passage the scribe picks back up with where he left off the text of the first gospel, copying Matthew 27:37, once again followed by parallels underneath. The features that caused Lewis’ anonymous 1910 reviewer to propose that this manuscript might be a fragment of Ammonius’ work are obvious. No other Greek gospel manuscript to have survived gathers parallel gospel passages in this manner,

Appendix 3

313

and Ammonius is the only person from antiquity who we know engaged in this kind of work. Moreover, the presentation of parallels in CCR5 gives priority to Matthew, as also did Ammonius, according to Eusebius’ report in the Letter to Carpianus. Furthermore, there is significant overlap in the parallels presented, bearing in mind, however, that this is only a short fragment of text. To take the example just mentioned, here is Matthew 27:35 6 and parallels in CCR5 in comparison with those of Eusebius: CCR5 Eusebius

Mt 27:35 6 Mt 27:35 6

Mk 15:24 Mk 15:24

Lk 23:34b Lk 23:34b 35a

Jn 19:23 24a Jn 19:23 24b

As one can see, the parallels mostly correspond but disagree on where to conclude the sections from the final two gospels. From what I can tell, once again it is possible, but impossible to confirm, that this juxtaposition of parallels could be somehow related to the works of Ammonius and Eusebius. A thorough analysis of the new images of the manuscript might help to confirm or disprove any such connection, particularly if other marginal paratextual features are revealed. Whatever new discoveries come to light, there are, in my view, three issues that need to be considered. First, it is just as likely (in fact a priori more likely) that this is a system of parallels inspired by Eusebius’ creation as (than) it is that this is the work of Ammonius upon which Eusebius drew to create his paratextual apparatus. Secondly, the relation of this synopsis material to the wholly idiosyncratic enumeration scheme used throughout CCR5 is not readily apparent. As noted above, the enumeration of the text is not designed to facilitate the presentation of cross references and at best could be considered a step along the way towards the kind of numbering system Eusebius produced. It seems unlikely that these two unique features of CCR5 should be unrelated to each other, but it is not clear that they are serving the same purpose, particularly given that the whole of the text is enumerated while only a portion of it is presented in a synopsis format. Finally, it may be significant that it is only the Passion narrative that is presented in synopsis form. The Harklean translation of the Syriac gospels includes, following the usual text of the four gospels, something usually called a Passion harmony, which consists of a combined version of the Passion narratives drawn from all four gospels. The purpose for this Passion harmony is not stated but it would have an obvious liturgical utility, so it may be that it was created to be used for the gospel readings during Holy Week. At the very least, it demonstrates that in the early seventh century when the Harklean was translated from the Greek, some had a particular interest in harmonizing the Passion accounts, and this might be related to the Passion text presented in synopsis form in CCR5. In other words, we should not assume that simply because the Passion account of CCR5 is presented in this form it must derive from a synopsis of the entirety of the fourfold gospel or was intended to lead to such. It could simply be another instance of scribal experimenta tion with this most important of passages from the gospels, perhaps with some liturgical use in view.

APPENDIX 4

Theophanes the Grammarian’s Note about Canon Tables In his influential discussion of the Canon Tables apparatus, Hermann von Soden pointed out that in twenty five per cent of the codices he examined he discovered numbers in the Canon Tables that diverged from the original figures Eusebius had designed. This lamentable state of affairs he found confirmed by a much earlier scholar who had undertaken a similar investigation and who left behind a ‘short essay’ (‘kleinen Aufsatz’) about his own efforts at investigating the textual transmis sion of the Eusebian apparatus and correcting the mistakes that had crept into the tradition.¹ The relevant note occurs on fol.7v 8v in a fourteenth century gospelbook held by the National Library of Greece (no.92; GA number 1410),² and is attributed to an otherwise unknown ‘Theophanes the grammarian’. In 1984 Carl Nordenfalk again drew attention to this Theophanes and questioned whether his efforts at restoring the Eusebian original might have left traces in the manuscript tradition.³ Answering this question is beyond the scope of the present monograph but would naturally be taken into consideration in the preparation of a critical edition of the Eusebian tables. However, it does seem appropriate to include a translation of the relevant text, given the relative difficulty of tracking down Theophanes’ note and what it reveals about the reception history of the apparatus, namely that all of the copies of the Canon Tables he could find were so divergent from one another that he had to recreate the ten prefatory tables on the basis of closely studying the section division and enumeration within each gospel.⁴ A further division (ἐπιδιαίρεσις), exact and without error, of the numbers contained in the ten Canons of the four holy gospels, which was made by Theophanes the Grammarian of Kratrianos. Eusebius, named for piety, taking his starting point⁵ from the outline (ὑποτύπωσις) produced by the labour⁶ of Ammonius of Alexandria, with skill and accuracy ¹ Von Soden, Die Schriften des Neuen Testaments, pp. 392 3. A few years later the note of Theophanes was also highlighted by Nestle, ‘Die Eusebianische Evangeliensynopse’, pp. 45 6. ² On this manuscript, see Anna Marava Chatzinicolaou and Christina Toufexi Paschou, Catalogue of the Illuminated Byzantine Manuscripts of the National Library of Greece, Volume 2: Manuscripts of New Testament Texts 13th 15th Century (Athens: Publications Bureau of the Academy of Athens, 1985), pp. 223 224. ³ Nordenfalk, ‘Some Textual Problems’, pp. 101 2. ⁴ Translation based on the text printed in Alchibiades Sakkelion, ΚΑΤΑΛΟΓΟΣ ΤΩΝ ΧΕΙΡΟΓΡΑΦΩΝ ΤΗΣ ΕΘΝΙΚΗΣ ΒΙΒΛΙΟΘΗΚΗΣ ΤΗΣ ΕΛΛΑΔΟΣ (Athens, 1892), ms. no. 92, pp.17 18 of the catalogue. Digitized images of the manuscript may also be found at http:// www.csntm.org/manuscript/View/GA 1410 [Accessed 20 12 2017]. ⁵ ἀϕορμῆς, also used in the Letter to Carpianus to refer to what Eusebius ‘took’ from Ammonius. ⁶ πονηθείσης echoes Eusebius’ πολλὴν . . . ϕιλοπονίαν in reference to Ammonius.

Appendix 4

315

marked out⁷ the ten Canons of the four holy gospels, as he himself testifies in writing to Carpianus. By the numbers enclosed within each Canon, he clearly showed how many of the holy theologians and evangelists have said similar things⁸ on a given subject, as well as the location of these passages and their wording. And it was such a clear work, and so well divided, and most carefully arranged. But many of those who copy the divine book of the holy gospels, being overcome by inattention and considering it in a superficial and haphazard way, extended the numbers in the ten Canons without end and mixed them up: they inserted additions and multipli cations into it in some places and omissions in others, and even substituted in different [numbers] with great licence. So I conducted a more diligent reading and painstaking investigation, and, after gathering together many of the copies, I found that they didn’t match at all and were all corrupted⁹ by scribal error and in a state of disorder. I therefore realized that I had to attend to the text¹⁰ of the four holy gospels and to select from there the section numbers contained in each of the evangelists, and thus to learn the truth. Once this had been completed and the truth had been brought to light with certainty, in the present outline only those section numbers were included which Eusebius, who originally composed the Canons with skill and rigour, also marked down¹¹ in his; that is, for the first Canon all the [sections], along with their type and contents, that were inscribed in St Matthew and St Mark and St Luke and St John too. And proceeding on for the other Canons also, the same division and outline (ὑποτύπωσις) were made accurately and with love of the truth,¹² with every addition and omission driven out. Therefore, let anyone who studiously¹³ reads the present outline (ὑποτύπωσις) of the Canons clearly know that nothing extra or missing will be found in them, but that those who also want to include an outline of the Canons in any of the copied books will gain certainty and surety from this one.

⁷ ⁸ ⁹ ¹⁰ ¹¹ ¹² ¹³

διεχάραξεν, also used in the Letter to Carpianus. παραπλήσια, also used in the Letter to Carpianus. διεϕθαρμένα; used also in the Letter to Carpianus. ὕϕει, also used in the Letter to Carpianus. ὑπεσημειώσατο; Eusebius used ὑποσημείωσις in the Letter to Carpianus. ϕιλαλήθως; also used in the Letter to Carpianus to describe the evangelists. ϕιλοπόνως, echoing ϕιλοπονίαν in the Letter to Carpianus.

Bibliography Biblical Versions Childers, Jeff W. and George A. Kiraz. The Syriac Peshiṭta Bible with English Trans lation: Matthew. Piscataway, NJ: Gorgias Press, 2012. Childers, Jeff W. and George A. Kiraz. The Syriac Peshiṭta Bible with English Trans lation: Mark. Piscataway, NJ: Gorgias Press, 2012. Childers, Jeff W. and George A. Kiraz. The Syriac Peshiṭta Bible with English Trans lation: Luke. Piscataway, NJ: Gorgias Press, 2013. Childers, Jeff W., James Prather, and George A. Kiraz. The Syriac Peshiṭta Bible with English Translation: John. Piscataway, NJ: Gorgias Press, 2014. Fischer, Bonifatius, et al. Biblia sacra iuxta vulgatam versionem. 3rd edn. Stuttgart: Deutsche Bibelgesselschaft, 1983. Nestle, E., B. Aland, and K. Aland. Novum Testamentum Graece. 28th edn. Stuttgart: Deutsche Bibelgesellschaft, 2015. Pusey, P. E. and G. H. Gwilliam. Tetraeuangelium Sanctum: Simplex Syrorum Versio. Oxford: Clarendon Press, 1901. Reference Works Bedrossian, Matthias. New Dictionary, Armenian English. Venice: S. Lazarus Arme nian Academy, 1875 1879. Brock, Sebastian P. Aaron Michael Butts, George Anton Kiraz, and Lucas Van Rompay, eds. The Gorgias Encyclopedic Dictionary of the Syriac Heritage. Piscat away, NJ: Gorgias Press, 2011. LSJ: https://www.tlg.uci.edu/lsj/. Primary Sources The primary sources cited above are here collected along with the abbreviations and critical editions I have used. The translations I have consulted are also listed, though in most instances I have made modifications to the passages I have quoted, without noting in every instance when I have done so. ‘Abdisho’ bar Brikha, cat. = Catalogus Librorum Syrorum G. S. Assemani. Bibliotheca Orientalis, Vol. 3.1. Rome: Sacra Congreg. de Propaganda Fide, 1725. Acts of the Council of Chalcedon Richard Price and Michael Gaddis. The Acts of the Council of Chalcedon. 3 vols. Translated Texts for Historians 45. Liverpool: Liverpool University Press, 2005.

318

Bibliography

Ailerán of Clonard, Can. ev. = Canon Evangeliorum David Howlett. ‘Hiberno Latin Poems on the Eusebian Canons’. Peritia 21 (2010): 162 71. Ailerán of Clonard, Interpretatio mystica et moralis progenitorum domini Iesu Christi Aidan Breen. Ailerani Interpretatio mystica et moralis progenitorum domini Iesu Christi. Celtic Studies. Dublin: Four Courts Press, 1995. Alcuin, Textum si cupies canonis Donatien De Bruyne, Pierre Maurice Bogaert, and Thomas O’Loughlin. Prefaces to the Latin Bible. Studia traditionis theologiae 19. Turnhout: Brepols, 2015. Page 186. Anatolius of Laodicea, Canon paschalis PG 10, col. 209 22. Daniel P. McCarthy and Aidan Breen. The Ante Nicene Christian Pasch: De Ratione Paschali: The Paschal Tract of Anatolius, Bishop of Laodicea. Dublin: Four Courts Press, 2003. Anatolius of Laodicea, decad. = Περὶ δεκάδος καὶ τῶν ἐντὸς αὐτῆς ἀριθμῶν J. L. Heiberg. Anatolius. Sur les dix premiers nombres. Macon: Protat, 1901. Anatolius of Laodicea, Fragmenta ex libris arithmeticorum PG 10, col. 232 36. Aristotle, de Mem. = De memoria et reminiscentia W. D. Ross. Aristotle. Parva naturalia. Oxford: Clarendon Press, 1955. Athenaeus, Deip. = Deipnosophistae G. Kaibel. Athenaei Naucratitae deipnosophistarum libri xv. 3 vols. Leipzig: Teubner, 1887 1890. Augustine, conf. = Confessiones Lucas Verheijen. Sancti Augustini Confessionum libri XIII. CCSL 27. Turnhout: Brepols, 1981. Augustine, doc. Chr. = De doctrina Christiana Josef Martin and Klaus Detlef Daur. Sancti Aurelii Augustini De doctrina Christiana. De vera religione. CCSL 32. Turnhout: Brepols, 1962. Augustine, cons. ev. = De consensu evangelistarum Franz Weihrich. Sancti Avreli Avgvstini Opera (Sect. III Pars 4): De consensv evange listarvm libri qvattvor. CSEL 43. Vindobonae: F. Tempsky, 1904.

Bibliography

319

Kim Paffenroth. ‘Agreement among the Evangelists’. In New Testament I and II. The Works of Saint Augustine I/15 & I/16. Hyde Park, NY: New City Press, 2014. S. D. F. Salmond and M. B. Riddle. ‘St Augustin. The Harmony of the Gospels’. In NPNF 1st Series, vol. 6. Peabody, MA: Hendrickson, 2004. Augustine, ep. = epistulae Al. Goldbacher. S. Aureli Augustini Operum Sectio II. S. Augustini Epistulae Pars I. CSEL 34.1. Vindobonae: Tempsky, 1895. Al. Goldbacher. S. Aureli Augustini Operum Sectio II. S. Augustini Epistulae Pars II. CSEL 34.2. Vindobonae: Tempsky, 1898. Augustine, Io. ev. tr. = In Iohannis evangelium tractatus R. Willems. Sancti Aurelii Augustini In Iohannis Evangelium tractatus CXXIV. CCSL 36. Turnhout: Brepols, 1954. Bar Hebraeus, horr. = Horreum mysteriorum G. S. Assemani. Bibliotheca Orientalis, Vol. 1. Rome: Sacra Congreg. de Propaganda Fide, 1725. Boethius, arith. = De institutione arithmetica H. Oosthout and J. Schilling. Anicii Manlii Severini Boethii de arithmetica. CCSL 94A. Turnhout: Brepols, 1999. Carmen paschale PL 19, col. 591 Cassiodorus, inst. = Institutiones divinarum et saecularium litterarum R. A. B. Mynors. Institutiones. Oxford: Clarendon Press, 1961. Cicero, De or. = De oratore E. W. Sutton and H. Rackham. On the Orator: Books 1 2. LCL 348. Cambridge, MA: Harvard University Press, 1942. CI = Codex Iustinianus P. Krüger. Codex Iustinianus, editio maior. Berlin: Weidmann, 1877. Diodorus Siculus, bibl. hist. = Bibliotheca historica C. T. Fischer. Diodori bibliotheca historica, Vol. IV. Leipzig: Teubner, 1906. ps Dionysius the Areopagite, e.h. = De ecclesiastica hierarchia Robert W. Thomson. The Armenian Version of the Works Attributed to Dionysius the Areopagite. CSCO 488, Scriptores Armeniaci 17. Louvain: Peeters, 1987.

320

Bibliography

G. Heil and A. M. Ritter. Corpus Dionysiacum II: Pseudo Dionysius Areopagita. De coelesti hierarchia, de ecclesiastica hierarchia, de mystica theologia, epistulae. PTS 36. Berlin: De Gruyter, 1991. Colm Luibheid and Paul Rorem. Pseudo Dionysius: The Complete Works. New York: Paulist Press, 1987. Dionysius bar Salibi, comm. ev. = Commentarii in evangelia I. Sedlacek and I. B. Chabot. Dionysii bar Salibi commentarii in evangelia. CSCO 15. Paris: E. Typographeo reipublicae, 1906. Epiphanius, mens. = De mensuris et ponderibus E. Moutsoulas. ‘Τὸ “Περὶ μέτρων καὶ σταθμῶν” ἔργον Ἐπιϕανίου τοῦ Σαλαμῖνος’. Θεολογία 44 (1973): 157 98. Epiphanius, pan. = Panarion Karl Holl and Jürgen Dummer. Epiphanius II: Panario haer. 34 64. GCS 31. Berlin: Akademie Verlag, 1980. Eunapius, vit. = Vitae philosophorum et sophistarum Wilmer C. Wright. Philostratus: Lives of the Sophists. Eunapius: Lives of the Philo sophers and Sophists. LCL 134. London: Heinemann, 1921. Eusebius, ep. Carp. = Epistula ad Carpianum Armenian: P. A. Vardanian. Handes Amsorya 42 (1928): 291 302. Coptic: Ad. Hebbelynck. ‘Les κεϕάλαια et les τίτλοι des évangiles’. Le Muséon 41 (1928): 114 18. Ge‘ez: Alessandro Bausi. ‘La versione etiopica della Epistola di Eusebio a Carpiano’. In Aethiopia fortitudo ejus. Studi in onore di Monsignor Osvaldo Raineri in occasione del suo 80° compleanno, pages 107 35. Edited by Rafał Zarzeczny. Orientalia Christiana Analecta 298. Rome: Pontificio Istituto Orientale, 2015. Greek: E. Nestle, B. Aland, and K. Aland. Novum Testamentum Graece. 28th edn. Stuttgart: Deutsche Bibelgesellschaft, 2015. Pages 89* 90*. Harklean: Samer Soreshow Yohanna. The Gospel of Mark in the Syriac Harklean Version: An Edition Based Upon the Earliest Witnesses. Biblica et Orientalia 52. Rome: Gregorian & Biblical Press, 2015. Pages 69 72. Latin: Donatien De Bruyne, Pierre Maurice Bogaert, and Thomas O’Loughlin. Prefaces to the Latin Bible. Studia Traditionis Theologiae 19. Turnhout: Brepols, 2015. Page 157. Peshitta: P. E. Pusey and G. H. Gwilliam. Tetraeuangelium Sanctum: Simplex Syrorum Versio. Oxford: Clarendon Press, 1901. Pages 2 5. Eusebius, chron. = Chronicle Joannes Baptista Aucher. Eusebii Pamphili Caesariensis Episcopi Chronicon Biparti tum, Pars I. Venice: Typis coenobii PP. Armenorum in insula S. Lazari, 1818.

