The Survey of the Whole of England: Studies of the documentation resulting from the survey conducted in 1086 9781841719092, 9781407320564

Table of contents :
Front Cover
Title Page
Copyright
Dedication
Table of Contents
List of tables
List of figures
Preface
Chapter 1 Introduction
Chapter 2 Seriation of the DB booklets
Chapter 3 Reconstruction of the D volumes
Chapter 4 The surviving portion of the C text – Part I
Chapter 5 The surviving portion of the C text – Part II
Chapter 6 The geld accounts associated with the C text
Chapter 7 Some alien interventions in the C text
Chapter 8 The monks of Ely and the records of the survey – Part I
Chapter 9 The monks of Ely and the records of the survey – Part II
Chapter 10 The conduct of the survey: the fieldwork phase
Chapter 11 The conduct of the survey: the compilation phase
Appendices
Appendix I An outline reconstruction of B-Ca
Appendix II Summaries compiled in the Treasury
Bibliography
Index

Citation preview

BAR  405  2006   FLIGHT  

The Survey of the Whole of England

THE SURVEY OF THE WHOLE OF ENGLAND

Studies of the documentation resulting from the survey conducted in 1086

Colin Flight

BAR British Series 405 B A R

2006

The Survey of the Whole of England Studies of the documentation resulting from the survey conducted in 1086

Colin Flight

BAR British Series 405 2006

Published in 2016 by BAR Publishing, Oxford BAR British Series 405 The Survey of the Whole of England © C Flight and the Publisher 2006 The author's moral rights under the 1988 UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher.

ISBN 9781841719092 paperback ISBN 9781407320564 e-format DOI https://doi.org/10.30861/9781841719092 A catalogue record for this book is available from the British Library BAR Publishing is the trading name of British Archaeological Reports (Oxford) Ltd. British Archaeological Reports was first incorporated in 1974 to publish the BAR Series, International and British. In 1992 Hadrian Books Ltd became part of the BAR group. This volume was originally published by Archaeopress in conjunction with British Archaeological Reports (Oxford) Ltd / Hadrian Books Ltd, the Series principal publisher, in 2006. This present volume is published by BAR Publishing, 2016.

BAR

PUBLISHING BAR titles are available from: BAR Publishing 122 Banbury Rd, Oxford, OX2 7BP, UK E MAIL [email protected] P HONE +44 (0)1865 310431 F AX +44 (0)1865 316916 www.barpublishing.com

For Jennifer

Sed quid sibi dominus meus rex uult in re huiuscemodi? But why doth my lord the king delight in this thing? (2 Samuel 24:3)

Contents List of tables

vi

List of figures

viii

Preface

ix

1

Introduction

1

2

Seriation of the DB booklets

13

3

Reconstruction of the D volumes

25

4

The surviving portion of the C text – Part I

38

5

The surviving portion of the C text – Part II

49

6

The geld accounts associated with the C text

60

7

Some alien interventions in the C text

71

8

The monks of Ely and the records of the survey – Part I

81

9

The monks of Ely and the records of the survey – Part II

94

10

The conduct of the survey: the fieldwork phase

108

11

The conduct of the survey: the compilation phase

125

Appendices I II

An outline reconstruction of B-Ca

146

Summaries compiled in the Treasury

148

Bibliography

151

Index

155

v

List of tables Table 1. Division of DB into booklets.

14

Table 2. A preliminary classification of the DB booklets.

16

Table 3. Seriation of the DB booklets.

22

Table 4. Surviving original records of the survey.

26

Table 5. Estimated sizes of the D booklets.

28

Table 6. The bound volumes of D reconstructed.

30

Table 7. The bound volumes of D put into the sequence which determined the binding sequence for the DB booklets. 33 Table 8. Three versions of a passage from the survey of Kent.

35

Table 9. Published reproductions of sample scripts from Exeter Cathedral Library 3500. 41 Table 10. Simulated compilation of a set of C booklets (three scribes, four B booklets, five C booklets). 47 Table 11. Simulated compilation of a set of C booklets (four scribes, six B booklets, eight C booklets). 48 Table 12. Published identifications of the scribes represented in Exeter Cathedral Library 3500. 50 Table 13. Numbers of stints performed by each scribe in each section of the C text. 52 Table 14. The quires of C restored to the sequence recorded by the foliation of circa 1500. 54 Table 15. Batches of booklets contained in Exeter Cathedral Library 3500. 61 Table 16. Booklets and scribal stints in batches 3–4.

62

Table 17. Published reproductions of sample scripts from the batch 4 booklets. 68 Table 18. Four versions of the text for one Wiltshire hundred.

70

Table 19. Annotations made in C-So by a scribe at work on D-So.

75

Table 20. Three versions of DB-So chapter 4.

78

Table 21. Two versions of DB-So chapter 33.

78

vi

Table 22. Collation of Cambridge, Trinity College O. 2. 41.

83

Table 23. Contents of the cartulary initiated by scribe T1.

84

Table 24. Collation of Cambridge, Trinity College O. 2. 1.

86

Table 25. Collation of BL, Cotton Tib. A. vi, fos. 36–120.

88

Table 26. Reconstructing B-Ca with or without the evidence of V.

90

Table 27. A reconstruction of the program governing the survey.

92

Table 28. Contents of xEl as represented in T.

95

Table 29. Core segments of xEl as represented in T.

97

Table 30. Cambridgeshire hundreds.

99

Table 31. Oscillations in segment xEl-Sk-1.

103

Table 32. Program for generating the summaries in Exeter Cathedral 3500, fos. 527v–8r. 105 Table 33. Program for generating the summaries in Exeter Cathedral 3500, fo. 531r. 106 Table 34. Order of the entries in the schedule of farms (Lit. E 28, fos. 5va–c) compared with that in the description of the archbishop’s manors (fos. 2va–3va). 112 Table 35. Two descriptions of the town of Sandwich.

113

Table 36. Entries misplaced in DB because of headings omitted in C-NnWaStOx. 130 Table 37. Scribal stints in the surviving D booklets, as they were identified by Rumble (1987). 135 Table 38. The index for DB-Dn.

138

Table 39. The first fifty items from the index for D-Sk (281r) put into a DB-like order. 139 Table 40. Two indexes for DB-Yo: the index provided by the DB scribe (298vb) and an imaginary index, modelled on the first one, representing the actual contents of the text. 141 Table 41. Revised seriation of the DB booklets.

vii

142

List of figures Figure 1. Part of booklet DB-YoLi. Transition from T’ra ad .. car’ to T’ra .. car’. 17 Figure 2. Booklets DB-Mx and DB-Bd. Oscillation between Quando recep’: similit’ and et tntd’ quando recep’, followed by a transition to Val’ et ualuit .. sol’. 18 Figure 3. Booklet DB-Sx. Transition from ualeb’ to ualb’.

19

Figure 4. Booklets DB-St and DB-Wa. Disappearance of one formula coinciding with the appearance of another. 19 Figure 5. Booklets DB-Wi and DB-Do. Stabilization of word-order in the formula De hac t’ra s’t in d’nio .. hid˛e, followed by a transition from De hac t’ra to De ea. 20 Figure 6. Booklet DB-So. Transition from tenuit to teneb’.