Bibliography

321

Eusebius, ecl. = Eclogae propheticae T. Gaisford. Eusebii Pamphili episcopi Caesariensis eclogae propheticae. Oxford: Oxford University Press, 1842. Eusebius, l.C. = De laudibus Constantini Ivar A. Heikel. Eusebius Werke, Erster Band. GCS 7. Leipzig: J. C. Hinrich, 1902. H. A. Drake. In Praise of Constantine: A Historical Study and New Translation of Eusebius’ Tricennial Orations. Berkeley, CA: University of California Press, 1976. Eusebius, HE = Historia ecclesiastica Gustave Bardy. Eusèbe de Césarée. Histoire ecclésiastique. 4 vols. SC 31, 41, 55, 73. Paris: Éditions du Cerf, 1952; 1955; 1958; 1960. Eusebius, PE = Praeparatio evangelica Karl Mras and Édouard des Places. Eusebius Werke VIII. GCS 8.1 8.2. Berlin: Akademie Verlag, 1982 3. Eusebius, pin. = Pinax for the Psalms Martin Wallraff. ‘The Canon Tables of the Psalms: An Unknown Work of Eusebius of Caesarea’. Dumbarton Oaks Papers 67 (2014): 1 14. Eusebius, qu. Marin./Steph. = questioniones evangelicae ad Marinum/ad Stephanum Claudio Zamagni. Eusèbe de Césarée: Questions évangéliques. SC 523. Paris: Les Éditions du Cerf, 2008. Roger Pearse et al. Eusebius of Caesarea: Gospel Problems and Solutions. Ipswich: Chieftain Publishing, 2010. Eusebius, qu. Steph. suppl. = Supplementa ad quaestiones ad Stephanum PG 22, col. 957 976. Eusebius, v.C. = Vita Constantini F. Winkelmann. Eusebius Werke 1.1: Über das Leben des Kaisers Konstantin. GCS 7.1. Berlin: Akademie Verlag, 1975. Euth. = Euthalian Apparatus Vemund Blomkvist. Euthalian Traditions: Text, Translation and Commentary. TU 170. Berlin: De Gruyter, 2012. Fortunatianus, rhet. = Ars rhetorica C. Halm. Rhetores latini minores. Leipzig: Teubner, 1863.

322

Bibliography

Mary Carruthers and Jan M. Ziolkowski. The Medieval Craft of Memory: An Anthology of Texts and Pictures. Philadelphia, PA: University of Pennsylvania Press, 2004. Germanus, hist. = Historia mystica ecclesiae catholicae Paul Meyendorff. St. Germanus of Constantinople: On the Divine Liturgy. Crestwood, NY: St. Vladimir’s Seminary Press, 1984. Ps Iamblichus, arith. = Theologoumena arithmeticae V. de Falco. [Iamblichi] theologoumena arithmeticae. Leipzig: Teubner, 1922. Robin Waterfield. The Theology of Arithmetic. Grand Rapids, MI: Kairos, 1988. In primo certe canone Donatien De Bruyne, Pierre Maurice Bogaert, and Thomas O’Loughlin. Prefaces to the Latin Bible. Studia Traditionis Theologiae 19. Turnhout: Brepols, 2015. Page 186. Irenaeus, haer. = Adversus haereses Adelin Rousseau and Louis Doutreleau. Irénée de Lyon. Contre les hérésies, livre 3. Sources Chrétiennes 210, 211. Paris: Éditions du Cerf, 1974. Irish Reference Bible (Pauca problesmata de enigmatibus ex tomis canonicis) G. MacGinty. The Reference Bible * Das Biblewerk: Pauca Problesmata de Enigmatibus ex Tomis Canonicis: Praefatio et Libri de Pentateucho Moysi. CCCM 173. Turnhout: Brepols, 2000. Paris, Bibliothèque nationale, lat. 11561. Isidore, etymol. = Etymologiae W. M. Lindsay. Isidori Hispalensis episcopi Etymologiarvm sive Originvm Libri XX. Oxford: E Typographeo Clarendoniano, 1911. Stephen A. Barney, W. J. Lewis, J. A. Beach, and Oliver Berghof. The Etymologies of Isidore of Seville. Cambridge: Cambridge University Press, 2006. Jerome, chron. = Chronicle Rudolf Helm. Die Chronik des Hieronymus. GCS 47. Berlin: De Gruyter, 1956. Jerome, comm. Mt. = Commentary on Matthew [the preface is known by its opening lines as Plures fuisse] Émile Bonnard. Jérôme. Commentaire sur saint Matthieu. 2 vols. SC 242, 259. Paris: Éditions du Cerf, 1977 1979. Jerome, nov. op. = Novum opus Fischer, Bonifatius, et al., Biblia sacra iuxta vulgatam versionem. 3rd edn. Stuttgart: Deutsche Bibelgesellschaft, 1983. Pages 1515 16.

Bibliography

323

Jerome, QHG = Liber quaestionum hebraicarum in Genesim C. T. R. Hayward. Saint Jerome’s Hebrew Questions on Genesis. OECS. Oxford: Clarendon Press, 1995. Jerome, vir. = De viris E. Richardson. Liber De viris inlustribus: Gennadius liber De viris inlustribus. TU 14, 1a 1b. Leipzig: J. C. Hinrichs, 1896. ps Jerome, Commentary on Mark Michael Cahill. Expositio Evangelii secundum Marcum. CCSL 82. Turnhout: Brepols, 1997. Michael Cahill. The First Commentary on Mark: An Annotated Translation. Oxford: Oxford University Press, 1998. ps Jerome, Sciendum etiam Donatien De Bruyne, Pierre Maurice Bogaert, and Thomas O’Loughlin. Prefaces to the Latin Bible. Studia Traditionis Theologiae 19. Turnhout: Brepols, 2015. Page 158. Joannes Tzetzes, Arist. = Prolegomena de comoedia Aristophanis W. J. W. Koster. Prolegomena de comoedia. Scholia in Acharnenses, Equites, Nubes. Scholia in Aristophanem 1.1A. Groningen: Bouma, 1975. Martianus Capella, nupt. = De nuptiis Philologiae et Mercurii J. Willis. Martianvs Capella. Leipzig: Teubner, 1983. William Harris Stahl, Richard Johnson, and E. L. Burge. Martianus Capella and the Seven Liberal Arts, Volume II: The Marriage of Philology and Mercury. New York: Columbia University Press, 1977. Marutha of Maypherqat Arthur Vööbus. The Canons Ascribed to Mārūthā of Maipherqaṭ. CSCO 439, Scriptores Syri 191. Louvain: Peeters, 1982. Arthur Vööbus. The Canons Ascribed to Mārūthā of Maipherqaṭ. CSCO 440, Scriptores Syri 192. Louvain: Peeters, 1982. Oscar Braun. De sancta nicaena synod. Syrische Texte des Maruta von Maipherkat nach einer Handschrift der Propaganda zu Rom übersetzt. Münster: H. Schöningh, 1898. Moses bar Kepha, comm. Mt. = Commentary on the Gospel of Matthew J. Rendel Harris. Fragments of the Commentary of Ephrem Syrus upon the Diatessaron. London: C. J. Clay and Sons, 1895. Pages 20 1. M. Kmosko. ‘Analecta Syriaca e codicibus Musei Britannici excerpta’. Oriens Chris tianus 3 (1903): 91 125, at 104 5.

324

Bibliography

Nerses Šnorhali, comm. Mt. = Commentary on the Gospel of Matthew Nerses Šnorhali and Yovhannes Erznkac‘i. Meknut‘iwn Surb Awetaranin or ěst Mat t‘ēosi. Constantinople, 1825. Vigen H. Ghazaryan. ԽՈՐԱՆՆԵՐԻ ՄԵԿՆՈՒԹՅՈՒՆՆԵՐ = Commentaries of Canon Tables. Yerevan: Sargis Khachents, 1995. Thomas F. Mathews and Avedis K. Sanjian. Armenian Gospel Iconography: The Tradition of the Glajor Gospel. Dumbarton Oaks Studies 29. Washington, DC: Dumbarton Oaks Research Library and Collection, 1991. Pages 206 11. [translation James R. Russell] Nerses Šnorhali, Jésus Fils Unique du Père Isaac Kéchichian. Nersès Šnorhali. Jésus Fils Unique du Père. SC 203. Paris: Les Éditions du Cerf, 1973. Origen, Jo. = Commentarii in Joanem Cécile Blanc. Origène: Commentaire sur Saint Jean, Tome I (Livres I V). SC 120. Paris: Les Éditions du Cerf, 1966. Pauc. lib. cath. = Pauca de libris catholicorum scriptorum in evangelia excerpta Robert McNally. Scriptores Hiberniae Minores, Pars I. CCSL 108B. Turnhout: Brepols, 1973. Peter of Celle, On Conscience Hugh Feiss. Peter of Celle: Selected Works. Cistercian Studies Series 100. Kalamazoo, MI: Cistercian Publications, 1987. Philo, qu. Ex. = Questions and Answers on Exodus J. B. Aucher. Philonis Judæi Paralipomena Armena. Venice, 1826. Ralph Marcus. Philo in Ten Volumes. Supplement II: Questions and Answers on Exodus. LCL 401. Cambridge, MA: Harvard University Press, 1987. Philodemus, On the Good King according to Homer T. Dorandi. Filodemo. Il buon re secondo Omero. La scuola di Epicuro 3. Naples: Bibliopolis, 1982. Philostratus, vit. = Vitae sophistarum Wilmer C. Wright. Philostratus: Lives of the Sophists. Eunapius: Lives of the Philo sophers and Sophists. LCL 134. London: Heinemann, 1921. Photius, hom. = Homiliae B. Laourdas. ΦΩΤΙΟΥ ΟΜΙΛΙΑΙ. Thessalonica: 1959.

Bibliography

325

C. Mango. The Homilies of Photius, Patriarch of Constantinople. Dumbarton Oaks Studies 3. Cambridge, MA: Harvard University Press, 1958. Plutarch, Ant. = Antonius Bernadotte Perrin. Plutarch’s Lives, Vol. 9. LCL 101. Cambridge, MA: Harvard University Press, 1920. Plutarch, Num. = Numa Bernadotte Perrin. Plutarch’s Lives, Vol. 1. LCL 46. Cambridge, MA: Harvard University Press, 1914. Plutarch, Sol. = Solon Bernadotte Perrin. Plutarch’s Lives, Vol. 1. LCL 46. Cambridge, MA: Harvard University Press, 1914. Porphyry, Against the Christians Adolf von Harnack. Porphyrius, »Gegen die Christen«, 15 Bücher, Zeugnisse, Fragmente, und Referate. Abhandlungen der königlich preussischen Akademie der Wissenschaften, Philosophisch Historische Klasse 1. Berlin: Verlag der königl. Akademie der Wissenschaften, 1916. Robert M. Berchman. Porphyry Against the Christians. Studies in Platonism, Neopla tonism, and the Platonic Tradition 1. Leiden: Brill, 2005. Matthias Becker. Porphyrios, ‘Contra Christianos’. Neue Sammlung der Fragmente, Testimonien und Dubia. Texte und Kommentare 52. Berlin: De Gruyter, 2016. Porphyry, vit. = Vita Plotini P. Henry and H. R. Schwyzer. Plotini opera, Vol. 1. Museum Lessianum, Series philosophica 33. Leiden: Brill, 1951.

Praefacio secundum Marcum Robert McNally. Scriptores Hiberniae Minores, Pars I. CCSL 108B. Turnhout: Brepols, 1973. Priscillian of Avila, Prisc. prol. = Priscillian Prologues Marco Conti. Priscillian of Avila: The Complete Works. OECT. Oxford: Oxford University Press, 2010. Pages 250 7. Priscillian of Avila, can. Paul. = Canones in Pauli Apostoli Marco Conti. Priscillian of Avila: The Complete Works. OECT. Oxford: Oxford University Press, 2010. Pages 164 209.

326

Bibliography

Ptolemy, alm. = Almagest J. L. Heiberg. Claudii Ptolemaei opera quae exstant omnia, Syntaxis mathematica. Leipzig: Teubner, 1898 1903. G. J. Toomer. Ptolemy’s Almagest. London: Duckworth, 1984. Ptolemy, geo. = Geographia Alfred Stückelberger and Gerd Graßhoff. Klaudios Ptolemaios. Handbuch der Geogra phie. 2 vols. Basel: Schwabe, 2006. Ptolemy, δια. = Προχείρων κανόνων διάταξις καὶ ψηϕοϕορία J. L. Heiberg. Claudii Ptolemaei opera quae exstant omnia, vol. 2 opera astronomica minora. Leipzig: Teubner, 1907. Ptolemy, Προ. καν. = Πρόχειροι Κανόνες A. Tihon. Πτολεμαίου Πρόχειροι Κανόνες. Ptolemy’s Handy Tables, Volume 1a: Tables A1 A2. Louvain: Peeters, 2011. R. Mercier. Πτολεμαίου Πρόχειροι Κανόνες. Ptolemy’s Handy Tables, Volume 1b: Tables A1 A2. Louvain: Peeters, 2011. Quintilian, inst. = Institutio oratoria Donald A. Russell. Quintilian: The Orator’s Education, Books 11 12. LCL 494. Cam bridge, MA: Harvard University Press, 2001. Rabbula of Edessa, can. = Canons Arthur Vööbus. Syriac and Arabic Documents Regarding Legislation Relative to Syrian Asceticism. Papers of the Estonian Theological Society in Exile 11. Stockholm: Etse, 1960. rhet. Her. = Rhetorica ad Herennium Harry Caplan. [Cicero] Rhetorica ad Herennium. LCL 403. Cambridge, MA: Harvard University Press, 1954. Rufinus, HE = Historia Ecclesiastica Eduard Schwartz and Theodor Mommsen. Eusebius Werke. Zweiter Band, Zweiter Teil. GCS 9.2. Leipzig: J. C. Hinrich, 1908. sac. Gel. = Sacramentarium Gelasianum L. C. Mohlberg. Liber Sacramentorum Romanae Aeclesiae Ordinis Anni Circuli. Rerum Ecclesiasticarum Documenta, Series Maior, Fontes IV. Rome: Herder, 1960.

Bibliography

327

Sedulius Scotus, De rectoribus Christianis Edward Doyle. On Christian Rulers and the Poems. Binghamton, NY: Medieval & Renaissance Texts & Studies, 1983. R. W. Dyson. De Rectoribus Christianis. Woodbridge: Boydell & Brewer, 2010. Sedulius Scotus, ex. Eus. = Expositio Eusebii in Decem Canones Mario Esposito. ‘Hiberno Latin Manuscripts in the Libraries of Switzerland. Part I’. Proceedings of the Royal Irish Academy 28 (1910): 62 95, at 83 91. Sedulius Scotus, exp. in praef. Hier. = Explanationes in praefationes sancti Hieronymi ad evangelia PL 103, col. 331 52. Sextus Empiricus, math. = Adversus mathematicos J. Mau and H. Mutschmann. Sexti Empirici Opera, Vol. 3, Adversus mathematicos libros I VI continens. Leipzig: Teubner, 1961. D. L. Blank. Sextus Empiricus: Against the Grammarians (Adversus Mathematicos I). Clarendon Later Ancient Philosophers. Oxford: Clarendon Press, 2007. Step‘anos of Siwnik‘, Commentary on the Four Evangelists Michael B. Papazian. Step‘anos Siwnec‘i: Commentary on the Four Evangelists. New York: SIS Publications, 2014. Step‘anos of Siwnik‘l, pray. or. = How our Prayers are Organized Michael Daniel Findikyan. The Commentary on the Armenian Daily Office By Bishop Step‘anos Siwneci‘ (735): Critical Edition and Translation With Textual and Liturgical Analysis. Orientalia Christiana Analecta 270. Rome: Pontificio Istituto Orientale, 2004. Step‘anos of Siwnik‘, comm. xor. = Commentary on the Xorank‘ Sirarpie Der Nersessian. Armenian Manuscripts in the Freer Gallery of Art. Oriental Studies 6. Washington, DC: Freer Gallery of Art, 1963. Pages 103 4. Vigen H. Ghazaryan. ԽՈՐԱՆՆԵՐԻ ՄԵԿՆՈՒԹՅՈՒՆՆԵՐ = Commentaries of Canon Tables. Yerevan: Sargis Khachents, 1995. Thomas F. Mathews and Avedis K. Sanjian. Armenian Gospel Iconography: The Tradition of the Glajor Gospel. Dumbarton Oaks Studies 29. Washington, DC: Dumbarton Oaks Research Library and Collection, 1991. Pages 206 11. [translation by James R. Russell] Suda A. Adler. Suidae lexicon. 4 vols. Lexicographi Graeci 1.1 1.4. Leipzig: Teubner, 1928 1935.

328

Bibliography

George Syncellus, chron. = Ecloga chronographica A. A. Mosshammer. Georgius Syncellus. Ecloga chronographica. Leipzig: Teubner, 1984. Teaching of Saint Gregory Robert W. Thomson. The Teaching of Saint Gregory: An Early Armenian Catechism. Harvard Armenian Texts and Studies 3. Cambridge, MA: Harvard University Press, 1970. Theophanes the Grammarian, Ἐπιδιαίρεσις ἀκριβὴς καὶ ἄπταιστος τῶν ἐγκειμένων ἀριθμῶν ἐν τοῖς δέκα κανόσι τῶν Δ´ ἁγίων Εὐαγγελίων Alchibiades Sakkelion. ΚΑΤΑΛΟΓΟΣ ΤΩΝ ΧΕΙΡΟΓΡΑΦΩΝ ΤΗΣ ΕΘΝΙΚΗΣ ΒΙΒΛΙΟΘΗΚΗΣ ΤΗΣ ΕΛΛΑΔΟΣ. Athens, 1892. Ms no. 92, p. 17 18. Victor of Capua, praef. = Preface to Codex Fuldensis Ernst Ranke. Codex Fuldensis. Marburg: Sumtibus N.G. Elwerti Bibliopolae Acade mici, 1868. Vrt‘anes K‘ert‘oł, yał. = Yałags Patkeramartic E. Tourian. ‘Vrt’anēaw Kert’ołi, Yałags Patkeramartic’. Sion (1927): 22 5, 61 3. Sirarpie Der Nersessian. ‘Une apologie des images du septième siècle’. Byzantion 17 (1944 1945): 58 87. Secondary Literature Alexander, Philip S., Armin Lange, and Renate Pillinger, eds. In the Second Degree: Paratextual Literature in Ancient Near Eastern and Ancient Mediterranean Culture and Its Reflections in Medieval Literature. Leiden: Brill, 2010. Amirkhanian, Rouzanna. ‘Les tables de canons arméniennes et le thème icono graphique de la Jérusalem céleste’. Revue des Études Arméniennes 31 (2008): 181 232. Amphoux, Christian B. ‘La division du texte grec des Évangiles dans l’Antiquité’. In Titres et articulations du texte dans les oeuvres antiques, ed. Jean Claude Fredouille, Marie Odile Glouet Cazé, Philippe Hoffman, Pierre Petitmengin, and Simone Deléani, 301 12. Paris: Institut d’Études Augustiniennes, 1997. Anderson, Benjamin. Cosmos and Community in Early Medieval Art. New Haven, CT: Yale University Press, 2017. Arevshatyan, Anna. ‘L’identité de Step‘anos Siwnec‘i’. Revue des Études Arméniennes 30 (2005): 401 10. Asmis, Elizabeth. ‘Philodemus’s Poetic Theory and “On the Good King According to Homer” ’. Classical Antiquity 10 (1991): 1 45. Bagnall, Roger S. Early Christian Books in Egypt. Princeton, NJ: Princeton University Press, 2009. Bakhtin, M. M. ‘Discourse in the Novel’. In Dialogic Imagination: Four Essays, ed. Michael Holquist, 259 422. Austin, TX: University of Texas Press, 1981.

Bibliography

329

Bandmann, Günter. ‘Beobachtung zum Etschmiadzin Evangeliar’. In Tortulae: Studien zu altchristlichen und byzantinischen Monumenten, ed. Walter Nikolaus Schumacher, 11 29. Rome: Herder, 1966. Banissi, Ebad, Francis T. Marchese, Camilla Forsell, and Jimmy Johannson, eds. Information Visualisation: Techniques, Usability and Evaluation. Newcastle upon Tyne: Cambridge Scholars Publishing, 2014. Barber, Charles. ‘Writing on the Body: Memory, Desire, and the Holy in Iconoclasm’. In Desire and Denial in Byzantium, ed. Liz James, 111 20. New York: Routledge, 2016. Barbet Massin, Dominique. L’enluminure et le sacré: Irlande et Grande Bretagne VIIe VIIIe siècles. Paris: Presses de l’université Paris Sorbonne, 2013. Bardenhewer, Otto. Geschichte der altkirchlichen Literatur. Zweiter Band: Vom Ende des zweiten Jahrhunderts bis zum Beginn des vierten Jahrhunderts. Freiburg im Breisgau: Herdersche Verlagshandlung, 1914. Barnes, T. D. ‘Some Inconsistencies in Eusebius’. Journal of Theological Studies 35 (1984): 470 5. Barnes, Timothy D. Constantine and Eusebius. Cambridge, MA: Harvard University Press, 1981. Barth, Karl. Church Dogmatics. Translated by G. T. Thomson, G. W. Bromiley, and Harold Knight. London: T&T Clark, 2009. Bausi, Alessandro. ‘A Case for Multiple Text Manuscripts Being “Corpus Organizers” ’. Manuscript Cultures Newsletter 3 (2010): 34 6. Bergmann, Bettina. ‘The Roman House as Memory Theater: The House of the Tragic Poet in Pompeii’. The Art Bulletin 76 (1994): 225 56. Bernabò, Massimo. Il Tetravangelo di Rabbula. Firenze, Biblioteca medicea laurenziana, Plut. 1.56: L’illustrazione del Nuovo Testamento nella Siria del VI secolo. Folia Picta 1. Rome: Edizioni di Storia e Letteratura, 2008. Bernabò, Massimo. ‘The Miniatures in the Rabbula Gospels: Postscripta to a Recent Book’. Dumbarton Oaks Papers 68 (2014): 343 58. Betancourt, Roland. ‘Why Sight is Not Touch: Reconsidering the Tactility of Vision in Byzantium’. Dumbarton Oaks Papers 70 (2016): 1 23. Bischoff, Bernhard. ‘Wendepunkte in der Geschichte der lateinischen Exegese im Frühmittelalter’. Sacris Erudiri 6 (1954): 189 279. Bischoff, Bernhard. ‘Turning Points in the History of Latin Exegesis in the Early Middle Ages’. In Biblical Studies: The Medieval Irish Contribution, ed. Martin McNamara, 74 160. Dublin: Dominican Publications, 1976. [English translation of the previous entry] Black, Matthew. ‘Rabbula of Edessa and the Peshitta’. Bulletin of the John Rylands University Library of Manchester 33 (1951): 203 10. Bloom, Harold. A Map of Misreading. New York: Oxford University Press, 1975. Briquel Chatonnet, François and Muriel Debié, eds. Manuscripta syriaca: des sources de première main. Cahiers d’études syriaques 4. Paris: Geuthner, 2015. Brock, Sebastian. The Bible in the Syriac Tradition. 2nd rev. edn. Piscataway, NJ: Gorgias Press, 2006. Brown, Michelle. ‘Embodying Exegesis: Depicting Evangelists in “Insular” Manu scripts’. In Le Isole britanniche e Rome in éta romanobarbarica, ed. A. M. Luiselli Fadda and É. Ó Carragáin, 109 27. Rome: Herder, 1998. Brown, Michelle. The Lindisfarne Gospels: Society, Spirituality and the Scribe. London: The British Library, 2003.

330

Bibliography

Brubaker, Leslie. Vision and Meaning in Ninth Century Byzantium: Image as Exegesis in the Homilies of Gregory of Nazianzus. Cambridge Studies in Palaeography and Codicology 6. Cambridge: Cambridge University Press, 1999. Brubaker, Leslie. ‘Pictures Are Good to Think With: Looking At Byzantium’. In L’écriture de la mémoire: la littérarité de l’historiographie, ed. Paolo Odorico, Panagiotis A. Agapitos, and Martin Hinterberger, 221 40. Paris: Centre d’études byzantines, néo helléniques et sud est européennes, École des hautes études en sciences sociales, 2006. Brubaker, Leslie. ‘Critical Approaches to Art History’. In Oxford Handbook of Byzan tine Studies, ed. Robin Cormack, John F. Haldon, and Elizabeth Jeffreys, 59 66. Oxford: Oxford University Press, 2008. Bruns, J. Edgar. ‘The “Agreement of Moses and Jesus” in the “Demonstratio Evange lica” of Eusebius’. Vigiliae Christianae 31 (1977): 117 25. Buchthal, Hugo. ‘Studies in Byzantine Illumination of the Thirteenth Century’. Jahrbuch der Berliner Museen 25 (1983): 27 102. Burgess, R. W. ‘The Dates and Editions of Eusebius’ Chronici Canones and Historia Ecclesiastica’. Journal of Theological Studies 48 (1997): 471 504. Burgess, R. W. and Michael Kulikowski. Mosaics of Time: The Latin Chronicle Tradi tions from the First Century BC to the Sixth Century AD. Studies in the Early Middle Ages 33. Turnhout: Brepols, 2013. Burgess, Richard W. and Witold Witakowski. Studies in Eusebian and Post Eusebian Chronography. Historia, Einzelschriften 135. Stuttgart: Franz Steiner, 1999. Burgon, John W. The Last Twelve Verses of the Gospel According to S. Mark Vindicated Against Recent Critical Objectors and Established. London: James Parker and Co., 1871. Burkitt, F. C. The Old Latin and the Itala. Texts and Studies 4. Cambridge: The University Press, 1896. Burkitt, F. C. Evangelion da Mepharreshe. 2 vols. Cambridge: The University Press, 1904. Burkitt, F. C. ‘Saint Augustine’s Bible and the Itala’. Journal of Theological Studies 11 (1910): 258 68, 447. Butler, Shane. ‘Cicero’s Capita’. In The Roman Paratext: Frame, Texts, Readers, ed. Laura Jansen, 73 111. Cambridge: Cambridge University Press, 2014. Cameron, Michael. ‘ “She Arranges All Things Pleasingly” (Wis. 8:1): The Rhetorical Basis of Augustine’s Hermeneutic’. Augustinian Studies 41 (2010): 55 67. Campbell Kelly, M., M. Croarken, R. Flood, and E. Robson. The History of Mathematical Tables: From Sumer to Spreadsheets. Oxford: Oxford University Press, 2003. Caroli, Menico. Il titolo iniziale nel rotolo librario greco egizio: con un catalogo delle testimonianze iconografiche greche e di area vesuviana. Bari: Levante, 2007. Carriker, Andrew. The Library of Eusebius of Caesarea. VCSup 67. Leiden: Brill, 2003. Carruthers, Mary. The Craft of Thought: Meditation, Rhetoric, and the Making of Images, 400 1200. Cambridge Studies in Medieval Literature 34. Cambridge: Cambridge University Press, 1998. Carruthers, Mary. ‘Late Antique Rhetoric, Early Monasticism, and the Revival of School Rhetoric’. In Latin Grammar and Rhetoric: From Classical Theory to Medi eval Practice, ed. Carol Lanham, 239 57. New York: Continuum, 2002. Carruthers, Mary. The Book of Memory: A Study of Memory in Medieval Culture. 2nd edn. Cambridge Studies in Medieval Literature. Cambridge: Cambridge Univer sity Press, 2008.

Bibliography

331

Cataldi Palau, Annaclara. ‘A Little Known Manuscript of the Gospels in “Maiuscola Biblica”: Basil. Gr. A. N. III. 12’. Byzantion 74 (2004): 463 516. Chadwick, Henry. Priscillian of Avila: The Occult and the Charismatic in the Early Church. Oxford: Clarendon Press, 1976. Chapman, John. Notes on the Early History of the Vulgate Gospels. Oxford: Clarendon Press, 1908. Charlesworth, Scott D. Early Christian Gospels: Their Production and Trasmission. Papyrologica Florentina 47. Florence: Edizioni Gonnelli, 2016. Chartier, Roger. Forms and Meanings: Texts, Performances, and Audiences From Codex to Computer. Philadelphia, PA: University of Pennsylvania Press, 1995. Chesnut, Glenn. The First Christian Histories. 2nd edn. Macon: Mercer University Press, 1986. Chin, Catherine M. ‘Cosmos’. In Late Ancient Knowing: Explorations in Intellectual History, ed. Catherine M. Chin and Moulie Vidas, 99 116. Oakland, CA: University of California Press, 2015. Chin, Catherine M. and Moulie Vidas. ‘Introduction: Knowing’. In Late Ancient Knowing: Explorations in Intellectual History, ed. Catherine M. Chin and Moulie Vidas, 1 13. Oakland, CA: University of California Press, 2015. Ciotti, Giovanni and Hang Lin, eds. Tracing Manuscripts in Time and Space Through Paratexts. Studies in Manuscript Cultures 7. Berlin: De Gruyter, 2016. Coogan, Jeremiah. ‘Mapping the Fourfold Gospel: Textual Geography in the Eusebian Apparatus’. Journal of Early Christian Studies 25 (2017): 337 57. Cook, John Granger. The Interpretation of the New Testament in Greco Roman Paganism. STAC 3. Tübingen: Mohr Siebeck, 2000. Coombs, Clayton. A Dual Reception: Eusebius and the Gospel of Mark. Emerging Scholars. Minneapolis, MN: Fortress Press, 2016. Cowe, S. Peter. ‘The Bible in Armenian’. In The New Cambridge History of the Bible. Volume 2: From 600 to 1450, ed. Richard Marsden and E. Ann Matter, 143 61. Cambridge: Cambridge University Press, 2012. Crawford, Matthew R. ‘Diatessaron, a Misnomer? The Evidence From Ephrem’s Commentary’. Early Christianity 4 (2013): 362 85. Crawford, Matthew R. ‘Ammonius of Alexandria, Eusebius of Caesarea and the Origins of Gospels Scholarship’. New Testament Studies 61 (2015): 1 29. Crawford, Matthew R. ‘The Fourfold Gospel in the Writings of Ephrem the Syrian’. Hugoye 18 (2015): 9 51. Crawford, Matthew R. ‘Scholarly Practices: The Eusebian Canon Tables in the Hiberno Latin Tradition’. In Producing Christian Culture: Medieval Exegesis and Its Interpretative Genres, ed. Giles E. M. Gasper, Francis Watson, and Matthew R. Crawford, 65 88. London: Routledge, 2017. Crawford, Matthew R. ‘A New Witness to the “Western” Ordering of the Gospels: GA 073+084’. Journal of Theological Studies 69 (2018): 477 83. Crawford, Matthew R. and Nicholas J. Zola, eds. The Gospel of Tatian: Exploring the Nature and Text of the Diatessaron. The Reception of Jesus in the First Three Centuries. London: T&T Clark, 2019. Crisci, Edoardo. ‘I più antichi codici miscellanei greci: materiali per una riflessione’. In Il codice miscellaneo: Tipologie e funzioni. Atti del Convegno internazionale Cassino

332

Bibliography

14 17 maggio 2003, ed. Edoardo Crisci and Oronzo Pecere, 109 44. Cassino: Università degli studi di Cassino, 2004. Croke, Brian. ‘The Originality of Eusebius’ Chronicle’. American Journal of Philology 103 (1982): 195 200. Croke, Brian. ‘The Origins of the Christian World Chronicle’. In History and Histor ians in Late Antiquity, ed. Brian Croke and Alanna M. Emmett, 116 31. Sydney: Pergamon Press, 1983. Crum, W. E. and H. G. Evelyn White. The Monastery of Epiphanius at Thebes: Part II. New York: Metropolitan Museum of Art, 1926. Davies, Luned Mair. ‘Sedulius Scottus (fl. 840x51 860x74)’. In Oxford Dictionary of National Biography, Oxford: Oxford University Press, 2004. Online. Available at http://www.oxforddnb.com/view/article/50134. De Bruyne, Donatien. ‘Une poésie inconnue d’Aileran le Sage’. Revue Bénédictine 29 (1912): 339 40. De Bruyne, Donatien. Sommaires, divisions et rubriques de la Bible latine. Namur: A. Godenne, 1914. De Bruyne, Donatien. Préfaces de la Bible latine. Namur: A. Godenne, 1920. De Bruyne, Donatien. ‘La préface du Diatessaron Latin avant Victor de Capoue’. Revue Bénédictine 39 (1927): 5 11. De Bruyne, Donatien, Pierre Maurice Bogaert, and Thomas O’Loughlin. Summaries, Divisions and Rubrics of the Latin Bible. Studia Traditionis Theologiae 18. Turnh out: Brepols, 2015. De Bruyne, Donatien, Pierre Maurice Bogaert, and Thomas O’Loughlin. Prefaces to the Latin Bible. Studia Traditionis Theologiae 19. Turnhout: Brepols, 2015. de Jonge, H. J. ‘Augustine on the Interrelations of the Gospels’. In The Four Gospels (Fs Frans Neirynck), vol. 3, ed. F. Van Segbroeck. Louvain: Louvain University Press, Peeters, 1992. Der Nersessian, Sirarpie. Manuscrits arméniens illustrés des XIIe, XIIIe et XIVe siècles de la Bibliothèque des pères mekhitharistes de Venise. Paris: E. de Boccard, 1937. Der Nersessian, Sirarpie. Armenian Manuscripts in the Freer Gallery of Art. Oriental Studies 6. Washington, DC: Freer Gallery of Art, 1963. De Vegvar, Carol Neuman. ‘Remembering Jerusalem: Architecture and Meaning in Insular Canon Table Arcades’. In Making and Meaning in Insular Art, ed. Rachel Moss, 242 56. Dublin: Four Courts, 2006. Dickey, Eleanor. Ancient Greek Scholarship: A Guide to Finding, Reading, and Under standing Scholia, Commentaries, Lexica, and Grammatical Treatises, From Their Beginnings to the Byzantine Period. Oxford: Oxford University Press, 2006. Dickey, Eleanor. ‘Columnar Translation: An Ancient Interpretive Tool That the Romans Gave the Greeks’. Classical Quarterly 65 (2015): 807 21. Digeser, Elizabeth DePalma. A Threat to Public Piety: Christians, Platonists, and the Great Persecution. Ithaca, NY: Cornell University Press, 2012. Dillon, John M. The Middle Platonists, 80 B.C. to A.D. 220. Rev. ed. Ithaca, NY: Cornell University Press, 1996. Döllinger, J. Hippolytus und Kallistus. Regensburg: G. J. Manz, 1853.