21

Figure 7. The sequence of DB booklets represented as a tour of the country. 23 Figure 8. Changing perceptions of the documentation resulting from the survey. 27 Figure 9. Order of the entries in xEl-Ex-2 mapped onto the order of hundreds in B-Ex. 100 Figure 10. Order of the entries in xEl-Nk-1 and xEl-Nk-2 mapped onto the order of hundreds in B-Nk. 101 Figure 11. Order of the entries in xEl-Sk-1 and xEl-Sk-2 mapped onto the order of hundreds in B-Sk. 102 Figure 12. The sequence of DB booklets construed as an approximate map of the progress of stage 2 of the survey. 119

viii

Preface This is not the book that I set out to write. About ten years ago, after I had finished correcting the proofs of a book which had kept me busy for some inordinate length of time, I had to make up my mind what I was going to do next. Casting around for some topic which I might have the resources to cope with from a distance (specifically from a base in Clemson, South Carolina), I decided that I could usefully do some work on the evidence relating to the survey of Kent conducted in 1086. With most of this evidence I was already at least superfically acquainted; and I had easy access (in Clemson), or fairly easy access (in Columbia), to the facsimile editions of some of the crucial manuscripts, and to most of the other published material that I should need to see. When I started work, I set myself two objectives, neither of them very ambitious: I wanted to make accurate transcriptions of all the relevant documents; and I hoped to be able to make some progress – beyond that achieved by previous commentators, from Hasted (1797–1801) through to Morgan (1983) and Williams and Martin (1992) – in mapping the evidence onto the actual landscape. Within a few years, I felt that I was far enough advanced to start to think of putting together a book. The introduction that I had in mind was going to include a few pages – no more than that – explaining how the survey of Kent fitted into the larger scheme of which it was part, the enterprise known to contemporaries as the survey of the whole of England. I did not expect these pages to be hard to write. To the extent that I had already had to think about it, I knew that some of the secondary literature was wrong, often wrong to the point of perversity. Nevertheless, as far as I understood things at the time, there seemed to exist a fair measure of consensus, focused on the book by Galbraith (1961). I was not impressed by Galbraith’s treatment of the evidence from Kent, but this, for him, was a matter of minor significance, and I did not think of judging the book on that basis. A summary of Galbraith’s interpretation, to the extent that it seemed to be generally accepted, was all that I intended to write. In drafting these pages, however, I discovered that what I was saying did not make sense. It did not cohere; it did not engage convincingly with the evidence. Reading and rereading what Galbraith had written, not just in this book but also in other publications, I began to see that his interpretation was fundamentally flawed. At that point, I suppose, I might have decided to drop this portion of the introduction, ignore the problem, and deal with the evidence from Kent as if it stood alone. But I never really thought that this was an acceptable option. It seemed clear to me that I should have to go back to the primary sources, look at the evidence for myself, and see what conclusions I could come to. There have been times, I confess, when I have cursed myself for making this decision. One such time was the day when I sent off a cheque to pay for a microfilm copy of the Exeter manuscript. Not that I begrudged the money: I knew that I was condemning myself to a long spell of hard labour, without any guarantee that the labour would show any profit. There have been times, too, when I have cursed the rest of the world. Nothing that I have done could not have been done ix

by someone else – quite possibly done better by someone else – a long time ago. How was it decided that historians should be paid for practising pirouettes, while waiting for me to do the work? But I persevered, despite moments of anger and periods of despondency, taking comfort from the thought that I was, at least, incidentally, gaining some sharper insight into the evidence from Kent. Over the last six years or so, I have thus been engaged on two parallel projects, switching from one to the other from time to time. I have continued working on the survey of Kent, and am hopeful now that a year or two more will bring that project to completion. (Some of my conclusions are anticipated here, especially in the last two chapters.) In the intervals of that, however, I have been working on the larger problem, the survey in its entirety. This book is the culmination of that second project. Having said what I have to say, I do not expect to write anything further on the subject; but comments from interested readers would never not be welcome. (My e-mail address is .) For the most part, I have worked alone. During the years that I spent at the University of Birmingham (more than twenty years, I shudder to think), I had to inure myself to a solitary mode of existence. Over time I learned to appreciate the advantages of isolation; by now I do not think that I could work in any other way. There are, however, numerous debts which I have incurred in the course of writing this book. They are footnoted in the text, but I am glad to acknowledge some of them here as well. Caroline Thorn read draft versions of chapters 1–5 and gave me the benefit of some candid comments. Tessa Webber was kind enough to check my diagnosis of the Exeter manuscript, comparing it in detail with her own notes and saving me from several errors. In Exeter, Peter Thomas gave me access to the manuscript itself; both he and the assistant librarian, Michael Howarth, have been unfailingly helpful. During the last few months, as the book approached completion, David Davison has been a model editor, responding patiently and promptly to query after query. My thanks to each of them, and to all the librarians and archivists whose help I have had to call on, somewhere along the line. The two documents printed in Appendix II, both of them first edited by Ellis in 1816, are republished here by permission of the Dean and Chapter of Exeter and the Master and Fellows of Trinity College, Cambridge, respectively. Sea Point, December 2005

x

Chapter 1 Introduction

The chapters which follow were written over a period of about six years, in exactly the sequence in which they are printed here, except that this introduction was written last. Each chapter was completed before the next was begun; often some interval elapsed before I was sure what the next chapter ought to be about, let alone what it ought to say. At the time when I was writing chapter 2, I had no clear idea what chapter 3 would look like, no idea at all what the final chapter would look like, if I ever got so far. In drawing the schematic map which sums up the conclusions reached in chapter 2 (Fig. 7), it did not occur to me that this map would prove to have the deeper significance which accrues to it later on (Fig. 12). It was a slow, sometimes painful process, which at any moment might have ended in failure; but in the event I managed to keep moving forward, and I think that I have now reached the end.

concerned; but the reader may think it pathological. Then again, I have a deeply-rooted distrust of the power of words, which, though it seems perfectly justified to me, may seem perverse to the reader. Two words in particular – words which anyone who knows the literature will be expecting to find on almost every page – are largely absent here. I do not voluntarily use the word ‘Domesday’ (except when I am speaking of the use to which the records of the survey were put, in the late twelfth century and later), because it seems to me to beg too many questions.3 I do not use the word ‘circuit’, because, in the current discourse, it implies belief in a silly conjecture which has done a great deal of damage. Words of this sort are idols. We need not waste time subverting them; it is enough just to ignore them. On the other hand, I ask the reader to tolerate a certain amount of algebra. It seems to me indispensable to develop some notation which can mean exactly what we want it to mean. Without any further preamble, I will introduce the notation which I have been using for the last few years, and have found to work well enough (though I do not say that it cannot be improved on). For the four versions of the survey text – the versions defined and sequenced by Galbraith (see below) – I use the following symbols:

In revising the text for publication, I have been able to streamline it to some extent. I have consolidated the bibliography; I have cut out some repetitious passages, replacing them with cross-references. But I have not attempted to give the impression that these chapters were all written at a single stretch. By and large I have left them as I wrote them, even though there are passages here and there which I would have worded rather differently had I known at the time what I only discovered later. Occasionally I have allowed a passage to stand which I no longer think to be right, adding a footnote to explain to the reader why I have changed my mind.1 This is, I hope, not mere laziness or self-indulgence. It seems to me that the reader will be able to understand the interpretation more readily if he or she can reenact something of the process of exploration by which it was arrived at.