Bibliography

333

Dörrie, Heinrich. ‘Ammonios, der Lehrer Plotins’. Hermes 83 (1955): 439 77. Drost Abgaryan, Armenuhi. ‘The Reception of Eusebius of Caesarea (ca. 264 339) in Armenia’. In Greek Texts and Armenian Traditions: An Interdisciplinary Approach, ed. Francesca Gazzano, Lara Pagani, and Giusto Traina, 215 29. Berlin: De Gruyter, 2016. Dungan, David L. ‘Theory of Synopsis Construction’. Biblica 61 (1980): 305 28. Dungan, David L. A History of the Synoptic Problem: The Canon, the Text, the Composition, and the Interpretation of the Gospels. New York: Doubleday, 1999. Dunkle, Brian. ‘Humility, Prophecy, and Augustine’s Harmony of the Gospels’. Augustinian Studies 44 (2013): 207 25. Eco, Umberto. The Infinity of Lists: From Homer to Joyce. Translated by Alistair McEwen. London: MacLehose Press, 2012. Edwards, James R. ‘The Hermeneutical Significance of Chapter Divisions in Ancient Gospel Manuscripts’. New Testament Studies 56 (2010): 413 26. Edwards, Mark. ‘Ammonius, Teacher of Origen’. Journal of Ecclesiastical History 44 (1993): 169 81. Edwards, Mark. ‘One Origen or Two? The Status Quaestionis’. Symbolae Osloenses 89 (2015): 81 103. Edwards, Mark. Origen Against Plato. Ashgate Studies in Philosophy & Theology in Late Antiquity. Aldershot: Ashgate, 2002. Elsner, Jaś. ‘From the Culture of Spolia to the Cult of Relics: The Arch of Constantine and the Genesis of Late Antique Forms’. Papers of the British School at Rome 68 (2000): 149 84. Elsner, Jaś. ‘Late Antique Art: The Problem of the Concept and the Cumulative Aesthet ic’. In Approaching Late Antiquity: The Transformation From Early to Late Empire, ed. Simon Swain and Mark Edwards, 271 309. Oxford: Oxford University Press, 2004. Elsner, Jaś. ‘Introduction’. In Art and Rhetoric in Roman Culture, ed. Jaś Elsner and Michel Meyer, 1 34. Cambridge: Cambridge University Press, 2014. Elsner, Jaś and Michael Squire. ‘Sight and Memory: The Visual Art of Roman Mnemonics’. In Sight and the Ancient Senses, ed. Michael J. Squire, 180 203. London: Routledge, 2016. Evans, Helen C. ‘Canon Tables as an Indication of Teacher Pupil Relationships in the Career of T’oros Roslin’. In Medieval Armenian Culture, ed. Thomas J. Samuelian and Michael E. Stone, 272 90. Chico, CA: Scholars Press, 1984. Evans, Helen C., ed. Armenia: Art, Religion, and Trade in the Middle Ages. New York: The Metropolitan Museum of Art; distributed by Yale University Press, 2018. Falluomini, Carla. The Gothic Version of the Gospels and Pauline Epistles: Cultural Background, Transmission and Character. Berlin: De Gruyter, 2015. Farmer, William R. The Synoptic Problem: A Critical Analysis. New York: Macmillan, 1964. Formisano, Marco. ‘Towards an Aesthetic Paradigm of Late Antiquity’. Antiquité Tardive 15 (2007): 277 84. Forte, Anthony J. ‘Some Philological Observations on Codex Vindobonensis 940’. In The Scriptures and Early Medieval Ireland: Proceedings of the 1993 Conference of the Society for Hiberno Latin Studies on Early Irish Exegesis and Homiletics, ed. Thomas O’Loughlin, 109 15. Turnhout: Brepols, 1999.

334

Bibliography

Fredouille, Jean Claude, Marie Odile Glouet Cazé, Philippe Hoffman, Pierre Petit mengin, and Simone Deléani, eds. Titres et articulations du texte dans les oeuvres antiques. Paris: Institut d’Études Augustiniennes, 1997. Gallagher, Edmon L. and John D. Meade. The Biblical Canon Lists from Early Christianity: Texts and Analysis. Oxford: Oxford University Press, 2017. Gamble, Harry Y. Books and Readers in the Early Church: A History of Early Christian Texts. New Haven, CT: Yale University Press, 1997. Gathercole, Simon J. ‘The Titles of the Gospels in the Earliest New Testament Manu scripts’. Zeitschrift für die Neutestamentliche Wissenschaft 104 (2013): 33 76. Gearhart, Heidi C. ‘From Divine Word to Human Hand: Negotiating Sacred Text in a Medieval Gospel Book’. Word & Image 32 (2016): 430 58. Genette, Gérard. Introduction à l’architexte. Paris: Editions du Seuil, 1979. Genette, Gérard. Palimpsestes: la littérature au second degré. Paris: Editions du Seuil, 1982. Genette, Gérard. Seuils. Paris: Editions du Seuil, 1987. Genette, Gérard. The Architext: An Introduction. Translated by Jane E. Levin. Berkeley, CA: University of California Press, 1992. Genette, Gérard. Palimpsests: Literature in the Second Degree. Translated by Channa Newman and Claude Doubinsky. Lincoln, NE: University of Nebraska Press, 1997. Genette, Gérard. Paratexts: Thresholds of Interpretation. Translated by Jane E. Lewin. Literature, Culture, Theory 20. Cambridge: Cambridge University Press, 1997. Gibson, Roy. ‘Starting With the Index in Pliny’. In The Roman Paratext: Frame, Texts, Readers, ed. Laura Jansen, 33 55. Cambridge: Cambridge University Press, 2014. Gippert, Jost, Wolfgang Schulze, Zaza Aleksidze, and Jean Pierre Mahé, eds. The Caucasian Albanian Palimpsests of Mt. Sinai. 2 vols. Monumenta Palaeographica Medii Aevi. Series Ibero Caucasica. Turnhout: Brepols, 2008. Gleason, Maud W. ‘Afterword’. In Late Ancient Knowing: Explorations in Intellectual History, ed. Catherine M. Chin and Moulie Vidas, 285 91. Oakland, CA: University of California Press, 2015. Goodacre, Mark. ‘Did Thomas Know the Synoptic Gospels? A Response to Denzey Lewis, Kloppenborg and Patterson’. Journal for the Study of the New Testament 36 (2014): 282 93. Goody, Jack. The Domestication of the Savage Mind. Themes in the Social Sciences. Cambridge: Cambridge University Press, 1977. Goody, Jack. The Power of the Written Tradition. Smithsonian Series in Ethnographic Inquiry. Washington, DC: Smithsonian Institution Press, 2000. Gorman, Michael. ‘A Critique of Bischoff ’s Theory of Irish Exegesis: The Commentary on Genesis in Munich Clm 6302 (Wendepunkte 2)’. Journal of Medieval Latin 7 (1997): 178 233. Gorman, Michael. ‘The Myth of Hiberno Latin Exegesis’. Revue Bénédictine 110 (2000): 42 85. Goulet, Richard. ‘Ammonios dit Saccas’. In Dictionnaire des philosophes antiques, vol. 1, ed. Richard Goulet, 165 8. Paris: Éditions du Centre national de la recherche scientifique, 1989. Grafton, Anthony. The Footnote: A Curious History. Rev. ed. Cambridge, MA: Harvard University Press, 1997.

Bibliography

335

Grafton, Anthony and Megan Hale Williams. Christianity and the Transformation of the Book: Origen, Eusebius, and the Library of Caesarea. Cambridge, MA: Belknap Press of Harvard University Press, 2006. Grant, Robert M. Eusebius as Church Historian. Oxford: Clarendon Press, 1980. Gregory, Caspar René. Textkritik des Neuen Testamentes. 3 vols. Leipzig: J. C. Hinrichs, 1900 9. Grigoryan, Gohar. ‘The Roots of Tempietto and Its Symbolism in Armenian Gospels’. Iconographica 13 (2014): 11 24. Grünbart, Michael. ‘Zur Memorialkultur im byzantinischen Mittelalter’. In Byzantine Religious Culture: Studies in Honor of Alice Mary Talbot, ed. Denis Sullivan, Elizabeth Fisher, and Stratis Papaioannou, 373 94. Leiden: Brill, 2012. Gwilliam, G. H. ‘The Ammonian Sections, Eusebian Canons and the Harmonizing Tables in the Syriac Tetraevangelium’. Studia Biblica et Ecclesiastica 2 (1890): 241 72. Haelewyck, Jean Claude. ‘Les vieilles versions syriaques des Évangiles’. In Le Nouveau Testament en syriaque, ed. Jean Claude Haelewyck, 67 113. Paris: Geuthner, 2017. Harnack, Adolf von. Geschichte der altchristlichen Litteratur bis Eusebius, Erster Theil. Leipzig: J. C. Hinrichs, 1893. Harnack, Adolf von. Die Chronologie der altchristlichen Litteratur bis Eusebius. Zweiter Band. Leipzig: J. C. Hinrichs, 1904. Harrison, Carol. ‘ “Not Words But Things”: Harmonious Diversity in the Four Gos pels’. In Augustine: Biblical Exegete, ed. Frederick Van Fleteren and Joseph C. Schnaubelt, 157 73. New York: Peter Lang, 2001. Head, Peter M. ‘Is P⁴, P⁶⁴ and P⁶⁷ the Oldest Manuscript of the Four Gospels? A Response to T. C. Skeat’. New Testament Studies 51 (2005): 450 7. Hebbelynck, Ad. ‘Les κεϕάλαια et les τίτλοι des évangiles’. Le Muséon 41 (1928): 81 120. Heine, Ronald E. Origen: Scholarship in the Service of the Church. Oxford: Oxford University Press, 2010. Helm, Rudolf. Eusebius’ Chronik und ihre Tabellenform. Abhandlungen der preus sischen Akademie der Wissenschaften, Philosophisch historische Klasse 1923, Nr. 4. Berlin: 1924. Herren, Michael. ‘On the Earliest Irish Acquaintance With Isidore of Seville’. In Visigothic Spain: New Approaches, ed. Edward James, 243 50. Oxford: Clarendon Press, 1980. Herren, Michael. ‘Irish Biblical Commentaries Before 800’. In Roma magistra mundi: Itineraria culturae medievalis: Mélanges offerts à Père L.E. Boyle, ed. Jacqueline Hamesse, 391 407. Louvain: Fédération des Instituts d’Etudes Méd iévales, 1998. Higbie, Carolyn. ‘Divide and Edit: A Brief History of Book Divisions’. Harvard Studies in Classical Philology 105 (2010): 1 31. Hollander, A. A. den, Ulrich Schmid, and Willem F. Smelik, eds. Paratext and Megatext as Channels of Jewish and Christian Traditions: The Textual Markers of Contextual ization. Jewish and Christian Perspectives Series 6. Leiden: Brill, 2003. Hollerich, Michael J. ‘Eusebius’. In The New Cambridge History of the Bible. Volume 1: From the Beginnings to 600, ed. James Carleton Paget and Joachim Schaper, 629 52. Cambridge: Cambridge University Press, 2013.

336

Bibliography

Hombert, Pierre Marie. Nouvelles recherches de chronologie augustinienne. Collection des études augustiniennes, Série Antiquité 163. Paris: Institut d’Études Augusti niennes, 2000. Horner, G. The Coptic Version of the New Testament in the Northern Dialect. Oxford: Clarendon Press, 1898. Houghton, H. A. G. Augustine’s Text of John: Patristic Citations and Latin Gospel Manuscripts. OECS. Oxford: Oxford University Press, 2008. Houghton, H. A. G. The Latin New Testament: A Guide to Its Early History, Texts, and Manuscripts. Oxford: Oxford University Press, 2016. Houston, George W. Inside Roman Libraries: Book Collections and Their Management in Antiquity. Chapel Hill, NC: The University of North Carolina Press, 2014. Howlett, David. ‘Seven Studies in Seventh Century Texts’. Peritia 10 (1996): 1 70. Howlett, David. ‘Further Manuscripts of Ailerán’s Canon Euangeliorum’. Peritia 15 (2001): 22 6. Howlett, David. ‘Hiberno Latin Poems on the Eusebian Canons’. Peritia 21 (2010): 162 71. Humfress, Caroline. ‘Judging By the Book: Christian Codices and Late Antique Legal Culture’. In The Early Christian Book, ed. William E. Klingshirn and Linda Safran, 141 58. Washington, DC: The Catholic University of America Press, 2007. Hurley, Sean M. and Laura R. Novick. ‘Solving Problems Using Matrix, Network, and Hierarchy Diagrams: The Consequences of Violating Construction Conventions’. The Quarterly Journal of Experimental Psychology 63 (2010): 275 90. Hurtado, Larry W. The Earliest Christian Artifacts: Manuscripts and Christian Origins. Grand Rapids, MI: Eerdmans, 2006. Inowlocki, Sabrina and Claudio Zamagni, eds. Reconsidering Eusebius: Collected Papers on Literary, Historical, and Theological Issues. VCSup 107. Leiden: Brill, 2011. James, Liz. ‘Art and Lies: Text, Image and Imagination in the Medieval World’. In Icon and Word: The Power of Images in Byzantium, ed. Antony Eastmond, and Liz James 59 72. Burlington, VT: Ashgate, 2003. James, Liz. ‘Senses and Sensibility in Byzantium’. Art History 27 (2004): 45 59. James, Liz, ed. Art and Text in Byzantine Culture. Cambridge: Cambridge University Press, 2007. James, Liz and Ruth Webb. ‘ “To Understand Ultimate Things and Enter Secret Places”: Ekphrasis and Art in Byzantium’. Art History 14 (1991): 1 17. Jansen, Laura. ‘Introduction: Approaches to Roman Paratextuality’. In The Roman Paratext: Frame, Texts, Readers, ed. Laura Jansen, 1 18. Cambridge: Cambridge University Press, 2014. Jansen, Laura, ed. The Roman Paratext: Frame, Texts, Readers. Cambridge: Cambridge University Press, 2014. Johnson, Aaron P. ‘Eusebius’ Praeparatio Evangelica as Literary Experiment’. In Greek Literature in Late Antiquity: Dynamism, Didacticism, Classicism, ed. Scott Fitzgerald Johnson, 67 89. London: Routledge, 2006. Johnson, Aaron P. Eusebius. Understanding Classics. London: I. B. Tauris, 2014. Johnson, Aaron and Jeremy Schott, eds. Eusebius of Caesarea: Tradition and Innov ations. Hellenic Studies 60. Washington, DC: Center for Hellenic Studies, 2013. Johnson, Allan E. ‘Rhetorical Criticism in Eusebius’ Gospel Questions’. Studia Patri stica 18 (1983): 33 9.

Bibliography

337

Johnson, Scott F. ‘Introduction: The Social Presence of Greek in Eastern Christianity, 200 1200 CE’. In Languages and Cultures of Eastern Christianity: Greek, ed. Scott F. Johnson, 1 122. Farnham: Ashgate, 2015. Johnson, Scott F. Literary Territories: Cartographical Thinking in Late Antiquity. Oxford: Oxford University Press, 2016. Johnson, William A. ‘Toward a Sociology of Reading in Classical Antiquity’. The American Journal of Philology 121 (2000): 593 627. Johnson, William A. Bookrolls and Scribes in Oxyrhynchus. Toronto: University of Toronto Press, 2004. Johnson, William A. ‘The Ancient Book’. In The Oxford Handbook of Papyrology, ed. Roger S. Bagnall, 256 81. Oxford: Oxford University Press, 2009. Johnson, William A. Readers and Reading Culture in the High Roman Empire: A Study of Elite Communities. Classical Culture and Society. New York: Oxford University Press, 2010. Jolowicz, Herbert F. and Barry Nicholas. Historical Introduction to the Study of Roman Law. 3rd ed. Cambridge: Cambridge University Press, 1972. Jones, Alexander. Astronomical Papyri from Oxyrhynchus (P. Oxy. 4133 4300a), Volumes I and II. Memoirs of the American Philosophical Society 233. Philadelphia, PA: American Philosophical Society, 1999. Jones, Alexander. ‘Uses and Users of Astronomical Commentaries in Antiquity’. In Commentaries Kommentare, ed. Glenn W. Most, 147 72. Göttingen: Vanden hoeck & Ruprecht, 1999. Jones, Alexander. ‘A Classification of Astronomical Tables on Papyrus’. In Ancient Astronomy and Celestial Divination, ed. N. M. Swerdlow, 299 340. Cambridge, MA: MIT Press, 2000. Jongkind, Dirk. Scribal Habits in Codex Sinaiticus. Texts and Studies, Third Series, Volume 5. Piscataway, NJ: Gorgias Press, 2007. Kalligas, Paul. ‘Traces of Longinus’ Library in Eusebius’ Praeparatio Evangelica’. Classical Quarterly 51 (2001): 584 98. Kalvesmaki, Joel. The Theology of Arithmetic: Number Symbolism in Platonism and Early Christianity. Hellenic Studies 59. Washington, DC: Center for Hellenic Studies, 2013. Kashouh, Hikmat. The Arabic Versions of the Gospels: The Manuscripts and Their Families. Berlin: De Gruyter, 2012. Kelber, Werner H. ‘The History of the Closure of Biblical Texts’. Oral Tradition 25 (2010): 115 40. Kelly, J. F. ‘A Catalogue of Early Medieval Hiberno Latin Biblical Commentaries (I)’. Traditio 44 (1988): 537 71. Kelly, J. F. ‘A Catalogue of Early Medieval Hiberno Latin Biblical Commentaries (II)’. Traditio 45 (1989): 393 434. Kelly, Joseph. ‘Das Bibelwerk: Organisation and Quellenanalyse of the New Testament Section’. In Irland und die Christenheit: Biblestudien und Mission = Ireland and Christendom: The Bible and Missions, ed. Próinséas Ní Chatháin and Michael Richter, 113 23. Stuttgart: Klett Cotta, 1987. Kennedy, George A. A New History of Classical Rhetoric. Princeton, NJ: Princeton University Press, 1994.

338

Bibliography

Keuls, Eva. ‘Rhetoric and Visual Aids in Greece and Rome’. In Communication Arts in the Ancient World, ed. E. Havelock and J. P. Herschbell, 121 34. New York: Hastings House, 1978. Klauser, Theodor. ‘Das Ciborium in der älteren christlichen Buchmalerei’. Nachrichten der Akademie der Wissenschaften in Göttingen 7 (1961): 191 207. Klemm, E. ‘Die Kanontafeln der armenischen Handschriften Cod. 697 im Wiener Mechitaristenkloster’. Zeitschrift für Kunstgeschichte 35 (1972): 69 99. Koester, Craig R. The Dwelling of God: The Tabernacle in the Old Testament, Intertesta mental Jewish Literature, and the New Testament. The Catholic Biblical Quarterly Monograph Series 22. Washington, DC: Catholic Biblical Association of America, 1989. König, Jason and Tim Whitmarsh. ‘Ordering Knowledge’. In Ordering Knowledge in the Roman Empire, ed. Jason König and Tim Whitmarsh, 3 39. Cambridge: Cambridge University Press, 2007. Kouymjian, Dickran. ‘Armenian Iconography: A New Approach’. Journal of the Society for Armenian Studies 6 (1992): 201 22. Kouymjian, Dickran. ‘Armenian Manuscript Illumination in the Formative Period: Text Groups, Eusebian Apparatus, Evangelists’ Portraits’. In Il Caucaso: Cerniera fra Culture dal Mediterraneo alla Persia (secoli IV XI), 1015 49. Settimane di studio del Centro italiano di studi sull’alto Medioevo 43. Spoleto: Centro italiano di studi sull’alto Medioevo, 1996. Krasnodebska D’Aughton, Malgorzata. ‘The Four Symbols Page in Cracow Cathedral Library MS 140: An Image of Unity’. Peritia 12 (2000): 323 41. Krasnodebska D’Aughton, Malgorzata. ‘A Gemmarium for the Recognition of Pre cious Stones in the Cracow Chapter Library, MS 140: A Study on the Unity of Exegetical Themes’. In Listen, O Isle, unto Me: Studies in Medieval Word and Image in Honour of Jennifer O’Reilly, ed. Elizabeth Mullins and Diarmuid Scully, 47 59. Cork: Cork University Press, 2011. Krautheimer, Richard. ‘Introduction to an “Iconography of Medieval Architecture” ’. In his Studies in Early Christian, Medieval, and Renaissance Art, 115 50. New York: New York University Press, 1969. Kynes, Will. An Obituary for ‘Wisdom Literature’: The Birth, Death, and Intertextual Reintegration of a Biblical Corpus. Oxford: Oxford University Press, 2018. Lamb, William R. S. The Catena in Marcum: A Byzantine Anthology of Early Commen tary on Mark. Texts and Editions for New Testament Study 6. Leiden: Brill, 2012. Lang, T. J. and Matthew R. Crawford. ‘The Origins of Pauline Theology: Paratexts and Priscillian of Avila’s Canons on the Letters of the Apostle Paul’. New Testament Studies 63 (2017): 125 45. Langerbeck, H. ‘The Philosophy of Ammonius Saccas and the Connection of Aris totelian and Christian Elements Therein’. The Journal of Hellenic Studies 77 (1957): 67 74. Leatherbury, Sean V. ‘Writing, Reading and Seeing Between the Lines: Framing Late Antique Inscriptions as Texts and Images’. In The Frame in Classical Art: A Cultural History, ed. Verity J. Platt and Michael Squire, 544 81. Cambridge: Cambridge University Press, 2017. Leloir, Louis. Le témoignage d’Éphrem sur le Diatessaron. CSCO 227, Subsidia 19. Louvain: Secrétariat du CorpusSCO, 1962.

Bibliography

339

Leroy, Jules. Les manuscrits syriaques à peintures conservés dans les bibliothèques d’Europe et d’Orient. Bibliothèque archéologique et historique 77, Paris: Librairie orientaliste Paul Geuthner, 1964. Lewis, Agnes Smith. Codex Climaci Rescriptus. Horae Semiticae VIII. Cambridge: The University Press, 1909. Lincicum, David. ‘The Paratextual Invention of the Term “Apostolic Fathers” ’. Journal of Theological Studies 66 (2015): 139 48. Lowden, John. ‘The Word Made Visible: The Exterior of the Early Christian Book as Visual Argument’. In The Early Christian Book, ed. William E. Klingshirn and Linda Safran, 13 47. Washington, DC: The Catholic University of America Press, 2007. McArthur, Harvey K. ‘The Earliest Divisions of the Gospels’. In Studia Evangelica, Vol. III, ed. F. L. Cross, 266 72. Berlin: Akademie Verlag, 1964. McArthur, Harvey K. ‘Eusebian Sections and Canons’. Catholic Biblical Quarterly 27 (1965): 250 6. MacGinty, G. ‘The Pentateuch of the Reference Bible: The Problem Concerning Its Sources’. In The Scriptures and Early Medieval Ireland: Proceedings of the 1993 Conference of the Society for Hiberno Latin Studies on Early Irish Exegesis and Homiletics, ed. Thomas O’Loughlin, 163 77. Turnhout: Brepols, 1999. McGurk, Patrick. ‘The Canon Tables in the Book of Lindisfarne and in the Codex Fuldensis of St. Victor of Capua’. Journal of Theological Studies 6 (1955): 192 8. McGurk, Patrick. Latin Gospel Books: From A.D. 400 to A.D. 800. Paris: Aux Éditions «Erasme», 1961. McGurk, Patrick. ‘The Disposition of the Numbers in Latin Eusebian Canon Tables’. In Philologia Sacra: Biblische und patristische Studien für Hermann J. Frede und Walter Thiele zu ihrem siebzigsten Geburstag, ed. Roger Gryson, 242 58. Freiburg: Herder, 1993. McGurk, Patrick. ‘The Canon Tables in the Book of Kells’. In New Offerings, Ancient Treasures: Studies in Medieval Art for George Henderson, ed. Paul Binski and William Noel, 40 60. Stroud: Sutton, 2001. McKenzie, Judith. The Architecture of Alexandria and Egypt, 300 B.C. to A.D. 700. New Haven, CT: Yale University Press, 2007. McKenzie, Judith and Francis Watson. The Garima Gospels: Early Illuminated Gospel Books From Ethiopia. Oxford: Manar al Athar, University of Oxford, 2016. Mac Lean, Douglas. ‘Scribe as Artist, Not Monk: The Canon Tables of Ailerán “the Wise” and the Book of Kells’. Peritia 17 18 (2003): 433 70. McNally, Robert E. ‘The “Tres Linguae Sacrae” in Early Irish Bible Exegesis’. Theological Studies 19 (1958): 395 403. McNally, Robert E. ‘The Evangelists in the Hiberno Latin Tradition’. In Festschrift Bernhard Bischoff, ed. Johanne Autenrieth and Franz Brunhölzl, 111 22. Stuttgart: Anton Hiersemann, 1971. McNamara, Martin. ‘Plan and Source Analysis of Das Biblewerk, Old Testament’. In Irland und die Christenheit: Biblestudien und Mission = Ireland and Christendom: The Bible and Missions, ed. Próinséas Ní Chatháin and Michael Richter, 84 112. Stuttgart: Klett Cotta, 1987.

340

Bibliography

McNamara, Martin. ‘Celtic Scriptures: Text and Commentaries’. In An Introduction to Celtic Christianity, Edinburgh: T&T Clark, 1989. Madec, Goulven. ‘Le Christ des païens d’après le De consensu evangelistarum de saint Augustin’. Recherches Augustiniennes 26 (1992): 3 67. Magny, Ariane. Porphyry in Fragments: Reception of an Anti Christian Text in Late Antiquity. Burlington, VT: Ashgate, 2014. Maguire, Henry. ‘Art and Text’. In Oxford Handbook of Byzantine Studies, ed. Robin Cormack, John F. Haldon, and Elizabeth Jeffreys, 721 30. Oxford: Oxford Univer sity Press, 2008. Mallan, C. T. ‘The Book Indices in the Manuscripts of Cassius Dio’. The Classical Quarterly 66 (2017): 705 23. Mango, Marlia Mundell. ‘The Production of Syriac Manuscripts, 400 700 AD’. In Scritture, libri e testi nelle aree provinciali di Bisanzio, ed. Guglielmo Cavallo, Giuseppe De Gregorio, and Marilena Maniaci, 161 79. Spoleto: Centro italiano di studi sull’alto Medioevo, 1991. Männlein Robert, Irmgard. Longin, Philologe und Philosoph. Eine Interpretation der erhaltenen Zeugnisse. Beiträge zur Altertumskunde 143. Berlin: De Gruyter, 2011. Manor, T. Scott. Epiphanius’ Alogi and the Johannine Controversy: A Reassessment of Early Ecclesial Opposition to the Johannine Corpus. VCSup 135. Leiden: Brill, 2016. Mansfeld, J. and D. T. Runia. Aëtiana: The Method and Intellectual Context of a Doxographer. Volume 1: The Sources. Philosophia Antiqua 73. Leiden: E. J. Brill, 1997. Marava Chatzinicolaou, Anna and Christina Toufexi Paschou. Catalogue of the Illu minated Byzantine Manuscripts of the National Library of Greece. Volume 2: Manuscripts of New Testament Texts 13th 15th Century. Athens: Publications Bureau of the Academy of Athens, 1985. Marchese, Francis T. ‘Tables and Early Information Visualization’. In Knowledge Visualization Currents: From Text to Art to Culture, ed. Francis T. Marchese and Ebad Banissi. London: Springer, 2013. Marchese, Francis T. ‘The Beginnings of Medieval Information Visualization’. In Information Visualisation: Techniques, Usability and Evaluation, 2 23. Newcastle upon Tyne: Cambridge Scholars Publishing, 2014. Marchese, Francis T. and Ebad Banissi, eds. Knowledge Visualization Currents: From Text to Art to Culture. London: Springer, 2013. Martens, Peter W. Origen and Scripture: The Contours of the Exegetical Life. OECS. Oxford: Oxford University Press, 2012. Martin, Martin J. ‘Origen’s Theory of Language and the First Two Columns of the Hexapla’. Harvard Theological Review 97 (2004): 99 106. Mastro, Gianluca del. Titoli e annotazioni bibliologiche nei papiri greci di Ercolano. Naples: Centro internazionale per lo studio dei papiri ercolanensi, 2014. Mathews, Thomas F. ‘Vrt‘anēs K‘ert‘oł and the Early Theology of Images’. Revue des Études Arméniennes 31 (2008): 101 26. Mathews, Thomas F. and Avedis K. Sanjian. Armenian Gospel Iconography: The Tradition of the Glajor Gospel. Dumbarton Oaks Studies 29. Washington, DC: Dumbarton Oaks Research Library and Collection, 1991.

Bibliography

341

Mercati, Giovanni. Osservazioni a proemi del Salterio di Origene, Ippolito, Eusebio, Cirillo Alessandrino e altri, con frammenti inediti. ST 142. Vatican City: Biblioteca apostolica vaticana, 1948. Merkel, Helmut. Die Widersprüche zwischen den Evangelien. Ihre polemische und apologetische Behandlung in der Alten Kirche bis zu Augustin. WUNT 13. Tübingen: Mohr Siebeck, 1971. Miall, David S. ‘Confounding the Literary: Temporal Problems in Hypertext’. In From Codex to Hypertext: Reading at the Turn of the Twenty First Century, ed. Anouk Lang, 203 16. Amherst, MA: University of Massachusetts Press, 2012. Millar, Fergus. Religion, Language and Community in the Roman Near East: Constan tine to Muhammad. The Schweich Lectures of the British Academy. Oxford: Oxford University Press, 2013. Minard, P. ‘Témoins inédits de la vieille version latine des Évangiles. Les canons à initia des évangéliaires de Sainte Croix de Poitiers et de la Trinité de Vendôme’. Revue Bénédictine 56 (1945): 58 92. Moir, Ian. Codex Climaci Rescriptus Graecus. Texts and Studies 2. Cambridge: Cambridge University Press, 1956. Morlet, Sébastien. ‘Comment le problème du Contra Christianos peut il se poser aujourd’hui?’ In Le traité de Porphyre contre les chrétiens. Un siècle de recherches, nouvelles questions. Actes du colloque international organisé les 8 et 9 septembre 2009 à l’Université de Paris IV Sorbonne, ed. Sébastien Morlet, 11 49. Paris: Institut d’Études Augustiniennes, 2011. Morlet, Sébastien. ‘Eusebius’ Polemic Against Porphyry: A Reassessment’. In Recon sidering Eusebius: Collected Papers on Literary, Historical, and Theological Issues, ed. Sabrina Inowlocki and Claudio Zamagni, 119 50. Leiden: Brill, 2011. Mosshammer, Alden A. The Chronicle of Eusebius and Greek Chronographic Trad ition. Lewisburg, PA: Bucknell University Press, 1979. Moszynski, Leszek. ‘Kanony Euzebiusza w głagolskim rękopisie kodeksu Zografs kiego’. Slovo 25 26 (1976): 77 119. Mullins, Elizabeth. ‘The Insular Reception of the Eusebian Canon Tables: Exegesis and Iconography’. Ph.D. dissertation, University College Cork, 2001. Mullins, Elizabeth. ‘The Eusebian Canon Tables and Hiberno Latin Exegesis: The Case of Vienna, Österreichische Nationalbibliothek, lat. 940’. Sacris Erudiri 53 (2014): 323 44. Nersessian, Vrej. The Bible in the Armenian Tradition. London: The British Library, 2001. Nersessian, Vrej. Treasures From the Ark: 1700 Years of Armenian Christian Art. London: The British Library, 2001. Nestle, E. ‘Die Eusebianische Evangeliensynopse’. Neue kirchliche Zeitschrift 19 (1908): 40 51, 93. Netz, Reviel. ‘Authorial Presence in the Ancient Exact Sciences’. In Writing Science: Medical and Mathematical Authorship in Ancient Greece, ed. Markus Asper, 217 54. Berlin: De Gruyter, 2013. Netzer, Nancy. Cultural Interplay in the Eighth Century: The Trier Gospels and the Making of a Scriptorium At Echternach. Cambridge Studies in Palaeography and Codicology 3. Cambridge: Cambridge University Press, 1994.

342

Bibliography

Netzer, Nancy. ‘The Origin of the Beast Canon Tables Reconsidered’. In The Book of Kells, ed. Felicity O’Mahony, 322 32. Aldershot: Scolar Press for Trinity College Library, Dublin, 1994. Netzer, Nancy. ‘The Design and Decoration of Insular Gospel books and Other Litur gical Manuscripts, c. 600 c. 900’. In The Cambridge History of the Book in Britain, ed. Richard Gameson, 225 43. Cambridge: Cambridge University Press, 2011. Neugebauer, O. and T. C. Skeat. ‘The Astronomical Tables P. Lond. 1278. With a Note on the Paleography of the Fragments’. Osiris 13 (1958): 93 113. Neuschäfer, Bernhard. Origenes als Philologe. Schweizerische Beiträge zur Altertums wissenschaft 18/1 2. Basel: Friedrich Reinhardt Verlag, 1987. Newsom, Carol A. The Book of Job: A Contest of Moral Imaginations. Oxford: Oxford University Press, 2009. Niehoff, Maren. Jewish Exegesis and Homeric Scholarship in Alexandria. Cambridge: Cambridge University Press, 2011. Nordenfalk, Carl. Die spätantiken Kanontafeln. 2 vols. Göteborg: O. Isacsons bok tryckeri a. b., 1938. Nordenfalk, Carl. ‘The Beginning of Book Decoration’. In Beiträge für Georg Swar zenski, ed. Oswald Götz, 9 20. Berlin: Verlag Gebr. Mann, 1951. Nordenfalk, Carl. ‘The Apostolic Canon Tables’. Gazette des Beaux Arts 105 (1963): 17 34. Nordenfalk, Carl. Celtic and Anglo Saxon Painting: Book Illumination in the British Isles, 600 800. New York: George Braziller, 1977. Nordenfalk, Carl. ‘Canon Tables on Papyrus’. Dumbarton Oaks Papers 36 (1982): 29 38. Nordenfalk, Carl. ‘The Eusebian Canon Tables: Some Textual Problems’. Journal of Theological Studies 35 (1984): 96 104. Novick, Laura R. ‘Understanding Spatial Diagram Structure: An Analysis of Hierarch ies, Matrices, and Networks’. The Quarterly Journal of Experimental Psychology 59 (2006): 1826 56. Novick, Laura R. and Sean M. Hurley. ‘To Matrix, Network, or Hierarchy: That is the Question’. Cognitive Psychology 42 (2001): 158 216. Nünlist, René. The Ancient Critic At Work: Terms and Concepts of Literary Criticism in Greek Scholia. Cambridge: Cambridge University Press, 2009. Nyström, Eva. Containing Multitudes: Codex Upsaliensis Graecus 8 in Perspective. Acta Universitatis Upsaliensis, Studia Byzantina Upsaliensia 11. Uppsala: Uppsala Uni versitet, 2009. Ó Cróinín, Dáibhí. ‘Bischoff ’s Wendepunkte Fifty Years on’. Revue Bénédictine 110 (2000): 204 37. O’Donnell, James J. Avatars of the Word: From Papyrus to Cyberspace. Cambridge, MA: Harvard University Press, 1998. O’Loughlin, Thomas, ed. The Scriptures and Early Medieval Ireland: Proceedings of the 1993 Conference of the Society for Hiberno Latin Studies on Early Irish Exegesis and Homiletics. Turnhout: Brepols, 1999. O’Loughlin, Thomas. ‘The Eusebian Apparatus in Some Vulgate Gospel Books’. Peritia 13 (1999): 1 92. O’Loughlin, Thomas. ‘Harmonizing the Truth: Eusebius and the Problem of the Four Gospels’. Traditio 65 (2010): 1 29.