B = the version represented by a late twelfth-century copy from Ely (BL, Cotton Tib. A. vi, fos. 71–98) C = the version represented by Exeter Cathedral Library 3500, fos. 25–62, 83–494, 530–1 D = the version represented by PRO, E 31/1 DB = the version represented by PRO, E 31/2, fos. 0–372

Though I have tried to deal fairly with the reader, not trying his or her patience further than is necessary, not demanding assent where the evidence does not require it, there are some idiosyncrasies of mine which may cause irritation.2 I recognize this; I cannot help it. For a start, I seem to have an obsession with putting things in the right order. This is, for me, the font of all wisdom, as far as the survey is

The notation is arbitrary (there is no version A), but the reader who wants some mnemonic might consider making use of the following words: breves, codicelli, Descriptio, Descriptio brevis. To identify each county I use a two-letter code, which generally consists of the first and second or first and last letter

1

Additions of one kind or another, most numerous in chapters 2–3, are enclosed in double brackets: I hope that this makes them obvious enough but not excessively conspicuous.

3

The names ‘Great Domesday’ and ‘Little Domesday’ have only recently become fashionable, but are not of recent creation. Through Pollock (1895), I trace them back to a popularizing book by Morgan (1858). (Apart from his name, I know nothing about the author, and I have not seen the book. There is a review of it, however, in Gentleman’s Magazine, 205 (1858), 120–7.)

2

It annoys some people, I know (because they have told me so), that I have taken to spelling medieval names – Willelm, Rotbert, Goisfrid, and so on – in a contemporary manner. But I see no need to apologize for this: it is a matter of simple courtesy to try to spell a person’s name correctly, all the more so if the person is dead.

1

The survey of the whole of England of the name (Table 1).4 Thus I write D-Ex to denote the D text for Essex. By extension, D-Ex can mean the booklet containing this text, and D-ExNkSk can mean the volume containing the D booklets for Essex, Norfolk and Suffolk.

to B. The rule is, in a word, that no explanation should be deeper than it needs to be. In this respect, life has become much harder since Galbraith. For Eyton, for Round, for Maitland, it was easy to jump from DB to B, almost as easy as to jump from D to B. Take any feature of the DB text: if it seemed sufficiently improbable that this feature originated in DB, one could assume straight away that it originated in B. There was no other possibility. But now we know, thanks to Galbraith, that there are other possibilities which have to be considered, and which (if we remember the rule) have to be considered first. Except in those counties for which part of B or part of C survives, we are extrapolating into the unknown, with nothing to get a grip on. We cannot expect it to be an easy matter to identify textual features which originated in B.

Especially in dealing with B, it is useful to have some easy way of warning the reader – or oneself – that what is being said, though true for some copy that survives, is not necessarily true for the original. For this purpose I have got into the habit of using a slash, and again I find that the trick seems to work well enough. Thus B-Ca / V means the B text for Cambridgeshire as it is represented in the copy that I call V (the only copy, as it happens).5 The same notation may be found helpful in dealing with the various derivative texts which survive as copies from the archives of the monasteries which commissioned them. I have deliberately avoided discussing most of these texts; but there are two which I have cited often enough that some algebra seems to be justified, and this is what I have adopted:

With these points in mind, readers should be able to navigate the following chapters without much difficulty. They are, of course, at liberty to read the chapters in any sequence, or to read just some of them and ignore the rest. But I hope that they will read them all, and read them in their proper order. The analytical chapters (2–9) work backwards from the completion of the survey towards its obscure beginnings. The last two chapters (10–11), which aim towards some synthesis, start at the beginning and work forwards, ending with the compilation of DB itself.

xEl = edited excerpts from some version of the survey text for six counties (Ca, Ht, Ex, Nk, Sk, Hu) made for the monks of Ely, surviving as a mid twelfth-century copy (Cambridge, Trinity College O. 2. 41, fos. 92r–143r) xAug = edited excerpts from the B text for Kent made for the monks of Saint Augustine’s, surviving as an early fourteenth-century copy (PRO, E 164/27, fos. 17r–25r)

1 The interpretation summed up in those last two chapters is (to the best of my knowledge) original in some respects; but it would not upset me at all if anyone wished to call it a neo-Galbraithian interpretation. (From time to time I have called it that myself.) In consequence the reader may be surprised to see that Galbraith’s name is seldom mentioned in the following chapters. The reason is that to find the Galbraithian basis on which I have built one has to go back a long way – not just beyond Galbraith’s (1961) book, but also beyond the (1942) article which was his first published statement on the subject. At some very early stage in the evolution of his thinking, Galbraith was on the right track; but he was losing the thread by the time that he completed this article, and had lost it completely by the time that he completed his book.7

Derivative texts like these, surviving only as copies, are difficult material to handle, and historians who have looked at them have generally misunderstood them.6 As far as matters of methodology are concerned, I have mostly preferred to let any relevant issues emerge at the appropriate moment, in the course of the argument; but there is one point which should perhaps be alluded to briefly here. Because we are dealing with several successive versions, we have to make it a rule to work backwards, one step at a time, from the latest towards the earliest. If we find something in DB that seems to need explaining, we start by trying to explain it on the assumption that it originated in DB, through some decision (or some mistake) on the part of the DB scribe. If that fails, we go back to D; if that fails, we go back to C; if that fails, finally, we are allowed to go back

There are two Galbraithian ideas which seem to me to form a solid foundation: this whole book is built upon them. Before Galbraith, no one had been willing to see that there are four versions of the survey text, each distinctly different from each of the others. Galbraith saw that – and had the courage not to flinch away from it. He saw, furthermore, that these four versions can be arranged typologically to

4

In most cases the meaning will be obvious at a glance; only two codes are at much risk of being misread. ‘He’ is Herefordshire (not Hertfordshire, which is ‘Ht’); ‘Be’ is Berkshire (not Bedfordshire, which is ‘Bd’).

5 In principle I see no objection to the use of multiple slashes. Thus one might write B-Ca / V / Hamilton / Round to mean a passage from B-Ca as it is reproduced in the surviving manuscript, as it was edited from this manuscript by Hamilton, as it was quoted from this edition by Round. In practice, however, I seldom use more than one slash, and would be reluctant to use more than two.

7 This is Vivian Hunter Galbraith (1889–1976). The reader who wishes to know more about the man might start with his ‘lapse into autobiography’ (Galbraith 1970); there is a memoir by Southern (1978), who also wrote a shorter piece for the Dictionary of National Biography. (Southern was one of two people – the other was J(ohn) G(oronwy) Edwards – whose help was acknowledged by Galbraith (1942, p. 161).)

6 The word ‘satellite’ is another idol which had better be forgotten. Used casually by Maitland, it took on a life of its own; but it never had and cannot be given any useful meaning.