Bibliography

343

O’Loughlin, Thomas. ‘The Eusebian Apparatus in the Lindisfarne Gospels: Ailerán’s Kanon Euangeliorum as a Lens for Its Appreciation’. In The Lindisfarne Gospels: New Perspectives, ed. Richard Gameson, 96 111. Leiden: Brill, 2017. O’Reilly, Jennifer. ‘Exegesis and the Book of Kells: The Lukan Genealogy’. In Scriptural Interpretation in the Fathers: Letter and Spirit, ed. Thomas Finan and Vincent Twomey, 344 97. Dublin: Four Courts Press, 1995. O’Reilly, Jennifer. ‘Patristic and Insular Traditions of the Evangelists: Exegesis and Iconography’. In Le Isole britanniche e Rome in éta romanobarbarica, ed. A. M. Luiselli Fadda and É. Ó Carragáin, 49 94. Rome: Herder, 1998. O’Reilly, Jennifer. ‘St. John the Evangelist: Between Two Worlds’. In Insular Anglo Saxon Art and Thought in the Early Medieval Period, ed. Colum Hourihane, 189 218. Princeton, NJ: Index of Christian Art, Dept. of Art & Archaeology, Princeton University, 2011. Oliver, Harold H. ‘The Epistle of Eusebius to Carpianus: Textual Tradition and Translation’. Novum Testamentum 3 (1959): 138 45. Onians, John. ‘Abstraction and Imagination in Late Antiquity’. Art History 3 (1980): 1 24. Ossendrijver, Mathieu, Christine Proust, Micah Ross, and Nathan Sidoli. ‘Tables in Antiquity’. In History of Numerical Tables, ed. D. Tournès. New York: Springer, forthcoming. Papalexandrou, Amy. ‘Memory Tattered and Torn: Spolia in the Heartland of Byzan tine Hellenism’. In Archaeologies of Memory, ed. Ruth M. Van Dyke and Susan E. Alcock, 56 80. Malden, MA: Blackwell, 2003. Papalexandrou, Amy. ‘The Memory Culture of Byzantium’. In A Companion to Byzantium, ed. Liz James, 108 22. Malden, MA: Wiley Blackwell, 2010. Papazian, Michael. ‘The Commentary on Job of Step‘anos of Siwnik‘ and its Sources’. Le Muséon 130 (2017): 343 63. Papazian, Michael. ‘Origen’s Commentaries as Sources for Step‘anos Siwnec‘i’s Com mentary on the Gospels’. Le Muséon 117 (2004): 507 25. Parker, D. C. An Introduction to the New Testament Manuscripts and Their Texts. Cambridge: Cambridge University Press, 2008. Parker, D. C. Codex Sinaiticus: The Story of the World’s Oldest Bible. London: The British Library, 2010. Parkes, M. B. ‘The Influence of the Concepts of Ordinatio and Compilatio on the Development of the Book’. In Medieval Learning and Literature: Essays Presented to Richard William Hunt, ed. J. J. G. Alexander and M. T. Gibson, 115 41. Oxford: Clarendon Press, 1976. Peabody, David B. ‘Augustine and the Augustinian Hypothesis: A Reexamination of Augustine’s Thought in De consensu evangelistarum’. In New Synoptic Studies: The Cambridge Gospel Conference and Beyond, ed. William R. Farmer, 37 64. Macon, GA: Mercer University Press, 1983. Pelttari, Aaron. ‘Speaking From the Margins: Paratexts in Greek and Latin Poetry’. In Walking the Wire: Latin and Greek Poetry in Dialogue, ed. Tine Scheijnen and Berenice Verhelst. Cambridge: Cambridge University Press, forthcoming. Penna, Angelo. ‘Il «De Consensu Evangelistarum» ed i «Canoni Eusebiani»’. Biblica 36 (1955): 1 19.

344

Bibliography

Petersen, William L. Tatian’s Diatessaron: Its Creation, Dissemination, Significance, and History in Scholarship. VCSup 25. Leiden: E. J. Brill, 1994. Petrucci, A. ‘Dal libro unitario al libro miscellaneo’. In Tradizione dei classici trasfor mazioni della cultura, ed. Andrea Giardina, 173 87. Rome: Laterza, 1986. Pfeiffer, Rudolf. History of Classical Scholarship: From the Beginnings to the End of the Hellenistic Age. Oxford: Clarendon Press, 1968. Platt, Verity J. and Michael Squire, eds. The Frame in Classical Art: A Cultural History. Cambridge: Cambridge University Press, 2017. Puglia, Enzo. ‘Il catalogo di un fondo librario di Ossirinco del III d. C. (PSILaur. inv. 19662)’. Zeitschrift für Papyrologie und Epigraphik 113 (1996): 51 65. Pulliam, Heather. ‘Painting By Numbers: The Art of the Canon Tables’. In The Lindisfarne Gospels: New Perspectives, ed. Richard Gameson, 112 33. Leiden: Brill, 2017. Ramelli, Ilaria. ‘Origen, Patristic Philosophy, and Christian Platonism: Re Thinking the Christianisation of Hellenism’. Vigiliae Christianae 63 (2009): 217 63. Ramelli, Ilaria. ‘Origen the Christian Middle/Neoplatonist: New Arguments for a Possible Identification’. Journal of Early Christian History 1 (2011): 98 130. Rapp, Claudia. ‘Holy Texts, Holy Men, and Holy Scribes: Aspects of Scriptural Holiness in Late Antiquity’. In The Early Christian Book, ed. William E. Klingshirn and Linda Safran, 194 222. Washington, DC: The Catholic University of America Press, 2007. Ratzinger, Joseph. God’s Word: Scripture Tradition Office. Translated by Henry Taylor. Edited by Peter Hünermann and Thomas Söding. San Francisco, CA: Ignatius Press, 2008. Reudenbach, Bruno. ‘Der Codex als heiliger Raum: Überlegungen zur Bildausstattung früher Evangelienbücher’. In Codex und Raum, ed. Stephan Müller, Lieselotte E. Saurma Jeltsch, and Peter Strohschneider, 59 84. Wolfenbüttel: Herzog August Bibliothek, 2009. Reuss, Joseph. ‘Der Presbyter Ammonius von Alexandrien und sein Kommentar zum Johannes Evangelium’. Biblica 44 (1963): 159 70. Riggsby, Andrew M. ‘Guides to the Wor(l)d’. In Ordering Knowledge in the Roman Empire, ed. Jason König and Tim Whitmarsh, 88 107. Cambridge: Cambridge University Press, 2007. Riggsby, Andrew M. ‘Learning the Language of God’. In Modes of Knowing and Ordering Knowledge in Early Christianity, Volume 1, ed. Lewis Ayres, Michael Champion, and Matthew R. Crawford, forthcoming. Riggsby, Andrew M. Mosaics of Knowledge: Representing Information in the Roman World. Classical Culture and Society. Oxford: Oxford University Press, 2019. Robson, Eleanor. ‘Tables and Tabular Formatting in Sumer, Babylonia, and Assyria, 2500 BCE 50 CE’. In The History of Mathematical Tables, ed. M. Campbell Kelly, M. Croarken, R. Flood, and E. Robson, 19 47. Oxford: Oxford University Press, 2003. Robson, Eleanor. ‘Accounting for Change: The Development of Tabular Book Keeping in Early Mesopotamia’. In Creating Economic Order: Record Keeping, Standardization, and the Development of Accounting in the Ancient Near East, ed. Michael Hudson and Cornelia Wunsch, 107 44. Bethesda, MD: CDL Press, 2004.

Bibliography

345

Roby, Courtney. ‘Framing Technologies in Hero and Ptolemy’. In The Frame in Classical Art: A Cultural History, ed. Verity J. Platt and Michael Squire, 514 43. Cambridge: Cambridge University Press, 2017. Rondeau, M. J. Les commentaires patristiques du Psautier (IIIe Ve siècles). Rome: Pont. Institutum Studiorum Orientalium, 1982. Rouveret, Agnès. ‘Les Tables iliaques et l’art de la mémoire’. Bulletin de la Societé Nationale des Antiquaires de France (1988): 166 76. Salzman, Michele Renee. On Roman Time: The Codex Calendar of 354 and the Rhythms of Urban Life in Late Antiquity. The Transformation of the Classical Heritage 17. Berkeley, CA: University of California Press, 1990. Scherbenske, Eric W. Canonizing Paul: Ancient Editorial Practice and the Corpus Paulinum. New York: Oxford University Press, 2013. Schironi, Francesca. To Mega Biblion: Book Ends, End Titles, and Coronides in Papyri with Hexametric Poetry. American Studies in Papyrology 48. Durham, NC: American Society of Papyrologists, 2010. Schissel, O. ‘Die Technik des Bildeinsatzes’. Philologus 72 (1913): 83 114. Schmid, Ulrich. Unum ex quattuor: Eine Geschichte der lateinischen Tatianüberlieferung. Aus der Geschichte der lateinischen Bibel 37. Freiburg im Breisgau: Herder, 2005. Schmid, Ulrich. ‘The Diatessaron of Tatian’. In The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, ed. Bart D. Ehrman and Michael W. Holmes, 2nd edn, 115 42. Leiden: Brill, 2013. Schott, Jeremy. ‘Language’. In Late Ancient Knowing: Explorations in Intellectual History, ed. Catherine M. Chin and Moulie Vidas, 58 79. Oakland, CA: University of California Press, 2015. Schott, Jeremy M. ‘Textuality and Territorialization: Eusebius’ Exegeses of Isaiah and Empire’. In Eusebius of Caesarea: Tradition and Innovations, ed. Aaron Johnson and Jeremy Schott, 169 88. Washington, DC: Center for Hellenic Studies, 2013. Schröder, Bianca Jeanette. Titel und Text: Zur Entwicklung lateinischer Gedichtübers chriften. Mit Untersuchungen zu lateinischen Buchtiteln, Inhaltsverzeichnissen und anderen Gliederungsmitteln. Untersuchungen zur antiken Literatur und Geschichte 54. Berlin: Walter de Gruyter, 1999. Schroeder, Frederic M. ‘Ammonius Saccas’. Aufstieg und Niedergang der römischen Welt 2.36.1 (1987): 493 526. Scrivener, Frederick Henry Ambrose. A Plain Introduction to the Criticism of the New Testament for the Use of Biblical Students. 3rd edn. Cambridge: Deighton, Bell and Co., 1883. Sellew, Philip. ‘Eusebius and the Gospels’. In Eusebius, Christianity, and Judaism, ed. Harold W. Attridge, and Gohei Hata, 110 38. Leiden: E. J. Brill, 1992. Sidoli, Nathan. ‘Mathematical Tables in Ptolemy’s Almagest’. Historia Mathematica 41 (2014): 13 37. Skeat, T. C. ‘The Oldest Manuscript of the Four Gospels?’ New Testament Studies 43 (1997): 1 34. Skeat, T. C. ‘The Codex Sinaiticus, the Codex Vaticanus and Constantine’. Journal of Theological Studies 50 (1999): 583 625. Small, Jocelyn Penny. Wax Tablets of the Mind: Cognitive Studies of Memory and Literacy in Classical Antiquity. New York: Routledge, 1997.

346

Bibliography

Small, Jocelyn Penny. The Parallel Worlds of Classical Art and Text. Cambridge: Cambridge University Press, 2003. Smith, W. Andrew. A Study of the Gospels in Codex Alexandrinus: Codicology, Palaeography, and Scribal Hands. New Testament Tools, Studies and Documents 48. Leiden: Brill, 2014. Smyth, Marina. ‘Isidorian Texts in Seventh Century Ireland’. In Isidore of Seville and His Reception in the Early Middle Ages: Transmitting and Transforming Knowledge, ed. Andrew Fear and Jamie Wood, 111 30. Amsterdam: Amsterdam University Press, 2016. Soden, Hermann von. Die Schriften des Neuen Testaments in ihrer ältesten erreichba ren Textgestalt. Berlin: Verlag von Alexander Duncker, 1902. Spatharakis, I. ‘Some Observations on the Ptolemy Ms. Vat. Gr. 1291: Its Date and the Two Initial Miniatures’. Byzantinische Zeitschrift 71 (1978): 41 9. Squire, Michael. Image and Text in Graeco Roman Antiquity. Cambridge: Cambridge University Press, 2009. Squire, Michael. The Iliad in a Nutshell: Visualizing Epic on the Tabulae Iliacae. Oxford: Oxford University Press, 2011. Squire, Michael. ‘The Ordo of Rhetoric and the Rhetoric of Order’. In Art and Rhetoric in Roman Culture, ed. Jaś Elsner, and Michel Meyer, 353 417. Cambridge: Cambridge University Press, 2014. Squire, Michael. ‘Ecphrasis: Visual and Verbal Interactions in Ancient Greek and Latin Literature’. Oxford Handbooks Online (2015). Available at http://www.oxfordhand books.com/view/10.1093/oxfordhb/9780199935390.001.0001/oxfordhb 97801999353 90 e 58. Accessed 18 January 2018. Stancliffe, Clare. ‘Early “Irish” Biblical Exegesis’. Studia Patristica 12 (1975): 361 70. Stanton, Graham N. ‘The Fourfold Gospel’. New Testament Studies 43 (1997): 317 46. Starr, Raymond J. ‘The Circulation of Literary Texts in the Roman World’. The Classical Quarterly 37 (1987): 213 23. Streeter, B. H. The Four Gospels: A Study of Origins. London: Macmillan, 1924. Strom Olsen, Rolf. ‘The Propylaic Function of the Eusebian Canon Tables in Late Antiquity’. Journal of Early Christian Studies 26 (2018): 403 31. Stroumsa, Guy G. Barbarian Philosophy: The Religious Revolution of Early Christian ity. WUNT 112. Tübingen: Mohr Siebeck, 1999. Stückelberger, Alfred. Bild und Wort: das illustrierte Fachbuch in der antiken Natur wissenschaft, Medizin und Technik. Kulturgeschichte der antiken Welt 62. Mainz am Rhein: P. von Zabern, 1994. Swete, Henry Barclay. An Introduction to the Old Testament in Greek. Cambridge: University Press, 1900. Taylor, David G. K. ‘Répertoire des manuscrits syriaques du Nouveau Testament’. In Le Nouveau Testament en syriaque, ed. Jean Claude Haelewyck, 291 313. Paris: Geuthner, 2017. Thiele, Walter. ‘Beobachtungen zu den eusebianischen Sektionen und Kanones der Evangelien’. Zeitschrift für neutestamentliche Wissenschaft 72 (1981): 100 11. Thomson, R. W. ‘Number Symbolism and Patristic Exegesis in Some Early Armenian Writers’. Handes Amsorya 90 (1976): 117 38. Thomson, R. W. ‘Architectural Symbolism in Classical Armenian Literature’. Journal of Theological Studies 30 (1979): 102 14.

Bibliography

347

Thomson, R. W. ‘Is There an Armenian Tradition of Exegesis?’ Studia Patristica 41 (2006): 97 113. Thomson, Robert W. ‘The Formation of the Armenian Literary Tradition’. In East of Byzantium: Syria and Armenia in the Formative Period, ed. Nina G. Garsoïan, Thomas F. Mathews, and Robert W. Thomson, 135 50. Washington, DC: Dum barton Oaks Center for Byzantine Studies, 1982. Toda, Satoshi. ‘The Eusebian Canons: Their Implications and Potential’. In Early Readers, Scholars, and Editors of the New Testament: Papers From the Eighth Birmingham Colloquium on the Textual Criticism of the New Testament, ed. H. A. G. Houghton, 27 43. Piscataway, NJ: Gorgias Press, 2014. Toom, Tarmo. ‘Was Augustine an Intentionalist? Authorial Intention in Augustine’s Hermeneutics’. Studia Patristica 70 (2013): 185 93. Toom, Tarmo. ‘Augustine’s Hermeneutics: The Science of Divinely Given Signs’. In Patristic Theories of Biblical Interpretation: The Latin Fathers, ed. Tarmo Toom, 77 108. Cambridge: Cambridge University Press, 2016. Turner, James. Philology: The Forgotten Origins of the Modern Humanities. Princeton, NJ: Princeton University Press, 2014. Underwood, Paul A. ‘The Fountain of Life in Manuscripts of the Gospels’. Dumbarton Oaks Papers 5 (1950): 41 138. Vaccari, P. A. ‘Le sezioni evangeliche di Eusebio e il Diatessaron di Taziono nella letteratura siriaca’. Rivista degli Studi Orientali 32 (1957): 433 52. Vandendorpe, Christian. From Papyrus to Hypertext: Toward the Universal Digital Library. Translated by Phyllis Aronoff and Howard Scott. Topics in the Digital Humanities. Urbana, IL: University of Illinois Press, 2009. Vogels, Heinrich Joseph. St. Augustins Schrift De consensu evangelistarum unter vornehmlicher Berücksichtigung ihrer harmonistischen Anschauungen: Eine biblisch patristische Studie. Biblische Studien 13. Freiburg im Breisgau: Herdersche Verlagshandlung, 1908. Vogliano, Achille. Papiri della R. Università di Milano, Vol. 1. Milan: Hoepli, 1937. Vollandt, Ronny. ‘The Conundrum of Scriptural Plurality: The Arabic Bible, Polyglots, and Medieval Predecessors of Biblical Criticism’. In The Text of the Hebrew Bible and Its Editions: Studies in Celebration of the Fifth Centennial of the Complutensian Polyglot, ed. Andrés Piquer Otero and Pablo Torijano Morales, 56 85. Leiden: Brill, 2016. Vööbus, Arthur. Studies in the History of the Gospel Text in Syriac II: New Contribu tions to the Sources Elucidating the History of the Traditions. CSCO 496, Subsidia 79. Louvain: E. Peeters, 1987. Wainer, Howard. ‘Understanding Graphs and Tables’. Educational Researcher 21.1 (1992): 12 23. Wainer, Howard. ‘Improving Tabular Displays, with NAEP Tables as Examples and Inspirations’. Journal of Educational and Behavioral Statistics 22 (1997): 1 30. Wallraff, Martin. Kodex und Kanon: Das Buch im frühen Christentum. Berlin: De Gruyter, 2013. Wallraff, Martin. ‘The Canon Tables of the Psalms: An Unknown Work of Eusebius of Caesarea’. Dumbarton Oaks Papers 67 (2014): 1 14. Wallraff, Martin and Patrick Andrist. ‘Paratexts of the Bible: A New Research Project on Greek Textual Transmission’. Early Christianity 6 (2015): 237 43.