2

Introduction make a single sequence, which in my notation is B > C > D > DB. Let it be assumed (we can always discard the assumption if it proves to lead nowhere) that a single sequence does indeed exist: then it becomes a simple exercise to piece the sequence together. The B text has to represent one extremity of it: in this version, unlike the other three, the text is organized cadastrally (hundred by hundred, village by village), not feodally (barony by barony). We can be sure straight away that this extremity is the earlier one, i.e. the beginning of the sequence: without going into detail, it is easy to see how something like D or DB could be derived from something like B, impossible to see how the sequence could work in reverse. The DB text must represent the opposite extremity from B, i.e. the end of the sequence: in this version, unlike the other three, some categories of data (such as livestock statistics) are absent. C and D fall into place between these extremities, and it is clear that C is the earlier of the two. In D, as in DB, the text is written out continuously: except occasionally by chance, the beginning of a new chapter does not coincide with the beginning of a new quire. In C, by contrast (to simplify but not to misrepresent the facts), each chapter occupies a separate booklet. Since D and DB have a property in common which C does not possess, the sequence, for these three versions, has to be C > D > DB.

useful progress, as I hope that the following chapters will go to show.8 Without it, Galbraith’s interpretation does not have any purchase on the facts. These ideas of Galbraith’s are so familiar by now that it takes some effort to realize how novel they were. Some sense of Galbraith’s originality is most easily got by comparing his theory with Round’s. For Round (1895), the uniformity assumption covered only the B text: because it could be proved that the survey text originally took this form for Cambridgeshire, he saw no difficulty in assuming that the same was true for every county. The facts had to be written down in some form: what could be more likely than that they were written down – hundred by hundred, village by village – in a manner framed by the conduct of the meeting at which the hundred juries were brought before the commissioners? Here in B one could see the survey in action. His views on the rest of the compilation phase were more diffidently expressed. The C text was and remained a mystery for him: he took the risk of ignoring it. For D and DB he had only a tentative explanation to propose (Round 1895, pp. 140–2). Two attempts had been made, he thought, to produce a feodalized version of the survey text. The first attempt produced the D text, a rearranged but unabridged copy of the B text. But then, after only three counties had been dealt with, that attempt was given up. Sooner or later a second attempt was made; and now the policy was for the text to be abridged as well as rearranged. This is what produced the DB text. (For consistency, the clerks ought to have gone back and dealt with the first three counties again on this new plan, completing the DB text and discarding the aborted D text; but they did not feel obliged to do that.) On this view, in short, both D and DB are directly derived from B; the form in which they exist is the form in which they originally came into existence; the uniformity assumption does not apply.

By constructing this typological sequence, we have arrived at a theory as to how the compilation process worked. (To the extent that this theory makes sense, we have justified the assumption that we made to begin with.) For each county in turn, we start with B, where the entries are cadastrally organized. Extracting from B the entries for each baron in turn, we produce the collection of booklets which constitutes C. Arranging these booklets in a suitable order and turning them into a continuous text, we produce D. Making numerous additional changes – omitting some categories of information, rewriting the entries, altering the format – we end up by producing DB. That is the core of Galbraith’s interpretation.

As Round pointed out, it then became allowable to think that D and DB – especially DB – might be, by some significant margin, later than the survey itself. Enthused by that possibility, some historians began competing to push DB further and further forward in point of time;9 and Douglas (1936, p. 255) summed up this tendency by concluding that it would, at the least, be ‘unwise’ to suppose that DB was completed before 1100. This article of Douglas’s seems to have been the final provocation which stirred Galbraith into action. Nothing enraged him more than any hint of a suggestion that DB was an afterthought.

However plausible we think this looks, it does not hold together unless we are willing to suppose that every version of the text (with the possible exception of DB) was originally more comprehensive than it is now. As things stand, two versions are the most that exist for any county, and two successive versions do not exist anywhere. We may be inclined to assume that the surviving portion of the C text was the source for a lost portion of the D text, or that the surviving portion of the D text was derived from a lost portion of the C text, but we are not in a position to prove it. Given that we have to start making assumptions, there is no virtue in making them piecemeal. To gain as much leverage as possible, we have to be ready to make it our working hypothesis that the compilation process was everywhere the same. Under this hypothesis, what is known to be true for one county can (in the absence of proof to the contrary) be taken to be true for every county. This is what I propose to call the uniformity assumption: it is the second element which I borrow from Galbraith. With it, we can make some

There were weaknesses in Round’s interpretation – most obviously his failure to come to grips with the Exeter manuscript – which Galbraith was able to exploit.10 Es8

But there is, I think, something to be said in its favour a priori. A job as complicated as this could hardly have been done at all unless it was done systematically.

9

The winners were Johnson and Jenkinson (1915), who thought that DB might perhaps be as late as circa 1130.

10

3

There is a footnote of Round’s – ‘It will be observed that I do not touch

The survey of the whole of England sentially what he did was to extend the uniformity assumption so that it covered not only B but also C and D. He had hardly made the suggestion before he started changing his mind; but it seems to me that he was right in the first place. The C text, though by accident it only survives for five counties (and is complete only for one) did originally exist for every county; the D text, though by accident it only survives for three counties, did originally exist for every county.

that is the right interpretation, this evidence does not necessarily mean very much: it might mean that a fair copy was made only of C-Dn (or only of some portion of C-Dn). Even if we took this evidence to mean that a fair copy was made of the entire C text, there would be nothing to prove that this fair copy was the source text used by the compilers of DB. Galbraith, throwing caution to the winds, pounced on this remark.14 Explicit proof that C gave birth to D, not just here but everywhere, was hardly to be expected, but here was one good hint – all the better because its authors had not been trying to prove a theory, just noting a fact which might turn out to be of interest.

To the extent that the C text survives, it can be compared with the DB text for the same counties; and the comparison shows that DB was derived (perhaps immediately, perhaps not immediately) from C. The point had been proved by Baring (1912), and Galbraith contented himself with citing that paper. But Baring himself referred back to an article by Whale (1905), which, despite its chaotic appearance, has the merit of including the first report of one crucial fact. By and large, the order of the entries in each chapter of DB is the same as in the corresponding booklet of C. Nothing can be argued from that, because the same might be true if C and DB were (as Eyton and others had supposed) derived independently from B. But there is one chapter in DB-Dn which seems to be an exception to the rule, and the order here, as Whale (1905, p. 266) discovered, was produced by a temporary transposition of two quires of C. That evidence is conclusive: DB derives from C.11 Given that, what Baring envisaged was a single version of the text, intermediate between B and DB, variably C-like or D-like. The recognition that C and D are typologically so different that they have to be taken to represent two separate stages of the compilation process is, I think, original with Galbraith. In every county (or group of counties), a ‘rough draft’ in the form of C was compiled from B; this was superseded by a ‘fair copy’ in the form of D; and this was the source text used by the compilers of DB.12

I take the discussion of these points no further here, because the rest of this book is an extended commentary on them. The thrust of it is Galbraithian. In a sense it is more Galbraithian than Galbraith, because he, in this sense, was only a Galbraithian at the very beginning. The typological sequence, B > C > D > DB, is his; his too is the willingness to drive the uniformity assumption as far as it will go. These are the points which seem to me to give this article what permanent value it has. Having made them, however, Galbraith started to back away from them. He was in retreat before he had even finished writing this paper; the retreat became a rout later on.

Some of the reasons why things went wrong are obvious. One striking feature of this article is its failure to make use of a crucial piece of evidence – a strictly contemporary account of the survey, written by no less a person than the bishop of Hereford – which had been discovered and published by Stevenson (1907). This evidence proves (as Stevenson saw) that the survey was conducted in two separate stages; and that had never been known or even suspected. Round, writing in 1895, did not have any way of realizing this; nor did Maitland, or Eyton, or Ellis, or anyone else. Since 1907, historians who drew their inspiration from Round had failed to exploit this new evidence; Galbraith had the chance to do better – had the chance but wasted it. In this article, bishop Rotbert is barely even mentioned.15 It was only at a later stage that Galbraith (1950) began try-

Galbraith could cite some evidence which tended to confirm the idea that C was not the immediate source for DB. When facsimiles of two pages from the Exeter manuscript were published by the Palaeographical Society, the editors (or one of the editors) mentioned the existence of two marginal notes elsewhere in the manuscript which appeared to be ‘the memoranda of persons engaged on a fair copy’.13 Even if

are probably the memoranda of persons engaged on a fair copy’ (Bond, Thompson and Warner 1884–94, vol. 1, letterpress to plates 70–1). This remark is quoted by Galbraith in an exaggerated and inaccurate form: ‘The marginalia, the editors say, “are in different hands from those of the text, from which it is evident that they cannot refer to the compilation of the present manuscript, but are probably the memoranda of those engaged on a fair copy” ’ (Galbraith 1942, p. 165, note 2).

the Liber Exoniensis’ (1895, p. 146) – which Galbraith delighted in quoting. 11 Strictly speaking, the conclusion is valid only for this one chapter; how far to go in generalizing from it is another question. 12 Some excerpts survive (in the Ely text which I call xEl) from a version of the survey text for Huntingdonshire more primitive than DB-Hu. Round (1895, p. 135) had left this evidence unexplained; Galbraith (1942, pp. 168–9) saw it as proof of the existence of C-Hu or D-Hu. In fact, it is not decidable where these excerpts came from; on balance they are (in my opinion) more likely to have come from B, rather than from C or D. (In Galbraith’s view of the case, there was a good chance that they might have come from C. In mine there is hardly any chance of that: the only likely sources are B and D.)

14 As far as I can see, he had no warrant for assuming that these ‘marginalia’ – all two of them – were in ‘non-curial’ script (see below), though as a matter of fact they are. (If they were ‘curial’, they would mean something different.) There is, moreover, a price attached to this conclusion (below, note 24) which Galbraith evaded paying. 15 He is referred to by name just once, and only incidentally (Galbraith 1942, p. 175). (Here and later, Galbraith tended to call him ‘Robert of Hereford’, as if he were some local chronicler.) There is also one footnote reference (p. 171), which takes the cryptic form ‘Cf. Select charters (1913), p. 95’: an extract from the text printed by Stevenson (1907) had been included in that edition of Stubbs’s book.

13

‘As both these notes are in different hands from those of the text, it is evident that they cannot refer to the compilation of the present MS., but

4

Introduction ing to fit this evidence into the picture; and he found that there was no room for it. By 1961 he had come to the conclusion that bishop Rotbert was . . . mistaken. There is, we are told, ‘certainly no question of two successive panels of Domesday legati’. Rotbert must have been confusing two different operations. ‘Perhaps the first body of inquisitors were the Domesday commissioners and the second a special Treasury panel sent to collect the arrears of the tax’, i.e. the current geld (1961, pp. 94–5).16 In effect, we are asked to believe that Galbraith understands what was happening better than it was understood by a well-placed contemporary observer. Galbraith was, to a surprising degree, confident of this himself; but I do not see how we can feel the same.

a group of ‘non-curial’ scribes. Who these scribes were, where they came from, exactly how many of them there were, are questions which Galbraith does not seem to have bothered to ask himself; it was enough for him to know that they had not undergone the same ‘curial’ training as the scribes of DB. The interpretation which he developed depends absolutely on this distinction between ‘curial’ and ‘non-curial’ scribes. Second, he gambled on the rightness of a conjecture of Eyton’s. Thinking (as he did at the time) that the entire survey, including the compilation of DB, was completed within three months or so, Eyton had had to find some way of explaining how this might have been achievable; and he came up with the idea that the work was divided among nine ‘Corps of Commissioners’, each of which dealt with a group of neighbouring counties (Eyton 1877, pp. 107–8). Galbraith adopted this suggestion, to the extent that it suited his purpose. There is no discussion: it is, he says, ‘generally agreed that distinct commissions visited separate groups of counties’ (1942, p. 162).

Apart from ignoring this evidence, Galbraith committed two disastrous errors. First, he took it for granted, seemingly without hesitation, that DB was written by a plurality of scribes. As everyone agrees by now, this was a mistake. It was, at the time, an assumption which had never been questioned; but no one had given it as much importance as Galbraith was about to do. DB, it seemed to him, was the product of a very tightly disciplined scriptorium: the scribes employed here had all been trained to write a distinctive style of script. No one had ever been able to decide how many different scribes were involved – but that just went to show how thoroughly the scribes had been trained.17 Within narrow limits, they all used the same sort of script, the same abbreviations, the same technical terms, the same turns of phrase.18 In short, it seemed obvious that DB was produced by a group of government scribes – ‘curial’ scribes, as Galbraith preferred to call them.

Was it ‘generally agreed’? Though Galbraith was much better placed than I am to judge, I am not persuaded that this statement was true at the time. Neither Round nor Maitland had expressed support for the idea.20 So far as they gave it any thought, they would have regarded the suggestion as a guess of Eyton’s; and Eyton’s guesses had almost invariably turned out to be wrong. It is true that the suggestion had been taken up in a popularizing book by Ballard (1906) – this seems to be where Galbraith came across it – but that endorsement can hardly have counted for much.21 The article by Douglas (1936) mentioned above says not one word about Eyton’s suggestion; it does not even mention Eyton’s name.22 Stenton (1943) is silent on the subject.