348

Bibliography

Watson, Francis. Gospel Writing: A Canonical Perspective. Grand Rapids, MI: Eerdmans, 2013. Watson, Francis. The Fourfold Gospel: A Theological Reading of the New Testament Portraits of Jesus. Grand Rapids, MI: Baker Academic, 2016. Watson, Francis. ‘Towards a Redaction Critical Reading of the Diatessaron’. Early Christianity 7 (2016): 95 112. Webb, Ruth. ‘The Aesthetics of Sacred Space: Narrative, Metaphor, and Motion in Ekphraseis of Church Buildings’. Dumbarton Oaks Papers 53 (1999): 59 74. Webb, Ruth. Ekphrasis, Imagination and Persuasion in Ancient Rhetorical Theory and Practice. New York: Routledge, 2016. Weitzmann, Kurt. Ancient Book Illumination. Martin Classical Lectures 16. Cambridge, MA: Harvard University Press, 1959. Weitzmann, Kurt. Illustrations in Roll and Codex: A Study of the Origin and Method of Text Illustration. Studies in Manuscript Illumination 2. Princeton, NJ: Princeton University Press, 1970. Wessel, Klaus. ‘Kanontafeln’. In Reallexikon zur byzantinischen Kunst, vol. 3, ed. Klaus Wessel and Marcell Restle, 927 68. Stuttgart: Anton Hiersemann, 1978. Whitby, Mary. ‘Rhetorical Questions’. In A Companion to Byzantium, ed. Liz James, 239 50. Malden, MA: Wiley Blackwell, 2010. Wilken, Robert Louis. The Christians as the Romans Saw Them. Reprint 2nd edn. New Haven, CT: Yale University Press, 2003. Willard, Louis Charles. A Critical Study of the Euthalian Apparatus. Arbeiten zur neutestamentlichen Textforschung 41. Berlin: Walter de Gruyter, 2009. Williams, Peter J. ‘The Syriac Versions of the New Testament’. In The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis, ed. Bart D. Ehrman and Michael W. Holmes, 2nd edn, 143 66. Leiden: Brill, 2013. Wilson, Nigel G. ‘The Relation of Text and Commentary in Greek Books’. In Il Libro e il testo: atti del convegno internazionale: Urbino, 20 23 settembre 1982, ed. Cesare Questa and Renato Raffaelli, 105 10. Urbino: Università degli studi di Urbino, 1984. Wright, Charles D. ‘The “Enumerative Style” in Ireland and Anglo Saxon England’. In The Irish Tradition in Old English Literature, 49 105. Cambridge: Cambridge University Press, 1993. Wright, Charles D. ‘Bischoff ’s Theory of Irish Exegesis and the Genesis Commentary in Munich Clm 6302: A Critique of a Critique’. Journal of Medieval Latin 10 (2000): 115 75. Wright, David H. ‘The Date of the Vatican Illuminated Handy Tables of Ptolemy and of Its Early Additions’. Byzantinische Zeitschrift 78 (1985): 355 62. Wright, W. Catalogue of the Syriac Manuscripts in the British Museum Acquired Since the Year 1838. 3 vols. London: British Museum, 1870. Yates, Frances A. The Art of Memory. London: The Bodley Head, 2014. Yohanna, Samer Soreshow. The Gospel of Mark in the Syriac Harklean Version: An Edition Based Upon the Earliest Witnesses. BO 52. Rome: Gregorian & Biblical Press, 2015. Zahn, Theodor. Tatian’s Diatessaron. Forschungen zur Geschichte des neutes tamentlichen Kanons und der altkirchlichen Literatur, Tl. 1. Erlangen: Deichert, 1881.

Bibliography

349

Zahn, Theodor. Einleitung in das Neue Testament. 3rd edn. Leipzig: Deichert, 1907. Zahn, Theodor. Introduction to the New Testament, Volume II. Translated by Melanc thon Williams Jacobus et al. 3rd edn. Edinburgh: T. & T. Clark, 1909. Zahn, Theodor. ‘Der Exeget Ammonius und andere Ammonii’. Zeitschrift für Kirch engeschichte 38 (1920): 1 22. Zamagni, Claudio. ‘Eusebius’ Exegesis Between Alexandria and Antioch: Being a Scholar in Caesarea (a Test Case From Questions to Stephanos I)’. In Reconsidering Eusebius: Collected Papers on Literary, Historical, and Theological Issues, ed. Sabrina Inowlocki and Claudio Zamagni, 151 76. Leiden: Brill, 2011. Zuntz, Günther. The Ancestry of the Harklean New Testament. British Academy Supplemental Papers 7. London: Oxford University Press, 1945.

Index of Ancient Sources Page numbers with an italic t appended indicate a table. Page numbers in italics indicate illustrations. Biblical Citations Old Testament Genesis 1:1 83 49:9 212 Isaiah 40:22 253 Psalms 103:3 253 New Testament Matthew 1:1 189 1:1 16 112, 220 1:1 26:25 134 1:2 189 1:18 35 2:5 6 184 2:12 13:46 311 2:13 311 2:17 18 312 3:3 218n57 3:4 135 3:7 135 3:11 173, 186 3:11a c 174 3:11d 174 3:17 153 4:1 218n57 4:2 4 179 4:2 10 179 4:5 7 179 4:8 10 179 5:1 183 5:5 215 5:14 188 5:14 16 188 5:15 188 5:16 188 5:41 43a 180 5:42 180 6:11b 190 6:63 311 8:5 10 178 8:13 177 8:18 149 8:19 22 136 9:1 138 9:1 8 176 9:2a 176 9:2b 6a 176 9:6b 7 176

9:8 176 9:27 28a c 312 10:1 42 205 10:24 25a 214 10:33 185 11:29 211 12:14 178 12:14 15 178 12:22 136 13:36 43, 44, 45 312 13:54 58 136 13:57 214, 219 14:3 12 149 14:15 136 14:23 136 14:28 136 14:32 34 188 15:24 214 16:13 16 118, 175 16:16 175 16:17 19 183 16:18 183 16:19 183 16:28 17:9 214 18:8 218n57 18:18 183, 214 19:16 133n28 20:29 137 21:1 311 21:12 114, 137 21:12 13 114 21:22 191, 214 22:1 137 22:34 133n28 22:34 40 133n28 24:45 211 26:2 114, 137 26:6 137 26:26 117, 137, 170, 189 26:35 40 166 168 26:48 50 190 26:75 27:40 311, 312 27:27 29 191 27:27 30 181n44 27:34 133n30, 180, 181 27:35 36 215, 312, 313 27:37 311, 312 27:48 181 27:48 49 133n30, 181

352 Biblical Citations (cont.) 27:50 167t 27:51 167t 27:54 167t, 168n30 27:55 167t 28:8 169 28:9 169 28:18 184 28:18b 184 28:19 20 182 Mark 1:1 21 139 1:3 218n57 1:4 135 1:7 186, 224 1:7 1:8a 174 1:7b 8 173 1:8b 174 1:9 225n75 1:11 153 1:12 13b 218n57 1:21 139, 215 1:22 39 139 1:23 139 1:23 28 139 1:34 139 1:34b 139 1:35 39 139 1:40 3:12 140 1:45 218n57 2:1 138 2:1 12 176 2:3 4 176 2:5 10a 176 2:10b 12a 176 2:12b 176 3:6 178 3:6 7a 178 3:11b 12 140 3:13 5:20 140 4:21 188 4:34b 215 5:18 20 140 5:20 140 5:21 6:30 140 6:1 4 136 6:4 214 6:4 6a 219n59 6:12 13 140 6:17 29 149 6:30 140 6:31 7:37 140 6:34 136 6:35 44 4, 5, 35 6:47 136 6:51 53 188

Index of Ancient Sources 7:36b 37a 140 8:1 9:39 140 8:27 29a 118, 175 8:29b 175, 175n42 8:30 175n42 8:38 185 9:1 9 214 9:37 39 140, 140n47 9:38 40 140n47 9:40 49 140 9:41 50 138 9:43 7 218n57 10:1 12:44 140 10:17 133n28 10:46 137 11:15 114, 137 11:15b 17 114 11:24 191, 214 12:40 44 140 13:1 133n28 14 140 14:1 114, 137 14:1a 114, 115 14:3 137 14:7b 190 14:22 117, 137, 170, 189 14:44 46 190 15:10 12 312 15:16 167t 15:16 19 181n44, 191 15:19 167t 15:21 167t 15:22 167t 15:23 133n30, 167t, 180, 181 15:24 215, 312, 313t 15:26 311 15:36 133n30, 181 16:8 169, 182 16:12 13 186 16:15 182 Luke 1:1 189 1:1 4 189 1:5 189 1:5 5:4 141 1:35 35 3:3 6 218n57 3:7 135 3:16 186, 224 3:16a d 174 3:16b e 173 3:16e 174 3:19 20 149 3:22 153 3:23 38 112 4:1 2a 218n57

Index of Ancient Sources 4:2b 4 179 4:2b 13 179 4:5 8 179 4:9 12 179 4:22 136 4:24 136, 214, 219 4:31 215 4:33 139 4:41 139 5:4 7 141, 187 5:8 10a 215 5:18 138 5:18 19 176 5:18 26 176 5:20 24a 176 5:24b 25 176 5:26 176 6:6 11 178 6:11 178 6:24 25 87 6:26 87 6:30 180 6:40 214 7:1 9 178 7:10 177 7:11 16 87 7:11 17 189 7:17 87 7:36 137 8:16 188 8:37 140 9:12 136 9:12ff. 4, 6, 35 9:18 20 118, 175 9:20b 175 9:26 185 9:27 36 214 9:49 140 9:57 60 136 9:57 62 149 9:61 62 87, 136 10:1 87 10:1 24 205 10:4 140 10:25 133n28 10:25 28 133n28 11:14 136 12:9 185 12:42 211 13:1 5 87 13:1 13 189 13:6 13 87 14:16 137 18:14ff. 11 18:35 137 19:10 214

19:45 114, 137 19:45 46 114 20:47 140 22:1 114, 137 22:3 215 22:19 117, 137, 170, 189 22:32 186 22:32b 119, 187n49, 219 22:47b 48 190 22:48 190 23:4 186, 187 23:4, 14, 22 186 23:10ff. 9 23:13 186, 187 23:22 186, 187 23:34 167t, 312 23:34b 313t 23:34b 35a 215, 313t 23:36 181n44 23:36 37 133n30, 181 23:38 311 23:45 167t 23:53d 187 24:9 169 24:10 187 24:12 187 24:13 35 186 24:41 138, 186 24:41 43 219 John 1:1 5 8, 112, 220 1:1 14 112 1:6 8 112 1:9 10 112, 220 1:12 13 112 1:14 112 1:15 173n40, 186, 224 1:23 218n57 1:26 186 1:26 27 173, 174, 224 1:28 218n57 1:30 186 1:30 31 173n40, 224 1:32 34 174 1:33 174 1:41 42 118, 175 1:42b 183 1:43 51 190 1:49 190 1:49 2:11 175 1:49 51 190 2:1 190 2:1 11 190 2:11 190 2:13 114, 137 2:14 137

353

354

Index of Ancient Sources

Biblical Citations (cont.) 2:14 16 114 3:28 173n40, 186, 224 4:41 137 4:42b 175 4:44 137, 214, 219 4:46b 54 177, 178 4:50 54 177 5:1 138 5:1 4 176 5:1 10 176 5:5 7 176 5:8 9b 176 5:9c 10 176 6 116 117 6:3 183 6:4 114, 137 6:5 136 6:5ff. 4, 6, 35 6:15 136, 168n30 6:31 34 116 6:35 137 6:35 39 170 6:35a 117, 170, 189 6:35b 37 116 6:39 40 116 6:41 137 6:41 42 137 6:43 45 116 6:47 116 6:48 117, 137, 189 6:49 50 116 6:51 117, 137, 189 6:52 54 116, 167t, 168n30 6:53 21:2 311 6:55 117, 137, 189 6:56 61 116 6:63b 64a 116 6:65 67 116 6:68 89 118, 175 6:70 71 116 7:33 87, 190 7:34 49 87, 190 7:40 186 7:42 184 185 7:43 184 185 11:27b 175 11:53 54 178 11:55 137 11:55a 114 12:2 137, 190 12:26 215 13:2 215 13:16 17 214 13:26b 27a 215 14:13 14 167t, 191

14:13 21a 214 14:14 191 14:15 21 191 15:7 214 15:16b 214 15:20a 214 16:23b 24 214 18:38 186 19:4 186 19:4, 6 186 19:5 191 19:7 186 19:23 24a 312, 313t 19:23 24b 215, 313t 19:28 30a 133n30, 180 19:41c 187 20:2 187 20:2 3 181n44 20:6 10 187 20:21 184 20:21b 184 20:23 183, 214 21:1 6 141, 187 21:9 138 21:11 141, 187 21:12 138 21:13 138, 219 21:15 18 120 21:15c 119, 120, 187n49, 219 21:16c 119, 120, 187n49, 219 21:17c 119, 187n49, 219 Acts 1:7 108 2 Corinthians 3:2 3 278 Hebrews 10:16 278 Classical, Late Antique, and Medieval Texts ‘Abdisho‘ bar Brikha, Catalogue of Syriac Books 3 163n25 Aristotle, De Memoria et reminiscentia 452a 271n129 Athenaeus, Deipnosophistae 1.8 66n27 14.56 66n26 Augustine of Hippo De consensu evangelistarum 1.1.1 1.1.2 150n74 1.1.1 23 129 1.2.3 129n19 1.2.4 135, 142n50, 146, 147 2.5.16 132nn26 27 2.6.19 135n32 2.12.25 26 135n32 2.12.27 149n70, 152n86 2.12.28 150n76, 152n86 2.14.31 153nn91 92

Index of Ancient Sources 2.17.35 139 2.21.51 151n81 2.21.52 151n82 2.22.53 54 149n71 2.23.54 136n34 2.37.84 136n35 2.42.89 136n36 2.44.92 149n72 2.45.94 138n42 2.46.95 136n33 2.46.97 152n87 2.47.99 100 136n33 2.65.125 126 137n39 2.66.128 153n90, 293n19 2.67.129 137n40 2.71.139 137n40 2.73.141 142 133n28 2.78.152 137n38 2.78.153 2.79.154 137n40 3.1.1 134n31 3.1.2 137n37 3.2.8 152n89 3.4.14 152n86 3.8.35 150n75 3.13.48 150nn79 80 3.17.54 133n30 4.1.1 4.7.8 135 4.1.2 139 4.2.3 139 4.3.4 140 4.4.5 140nn45 46 4.5.6 140 4.6.7 138, 140 4.7.8 140, 141n48 4.8.9 4.9.10 135 4.9.10 141, 141n49 4.10.11 135, 146n59, 147 12.18.27 133n29 De doctrina christiana 1.26.27 133n29 Epistulae 24 130n23 47.3 130n23 71 130 In Iohannis evangelium tractatus 112.1 145n55 Retractiones 2.16 126n3 Boethius, De institutione arithmetica 2.48 66n24 Cassiodorus, Institutiones divinarum et saecularium litterarum 1.7.2 197n7 Cicero De oratoria 2.353 354 272n132 2.357 271n125 2.358 270nn119, 123 Diodorus Siculus, Bibliotheca historica 17.115.1 66n26

355

Dionysius bar Salibi, Commentarii in evangelia pref. 44 93n112 Epiphanius De mensuris et ponderibus 510 535 67n32 Panarion 51.22.1 113n53 64.3.5 7 67n32 Eunapius, Vitae philosophorum et sophistarum 4.1.2 4.1.6 72n46 Eusebius Chronicle (Aucher text) p.4 108n37 p.5 108n38 p.9 77n63 De laudibus Constantini 6.3 4 94n114 6.5 93n109, 95n115 6.14 93n109 9.15 248n50 Historia Ecclesiastica 1.1.6 75n57, 77n67 1.7.1 17 112n50 3.24.7 14 110n41 3.24.13 113n52 3.39.14 16 155n95 4.11.2 83n88 4.26.5 83n87 4.29.6 64n18, 69n37 4.39.6 69n37 5.27.1 83n87 6.2.7 83n88 6.14.7 155n96 6.16.4 64n19 6.19.1 10 60n7 6.19.10 59n5 6.19.12 14 71n43 6.32.3 79n70 6.43.14 83n87 7.32.6 50n101 7.32.14 19 50n104 7.32.20 50n103 10.1.3 93n110 Praeparatio evangelica 1.10.50 84n89 1.10.53 84n89 8.10.4 84n89 10.9.11 75n57 13.13.4 84n89 13.13.53 84n89 14.18.31 84n89 14.19.9 84n89 Quaestiones evangelicae ad Marinum 1.1 3 182n45