In any number of respects, C and D are very different from DB. Each is manifestly the work of a group of scribes, some of them quite good, some others barely competent. Not only does the script vary greatly from scribe to scribe: none of it bears much resemblance to the ‘curial’ script exemplified by DB.19 So C and D must each be the work of

And yet, whether or not the statement was true when Galbraith made it, it certainly did become true. As far as I can see, it went entirely unchallenged.23 Without bothering to look at the evidence for himself, without being made to justify the assertion, he was allowed to assume that Eyton’s conjecture was an established fact. The country was divided up into groups of contiguous counties; each group of coun-

16 By 1974, he had changed his mind (without explaining why, without saying that he had done so): now he preferred to reverse the sequence of events. Bishop Rotbert is quoted as saying that there were ‘two separate Inquests in 1086, of which the first was a geld inquest and the second the Domesday survey’ (1974, p. 23). The bishop says no such thing: if that is what really happened, the bishop got it wrong. Yet we are also told, with much more emphasis here than previously, that bishop Rotbert is an excellent witness: it is ‘difficult to imagine a man better fitted to testify’ (Galbraith 1974, pp. 22–3). (This statement is true in a way, but exaggerated to the point of becoming untrue. In fact it is easy to think of a man who would be ‘better fitted’: given the choice, we should have preferred to hear from the bishop of Lincoln – the only bishop who is known for certain to have served on one of the commissions of inquiry.)

of the original C text. 20 In a passing remark (‘It should be added, however, . . . ’), Round (1895, p. 134) mentions it as an alternative possibility that the bishop of Coutances and bishop Walkelin ‘may have been, respectively, the heads of two distinct commissions for adjoining groups of counties’. To cite this in paraphrase as ‘Round’s opinion’ (Galbraith 1942, p. 162) is, to put it politely, a careless piece of wording – not as careless, however, as a passage elsewhere (1948, p. 94), which seems to attribute the entire conjecture to Round.

17

Somewhere in my reading, I met with an anecdote about a nineteenthcentury government department whose clerks all wrote so similarly that only they could tell which of them had written what. Foolishly I failed to make a note of it. Can anyone tell me where to find it again?

21 Tait (1908) reviewed the book at some length without referring to this point. Galbraith at that time was a student in Manchester, and Tait was one of his teachers.

18

And, one might add, as Sawyer (1956) did, they all spelt English placenames in the same way. And, one might also add, they all spelt French words alike.

22

Ballard’s book (in its second edition) is listed in the ‘bibliographical note’ (Douglas 1936, p. 249) but never cited specifically.

19

23 I do not challenge it here; in the absence of any evidence for it, I take no notice of it.

There are, in fact, two or three stretches of text in the Exeter manuscript for which one might make an exception; but they are all additions, not part

5

The survey of the whole of England ties was dealt with by a separate group of commissioners; and – this is where the argument engages with the palaeographical evidence – each group of commissioners was accompanied by a separate group of ‘non-curial’ scribes.24

on two points which are rather obviously wrong: that the B text never came anywhere near the treasury,26 and that the DB text superseded all earlier versions. Having reached this point, Galbraith summed up the argument so far by restating his previous conclusion, B > C > D > DB, but now he had to preface it with an ‘if’ clause: ‘if any uniform system governed the actions of the various commissions’, the conclusion would still be valid (1942, p. 169). That is a fatal concession. If the compilation process was decentralized to the extent that Galbraith is suggesting, what justification can there be for assuming that the procedure was uniform? How can we suppose that uniformity prevailed when we cannot see that any means existed for enforcing it? The obvious agents to use would be the ‘curial’ scribes: let them be trained in the proper procedure and then sent out, one here, one there, to make sure that this procedure is understood and fully complied with by each group of ‘non-curial’ scribes. But the ‘curial’ scribes, in Galbraith’s interpretation, are not permitted to leave Winchester: they sit there, twiddling their thumbs, while B and C and D are being compiled.27 There is nothing whatever for the ‘curial’ scribes to do until portions of the D text start arriving. If anything has gone wrong (if the hundred headings, for example, have been omitted), it is, by now, too late to put it right. The ‘curial’ scribes can only grumble. The ‘non-curial’ scribes, for their part, have no contact with the treasury while they are actually at work: the first and only contact occurs at the moment when they deliver the finished D text. (If someone else makes the delivery, there is no contact at all.) Their job is over and done with before that of the ‘curial’ scribes has even begun.

For Galbraith, furthermore, each version of the survey text (other than DB) existed for only one purpose: to serve as a source for the next version. Once the data had been put into a feodal frame, B was no longer useful. The various portions of the B text were discarded wherever they happened to be at the time; if one portion survived, that was because it was rescued and taken home by the abbot of Ely. Once a fair copy had been made of it, C was no longer useful. The various portions of the C text were discarded wherever they happened to be at the time; if one portion survived, that was because someone in Exeter was able to get hold of it and thought it worth preserving. Officially B and C had both ceased to exist; so far as they survived at all, they were scattered around the country. Only the fair copy was delivered to the treasury. And finally, once DB had been compiled, D was no longer useful: if one portion survived, that was due to some accidental cause.25 This is just speculation, though Galbraith presents it in the sort of peremptory language which suggests that only an idiot will disagree with it. I disagree with most of it. As far as C is concerned, I agree that it was made solely for the purpose of allowing D to be made. (I do not agree, however, that any part of either C or D was made in Exeter.) But D contains a large amount of information which was not transferred into DB. Unless one thinks that this information was never really wanted and had only been collected by mistake – Galbraith was reduced to precisely that absurdity – D does not become redundant as soon as DB exists. On the contrary, D is the full record of the survey, and DB is just an epitome, condensed into one volume. Something similar is true (though I was slow to realize it) with regard to B. There is information in B which was not transferred, via C, into D. Unless one thinks that this information was worthless, it seems to follow that B would also have been worth keeping, at least for a while. There is nothing here which conflicts in any significant way with Galbraith’s interpretation; he had no logical reason for denying any of this. Instead, for reasons of his own, he preferred to insist

Galbraith had a choice to make, and this ‘if’ clause proves that he was aware of the fact. If he continued stressing the decentralized nature of the compilation process, he would have to back away from the uniformity assumption. If he wanted to hold on to this assumption, he would have to find some way of explaining how uniformity might have been maintained, in spite of decentralization. Perhaps more by drift than by conscious decision,28 he eventually made his choice: he preferred the first alternative. That was a miscalculation from which he never recovered. The reception of Galbraith’s article is hard to gauge. It is not to be forgotten, of course, that the article was written and published in the middle of a war – at a time, that

24

But this is not a fair conclusion. The only reason for thinking that certain marginal notes in the Exeter manuscript (above, note 13) are ‘probably the memoranda of persons engaged on a fair copy’ is the fact that they are ‘in different hands from those of the text’. If we think that a ‘fair copy’ was made, we must also think that the commissioners were accompanied by two groups of scribes: the first group was responsible for a ‘rough draft’ (the surviving C text), but this second group took over when the time arrived to make a ‘fair copy’ (the lost D text). At a stroke, we seem to have doubled the number of ‘non-curial’ scribes who would have to be employed. Are we comfortable with that? Do we see any rationale for such a division of labour? Galbraith disposes of these difficulties by ignoring them.

26

The point being that, in that case, the compilation process must have been started immediately, wherever each portion of the B text was available; otherwise it could hardly have been started at all. This is specious; but a rather more plausible argument became available to him later, once he had proved (Galbraith 1950) that the processing of the survey text was coordinated with that of the current geld account. 27

Though Galbraith did not know it at the time, two passages in the Exeter manuscript were written by a ‘curial’ scribe (below, p. 7). But they are later insertions, not properly part of the C text; so they hardly affect the issue.

25

Why this portion survives ‘we do not know. . . . This is only one of many questions . . . that we cannot answer’ (Galbraith 1948, pp. 97–8). Such candour, in my view, is much to be preferred to gratuitous remarks – inspired by Baring (1912, p. 310) – about the ‘extreme complexity of the free tenures in East Anglia’ (Galbraith 1961, p. 8).