356

Index of Ancient Sources

Classical, Late Antique, and Medieval Texts (cont.) Questioniones evangelicae ad Stephanum 1.9 113n52 4 112n50 Supplementa ad quaestiones ad Stephanum 3 81n76 Vita Constantini 3.34 241n30 4.33 241n30 4.36.2 286n3 4.45 241n30 Euthalian Apparatus, prol. Paul 81n76 Fortunatianus, rhet. 3.14 148n66 Gelasian Sacramentary, sac. Gel. XXXIIII, 305 212n40 Germanus, Historia mystica ecclesiae catholicae 4 97n4 4 5 248n50 Herodotus, Histories 2.116 30 Homer Iliad 1.9.4 31 1.10.4 31 6.289 292 30 14 31 Odyssey 4.227 230 30 4.351 352 30 Isidore of Seville, Etymologiae 1.2.4 205n23 6.15 197n8 6.16.1 217n56 Jerome Chronicle, praef. 77n63 Commentary on Matthew, praef. 210n36, 211n39 De viris 55 59n6 81 125n1 Joannes Tzetzes, Prolegomena de comoedia Aristophanis 2 68n33 Martianus Capella, De nuptiis Philologiae et Mercurii 5.539 148n67 Moses bar Kepha, Commentary on the Gospel of Matthew intro. §33 193n53 Nerses Šnorhali, Commentary on the Gospel of Matthew, prol. 2 282n172 3 256n87, 271n128 4 253n75, 256n88 5 258nn94 96 5 6 259n100 6 253n79, 260n102 7 260n103

8 260n104, 263n110, 265n112 10 260n105 12 262n107 13 262n108 14 260, 263n109, 264n111, 265n112 15 255n83, 260, 261n106, 263n109 16 260 17 256n89, 258n98, 259n99, 277n157 Origen, Commentarii in Joannem 10.31 33 119n59 10.112 115 114n54 10.119 209 114n57 Pauca de libris catholicorum scriptorum in evangelia excerpta §21 23 214n47 Philo, Questions and Answers on Exodus 2.83 253n77 Philodemus, On the Good King according to Homer, col. 43 84n90 Philostratus, Vitae sophistarum 2.27 61n11 Photius, Homiliae 17.5 274n146, 278n159 Plutarch, Lives Antonius 58 68n33 Numa 8.8 66n26 Solon 27.1 75n58 Porphyry, Vita Plotini 14, 19 20 71n45 20, 49 52 71n44 20, 49 51 61n11 Ps Dionysius, De ecclesiastica hierarchia 1.2 257n91 2.3.2 257n92 Ps Iamblichus, Theologoumena arithmetica 4 93n108 10 94n113 Ptolemy Almagest II.9, III.1, IV.3 45n81 III, III.1, III.5, IX.2, IX.3, XI 45n82 III.1 46n86, 51n106 V.7 45n80 Geographia 2.1.3 64n19 Quintilian, Institutio oratoria 3.6.66 42n68 3.6.70 71 42n68 11.2.1 148n65 11.2.18 271n126 11.2.21 270n120 11.2.22 270n124 11.2.28 272n133 Rabbula of Edessa, Canons 43 158n7 Rhetorica ad Herennium 3.16.29 270nn118, 121 3.17.30 271n130 3.18.30 272n134

Index of Ancient Sources 3.18.31 271n131 3.22.36 270n122 3.23.38 39 273n137 3.24.39 148n64 Rufinus, Historia Ecclesiastica 6.16.4 67n31 Sedulius Scottus Explanationes in praefationes sancti Hieronymi ad evangelia 12 223n69 13 14 223n70 15 16 223n71 16 18 224n72 20 21 224n73 23 226n79 25 26 225n75

357

Expositio Eusebii in Decem Canones 2 3 225n76 8 225n77 10 225n78 Sextus Empiricus, Adversus mathematicos 1.27.1 84n90 1.77 66n24 Step‘anos of Siwnik‘, Commentary on the Xorank‘ 1 255n85 2 260n103, 263n110 3 260nn104 105 5 262n107, 265n112 6 261n106, 262n108, 264n111

Index of Manuscripts and Papyri Manuscripts are listed by location, with names in parentheses. Page numbers with an italic t appended indicate a table. Page numbers in italics indicate illustrations. Manuscripts Athens, National Library of Greece, no.92, GA number 1410, fourteenth c. 314 Augsburg, Universitätsbibliothek, Cod. I.2.4°.2 (Augsburg Gospels), early eighth c. 200, 201, 206 Basel, Universitätsbibliothek, Codex Basilensis A.N.III.12 (Ee), ninth c. 115, 169, 170, 182n46 Brescia, Biblioteca civica Queriniana, ManoscrittoPurpureo (Codex Brixianus), sixth c. 36, 37 Cambridge, England Corpus Christi College, Parker Library MS 286 (St. Augustine Gospels), sixth c. 7, 9 University Library, Horae Semiticae VIII (Codex Climaci Rescriptus or CCR), ninth c. 310 313, 313t Cambridge, MA, Harvard University, Houghton Library, MS Georgian 1 (Bert’ay Gospels), tenth c. 13 Cheltenham, Cheltenham Ladies’ College, Dep. e 175 (z. Z. London, Brit. Libr., Loan ms 100/2; NT minuscule 717) 164n28 Dublin, Trinity College Library MS 57 (Book of Durrow), late seventh c. 231, 232 MS 58 (Book of Kells), c.800 201, 202, 207n27, 209n34, 244 Erevan, Matenadaran MS 1681 277n155 MS 2374 (Etchmiadzin Gospels) 989 240n19, 242, 242n35, 254, 255, 261 266 MS 6897 277n155 Ethiopia, Abba Garima Monastery AG I (Abba Garima I), sixth seventh c. 14, 15, 229n4, 238, 240, 241n31, 244n41 AG III (Abba Garima III), fifth seventh c. 14, 229n4, 234, 240, 244n42, 246n47 Florence, Biblioteca Laurenziana, MS Plut. 1.56 (Rabbula Gospels), sixth c. 7, 88, 160, 166, 181n44, 189, 244, 246, 247

Fulda, Germany, Landesbibliothek Bonifatianus 1 (Codex Fuldensis), sixth c. 7, 65n23, 196, 286n3 Jerusalem, Armenian Patriarchate MS 2555 (Second Etchmiadzin Gospels) 240n19 Laon, Bibliothèque municipale, 473 bis, ninth c. 36n56 London, British Library Add MS 43725 (Codex Sinaiticus), mid fourth c. 7, 8 Cotton MS Nero D IV (Lindisfarne Gospels), early eighth c. 7, 116, 120, 207n27, 233, 235 Egerton Papyrus 2 (Egerton Gospel), end of second c. 99 MS Add 17,274, 193n53 MS Add 5111/1 (London Canon Tables), sixth seventh c. 25 27, 26, 244 MS Royal 1. D. V VIII (Codex Alexandrinus), fifth c. 7 Los Angeles, J. Paul Getty Museum, MS Ludwig II 3 (Gospels of Helmarshausen), twelfth c. 277n156 Los Angeles, UCLA, Charles E. Young Research Library, Library Special Collections, Armenian MS 1 (Gladzor Gospels), c. 1300 12, 58, 265, 267, 276n152 Melbourne, National Gallery of Victoria, Felton Bequest, 1960 (710 5) (Gospel Book of Theophanes), second quarter of twelfth c. 4 6, 118, 236 Mestia, Svaneti Museum, Georgian National Museum, Adishi Gospels, 897 13, 239, 240, 241, 246 Mount Athos Dionysiou 4, thirteenth c. 246n47 Vatopedi 949 (1582 INTF), tenth c. 182n46 Munich, Bayerische Staatsbibliothek Clm 14276 + 14277 216n52 Clm 6235, middle of ninth c. 213 Naples, Biblioteca Nazionale “Vittorio Emmanuele III,” Cod. Neapol. ex Vind. 3 (NT minuscule 108) 164n28

Index of Manuscripts and Papyri Oxford, Bodleian Library Clarke 10, 246n47 MS Auct. D.4.1 78 MS Auct. T. inf. 1. 10 (Codex Ebnerianus), twelfth c. 246n47 Paris, Bibliothèque de l’Arsenal, MS 8407 222 Paris, Bibliothèque nationale de France MS Greek 9 (Codex Ephraemi Rescriptus), fifth c. 7 MS lat. 8850 (Gospels of St. Médard de Soissons), before 827 238, 244, 245 MS lat. 11561 (Irish Reference Bible), middle of ninth c. 213n43, 216, 217 MS NAL 1203 (Godescalc Gospel Lectionary), 781 783 243 244 MS syr. 33 (Peshitta Gospels), sixth c. 7, 10, 244, 246n45 Parma, Biblioteca Palatina, MS gr. 5, latter half of eleventh c. 89, 246n47 Poitiers, Médiathèque François Mitterrand, MS 17 (65) (Livre d’Evangiles de l’abbaye Sainte Croix), eighth c. 36n56, 196n4, 208 St. Gall, Stiftbibliothek 1395 (Codex Sangallensis 1395), first half of fifth c. 168, 169 St. Petersburg, National Library of Russia, MS Глаг 1 (Codex Zographensis), late tenth to early eleventh c. 14n14 Tbilisi, Georgia, Georgian National Center of Manuscripts, A 1335 (Vani Gospels), late twelfth to early thirteenth c. 246n47

359

Uppsala, Uppsala University Library, MS DG 1 (Codex Argenteus), sixth c. 9, 11, 171 Vatican, Biblioteca Apostolica Vat. Copt. 9 14n14 Vat. gr. 1209 (Codex Vaticanus), fourth c. 286n3 Vat. gr. 1291 46, 47 Vat MS Barberini lat. 2154 237 Vat. Reg. lat. 76, 216n52 Venice, Biblioteca Nazionale Marciana MS gr. I 8, ninth c. 239, 240 MS gr. Z 10 (394), fourteenth c. 182n46 Vienna, Österreichische Nationalbibliothek, Theol. Gr. 154 (NT minuscule 77) 164n28 Washington, D.C., Freer Art Gallery, MS 50.3 249 Papyri NT Papyri P⁴ 69n36 P⁴⁵ 69n36 P⁶⁴ 69n36 P⁶⁷ 69n36 P⁷⁵ 69n36 P. Cairo 65445 233 234 P. Egerton 2 99 P. Lond. 1278 46 P. Mil. Vogl. 1 19 31 P. Mon. Epiph. 584 35, 36, 47 P. Oxy. 4167 4171 46, 47, 48 P. Oxy. 4168 47 P. Oxy. 4169 48

Index of Modern Authors Amirkhanian, Rouzanna 242 Anderson, Benjamin 46n87, 52n108

Edwards, Mark 61, 62, 71, 73n51 Elsner, Jaś 273, 276n151, 282n171

Bakhtin, Mikhail 96, 109 Bandmann, Gunter 242n35, 246 Barnes, Timothy D. 30n28, 82 Barney, Stephen A. 197n8 Barth, Karl 293n20 Bedrossian, Matthias 259n99 Berchman, Robert M. 106 107n31, 126n3 Bergmann, Bettina 273 Bernabò, Massimo 7n10, 246n45 Bischoff, Bernhard 197n10, 198, 216, 218 Bloom, Harold 103n22 Brock, Sebastian 159, 169 Brown, Michelle P. 209n34 Brubaker, Leslie 280n166 Buchthal, Hugo 246n47 Burgess, Richard W. 53n112, 78n68, 92n103, 107 Burgon, John W. 81 82 Burkitt, F. C. 126, 129, 184n48 Butler, Shane 24 25n16

Farmer, William 146 Findikyan, Michael Daniel 249 250 Formisano, Marco 287

Cahill, Michael 198n12 Cameron, Michael 151n83 Carruthers, Mary J. 148n68, 149, 150n78, 231, 276, 278 281 Certeau, Michel de 100n12 Chesnut, Glenn 110n44 Chin, Catherine M. 288, 289n13 Condorcet, Marquis de 53n110 Coogan, Jeremiah 100nn12 13, 101 Cook, John Granger 107n31 de Bruyne, Donatien 25, 200n18 de Certeau, Michel 100n12 de Jonge, H. J. 146 147 de Vegvar, Carol Neuman 241n29, 242 Der Nersessian, Sirarpie 249 Dickey, Eleanor 57n3, 66, 73 Digeser, Elizabeth DePalma 61, 72n48 Dillon, John 62n12, 72n46 Dungan, David Laird 84 85nn91 92, 100n13, 106, 110n42, 125, 127n7, 128, 146n58, 153n91 Dunkle, Brian 144n53

Gamble, Harry Y. 49n95 Gearhart, Heidi G. 204n21, 277n156 Genette, Gérard 21 25, 27, 28, 101 102, 103n21, 229n3 Ghazaryan, Vigen 251 Gleason, Maud W. 289n12 Goodacre, Mark 99n11 Goody, Jack 292n17 Goulet, Richard 60n8 Grafton, Anthony 64n20, 67n30, 70, 73, 74, 91 92, 101, 107, 108, 109, 110, 285, 286n3, 292 Gregory, Caspar Renè 164n28, 169 Griesbach, Johann Jakob 102, 146 Grigoryan, Gohar 241n31, 242n35 Gwilliam, G. H. 161, 162, 181n44 Hale, Megan 64n20 Harnack, Adolf von 65 Harris, J. Rendell 193nn52 53 Harrison, Carol 126n3 Heine, Ronald E. 60n8 Herren, Michael 205n23 Higbie, Carolyn 30 Hollerich, Michael J. 30n28, 80n75, 91n102 Hombert, Pierre Marie 129n21 Houghton, Hugh 128 130 Houston, George W. 28, 97n1 Howlett, David 200n18, 207n27 James, Liz 275n149 Jansen, Laura 24, 28, 288 Johnson, Scott F. 286 287, 288 Johnson, William A. 29, 31n36, 49nn98 99 Joyce, James 101 102 Kelber, Werner H. 90n99, 154n93 Keuls, Eva 279n161, 280n165 Klauser, Theodor 241n31, 243n39 Kmosko, M. 193nn52 53 König, Jason 75, 95n118, 101, 102

Index of Modern Authors Kouymjian, Dickran 254n81, 283n173 Kulikowski, Michael 53n112 Kynes, Will 99n10 Lamb, William S. 90n99 Langerbeck, H. 61n10 Leatherbury, Sean V. 237n15, 242n31 Lewis, Agnes Smith 310, 312 McArthur, Harvey K. 92n104, 114n56 MacGinty, G. 216 McGurk, Patrick 195n2, 196n4 McKenzie, Judith 229n4, 230n7, 233, 240, 243n38, 244n42 McNally, Robert 212n40, 213, 214 Madec, Goulven 126n3 Magny, Ariane 126n3, 144n52 Maguire, Henry 243n38 Mango, Marlia Mundell 156n1 Manor, T. Scott 113n53 Mansfeld, J. 44, 65n21, 72n47 Marchese, Francis T. 34, 38, 168n32 Mathews, Thomas F. 230n7, 250, 254n80, 257, 259n101, 266, 272, 275n148, 277n155 Mercier, R. 53n111 Merkel, Helmut 106n31, 126n3 Miall, David S. 103n19 Moir, Ian 310 312 Morlet, Sébastien 106n28 Mullins, Elizabeth 198n11, 209n32, 211nn37 38, 212nn41 42, 215n49, 218n58, 222n68 Nestle, Eberhard 82 Netz, Reviel 50n100, 290 291 Neuschäfer, Bernhard 70 Nordenfalk, Carl 26 28, 43 44, 46, 49n97, 65n21, 80n73, 92, 93, 94, 113, 195n2, 228 230, 233, 237 242, 244, 254, 286n3, 314 Novick, Laura R. 38 O’Donnell, James J. 101 O’Loughlin, Thomas 3n5, 24n16, 106, 110, 112n49, 128n13, 168n31, 196nn3 4, 197n8, 201n19, 204n21, 210n35 Onians, John 276n152 O’Reilly, Jennifer 113n52, 199n14, 209n34, 210n36 Ossendrijver, Mathieu 43 44n73, 46n86, 192n51 Paffenroth, Kim 136n36, 142 143n50 Parker, D. C. 81n78 Peabody, David B. 128n12, 146n58

361

Penna, Angelo 127 128, 130n23, 131, 132, 137 Petersen, William L. 65n23, 69n37 Pfeiffer, Rudolf 57n3 Pulliam, Heather 207n25 Pusey, P. E. 181n44 Ramelli, Ilaria 61n10, 62n12 Rapp, Claudia 290 Ratzinger, Joseph 104n23 Reudenbach, Bruno 243 Riddle, M. B. 145n55 Riggsby, Andrew M. 29, 34, 35n52, 42, 50n100, 51, 73, 90n101, 95n118, 98, 285 286, 290 Robson, Eleanor 40 41 Roby, Courtney 53n111 Rouveret, Agnès 273 Runia, David T. 44, 65n21, 72n47 Russell, James R. 252, 257, 259n99 Salmond, S. D. F. 145n55 Sanjian, Avedis K. 230n7, 250, 254n80, 257, 259n