28 The same sort of drift which caused ‘Essex and East Anglia’ (1942, p. 166) to degenerate into ‘East Anglia’ (p. 168). The same sort of drift to which everyone is prone who teaches about the same subject year after year.

6

Introduction is, when it took an abnormal effort for any semblance of normality to be maintained. Douglas, writing an introduction for his facsimile edition of a manuscript from Canterbury, used the opportunity to make a riposte to Galbraith (Douglas 1944). Another historian, reviewing that edition, used the opportunity to declare himself ‘a convinced “Galbraithian” in most respects’ (Lennard 1946). At some stage (I do not know quite when), it would have become known that Galbraith was working on a book; and people aware of that would naturally prefer to wait for the book to appear. Excuses for doing nothing just yet are always welcome.29

have been working from the printed text. Time passed, the manuscript went back to Exeter, and Galbraith’s hope of writing about it ‘at greater length in the future’ (p. 1) remained unfulfilled.31 It was Finn (1951), not Galbraith, who noticed that there were two passages, added on blank pages of the manuscript, which were written by a ‘curial’ scribe – a discovery which obviously had to mean something important, though neither Finn nor Galbraith was ever able to make up his mind quite what that something might be. It was Finn, too, who made the first serious attempt to identify the numerous individual hands which participated in the writing of the main text. After looking at only part of the manuscript, he had already found a dozen different hands; but his results were greeted with so much incredulity – on Galbraith’s part, I assume – that Finn became discouraged and gave up. The article that he published on this subject includes a rather pathetic paragraph which amounts to an admission that he has probably got it all wrong (Finn 1959, p. 363); and when he wrote a whole book about this manuscript – a book which says so little to the point that I have not had to cite it elsewhere (Finn 1964) – he avoided the subject altogether.

At the time when he wrote his 1942 article, Galbraith was still assuming – like everybody else – that the geld accounts surviving at Exeter (below, pp. 60–9) were accounts of the six-shilling geld of 1083–4, referred to by a disgruntled English chronicler (Swanton 1996, p. 215). The question which seemed to need answering, therefore, was how a batch of two-year-old accounts might have become connected with the record of the survey (1942, p. 171). But some time later he saw that this was the wrong question. The dating proposed by Eyton – accepted even by Round, agreeing this once with Eyton – was in error: in fact it was possible to prove that these accounts are contemporary with the survey (Galbraith 1950). (There is also a remark, instantly forgotten, to the effect that the geld accounts appear to have been written in the treasury (p. 3). So indeed they do – because they were.) It has seemed to almost everyone, it seems to me, that Galbraith was perfectly right. Though no chronicler mentions the fact, it is clear that another sixshilling geld was being collected in 1085–6, and that the business of getting hold of the money and writing up the accounts was in progress concurrently with the business of the survey. Galbraith could produce enough evidence to prove the point; some further evidence (the significance of which is only to be seen in light of the subsequent history of the holdings concerned) was added soon afterwards by Mason (1954). When Darlington (1955), editing the Wiltshire accounts for the Victoria County History, professed to be still uncertain as to their date, Galbraith (1957), reviewing the volume in question, seems to have been more amused than annoyed by this gesture of recalcitrance.30

Galbraith’s book was finally published in 1961. From the preface, and from internal evidence (quite frequently a statement in one chapter is contradicted by a statement in another), it is clear that the book took a long time to write; apparently Galbraith had to wait till after his retirement in 1957 before he could concentrate on getting the book completed. Compared with what might have been hoped for, it is a meagre piece of work. Nineteen years on, it is still little more than a sketch. Much of the time, it gives the impression that Galbraith was writing from memory, without going back to look at the evidence again; and often his memory deceived him, not just on points of detail. Step by step, I have come to think that there is almost nothing of value in this book – nothing worth saying that had not been said before. By the time that Galbraith finished it, he had entirely lost his way: if he was right about anything, by now it was only by accident. This conclusion, so to speak, crept up on me. I did not invite it; I derive no pleasure from it. All the way through, I assumed that I would, after working through the evidence for myself, find out that I was merely rediscovering what Galbraith had already discovered. Looking back, however, I see that this never happened. Not once. Every time it turned out that Galbraith had got things wrong – had misstated the facts, had posed the wrong question or come up with the wrong answer, had waved away some serious objection with a facetious remark. As the reader will notice (below, p. 134), one of Galbraith’s ideas – an idea which I was hoping would be right – was given the benefit of the doubt until almost the final moment; but here again, when I examined the evidence

By establishing the date of the geld accounts, Galbraith had made an important contribution; but that, as I regret to say, was the last one. While he was working on that problem, the Exeter manuscript was loaned to the Bodleian by the dean and chapter, so that Galbraith could have the use of it; but nothing much came of that. There is a rather confused footnote regarding the various hands that appear in the geld accounts (Galbraith 1950, p. 6); but mostly he seems to 29

I apply this to myself. It is my excuse for writing nothing more on the subject of pseudo-Lanfranc (Flight 1997, pp. 187–90) that I am waiting for the critical edition of Bernard of Cluny, still said to be forthcoming one day.

30 But he accepted Darlington’s mistaken conclusion as to the sequencing of the different versions of the Wiltshire account (below, pp. 69–70).

31 One has only to look at Ker’s (1977) description of this manuscript to see how much needed to be done that Galbraith failed to do.

7

The survey of the whole of England more closely, I saw that his interpretation had to be rejected. Though I still wish that it had been right, I am sure now that it was not.

this procedure would be more efficient, that involves the assumption that the text was being dictated to the scribe, while the meeting was in progress; and that is highly inefficient.) As Galbraith got older, his antipathy for Round became increasingly overt,33 and this thought-experiment seems to originate in that. He could not deny that Round was right about Cambridgeshire; but he could deny that Round was right to generalize. To do that, however, he had to jettison the uniformity assumption.34

The basic problem had still not been (because it could not be) resolved. By stressing the decentralized nature of the compilation process, he was undermining the assumption that the process was largely the same for every county (or group of counties). Even in 1942, he had been aware that his argument was taking a turn which weakened that initial assumption; by now he had largely abandoned it. It is, we are told, ‘unlikely’ that any two commissions ‘proceeded entirely alike’ (1961, p. 35); to dare to assume that they did would be ‘unscholarly’ of us (p. 59). Instead we are invited to suppose that each group of ‘non-curial’ scribes was left to work out its own procedure, without guidance from the centre. One group of scribes submitted a fair copy of the survey text in a feodalized but unabridged form; but that does not mean that fair copies were submitted by every group of scribes. One group of scribes produced the collection of booklets which (mostly) survives at Exeter; but that does not mean that similar collections were being produced by other groups of scribes elsewhere. Nineteen years earlier, his intuition had told him, quite rightly, that the version of the text represented by the Exeter booklets was the vehicle used for transforming the ’original returns’ into a feodally organized version of the text. On that view the existence of D implies the existence of C (or something like it), and the existence of C implies the existence of D (or at least the intention to bring it into existence). Similarly the existence of C implies the existence of B. But that insight had been lost.

Preoccupied with the difficulties which he had created for himself, Galbraith failed to take full advantage of the new evidence which became available in 1952–3, when D and DB were rebound. That evidence was reported in a pamphlet published by the PRO (Jenkinson 1954). Galbraith’s annoyance with this pamphlet is only thinly disguised: he thought (with some reason) that he had not been given the prominence that he deserved.35 But there was information here which he could make use of, most of which would fit quite comfortably with his interpretation. With hindsight one can see which point was of greatest consequence. Citing the opinion of Alfred Fairbank, Jenkinson made the provocative suggestion – absolutely new, it seems – that DB might have been written by just one man. Most people seem to have ignored the suggestion; Galbraith was willing to consider it. He did not change his mind at once. It is one of the symptoms which go to show that his book had a long gestation that in some chapters he speaks of ‘the scribes’, in others of ‘the scribe’. Apparently through nothing more strenuous than introspection,36 he became increasingly convinced that Fairbank was right; and from there he went on to wonder (vainly) whether it might be possible to put a name to this man.37 It never occurred to him, as far as

By this time, in fact, he was making a positive effort to reduce the scope of the uniformity assumption (Galbraith 1961, pp. 64–6). The proven existence of a cadastrally organized version of the text for Cambridgeshire was no longer allowed to imply that a similar text existed for every county. Galbraith made the point that a feodalized version of the text (C-like in overall shape) could have been compiled during the survey itself: if the scribe who was servicing the meeting had a stack of booklets in front of him, one for each baron, he could write the facts recorded for each manor into the appropriate booklet.32 Would that not be an improvement on the procedure which – as admittedly Round had proved – was followed in Cambridgeshire? Instead of the whole text having to be copied out twice (first in a B-like and then in a C-like form), it is only copied out once (in a C-like form straight away). This argument is just a quibble: it would have been perfectly possible for things to be done in this way, but there is not the slightest reason for thinking that they were. (As for the suggestion that

33

Without having ever met him (1974, p. 9), Galbraith ‘had imbibed some morally well justified hostility to Round in the PRO’ (Southern 1978, p. 416).

34

The assumption is allowed to apply within a group of counties, to the extent that ‘the inherent differences between counties’ are overridden by ‘the tendency to uniformity imposed by the legates’ (Galbraith 1961, p. 167). In other words, it applies at best only within whatever limits are set by whatever version of Eyton’s conjecture one chooses to believe.

35 He was made to share a paragraph with Douglas, because ‘both have reviewed in some detail the conclusions of Maitland and others’ (Jenkinson 1954, p. 16). That is cruel; Galbraith was entitled to feel cross. For Jenkinson, as this comment shows, Maitland was still the standard authority. For Galbraith, Maitland was always a minor figure, just one more historian who had succumbed to ‘the dominating force of Round’s personality’ (1948, p. 99): adopting Round’s views, he carried them ‘to such extreme lengths’ that ‘even Round was rather embarrassed by the zeal of his distinguished disciple’ – or so ‘it may be surmised’ (1961, p. 15). This Maitland, the author of a book called Domesday and Beyond, published in 1907 (Galbraith 1948, p. 90), is an imaginary character, and his book will not be found in any catalogue. (Jenkinson’s Maitland is, of course, the real F. W. Maitland, the author of Domesday Book and beyond, published in 1897.)

32

Within each booklet, the order of the entries would reflect the order in which the hundred juries had made their appearance; so this order would be consistent from booklet to booklet, and eventually from chapter to chapter of D or DB. Hence proof of consistency in D or DB was not acceptable as proof of the existence of B, and one young historian – who had thought that he was doing something helpful by finding as much consistency as possible (Sawyer 1955) – was informed that he had wasted his time.

36 ‘The more one broods on the script . . . ’ (Galbraith 1961, p. 202). He did not have the courtesy, here or anywhere, to mention Fairbank by name. 37 The candidate proposed by Galbraith (1967) has attracted no support, as far as I am aware. An alternative suggestion by Chaplais (1987) has been more favourably received, but does not seem convincing to me.

8

Introduction I can see, that instead of trying to advance he ought to be retreating. Fairbank’s suggestion, if it is right, annihilates the distinction which Galbraith had made between ’curial’ and ’non-curial’ scribes. If DB was written by one scribe, the style of it has to be assumed to be an individual style: it cannot be taken to represent a collective ‘curial’ style.38 Conversely, a scribe cannot be said to be a ‘non-curial’ scribe just because he writes in a different manner from the DB scribe. (At this point, if he had reached it, Galbraith might perhaps have remembered what he had said about the geld accounts. That opening too was lost.) Yet this distinction was fundamental to Galbraith’s interpretation. If one cannot draw a line between ‘curial’ and ‘non-curial’ scribes, the interpretation has lost one of its principal supports. Sooner or later it was bound to fall flat – but not until people realized the implications of Fairbank’s suggestion, and that was slow to happen.

haps the whole) of DB. To make any genuine progress in understanding the survey, we have to grasp this fact and take full advantage of it. That was the message which Galbraith had been trying to get across in the 1940s, and he never forgot it himself. ‘These three’ – ‘three absolutely contemporary manuscripts’ (C, D, DB) – ‘taken in conjunction, and used properly, cannot mislead us’ (1970, p. 15).40 In the 1980s, around the time of the manuscripts’ nine hundredth anniversary, historians heard it again. To that extent at least, we are all convinced Galbraithians by now.41

2 Over the last four hundred years, the survey of 1086 has generated a volume of literature vastly in excess of this original documentation. I have not tried to read it all; I should doubt the sanity of anyone who did. Much of it was ephemeral, forgotten and deservedly forgotten almost as soon as it appeared.42 Much of it, good in its day, was, in the normal course of events, subsumed and superseded until it became of merely historical interest.

Because in the end he had rather little to say about the principal manuscripts, Galbraith’s book is padded out with some feeble discussion of various derivative texts – the same texts which had seemed so promising to Douglas, twenty-five years before. For some historians at least, Galbraith’s book seemed to imply that this was the way ahead: Galbraith himself had already said as much as needed to be said about the principal manuscripts, and the next step forward would be achieved by closer study of these derivative texts. There was briefly a time, during the 1970s, when this tendency appeared to be gaining ground. A few papers were published which were thought, not just by their authors, to be on the point of inaugurating a post-Galbraithian, neoDouglasian era. But the impetus soon died away.

There is, in any case, only a small proportion of this literature which touches on the fundamental issues, as I understand them to be: the logistics of the fieldwork phase and the mechanics of the compilation phase. Until these issues have been adequately grasped, there is little point, so it seems to me, in discussing anything else. Within this narrower field, I have tried to read everything relevant; but even here, no doubt, I have fallen short. Perhaps I should say that I have read much more than is listed in the bibliography. It is not my practice to cite any publication solely for the purpose of proving that I am aware of its existence. If a book or article seems to me to have some positive value, I have made a point of citing it; if not, I have preferred to ignore it. Over the last few years, I have spent a share of my time rereading publications that I had read before, to make sure that I was understanding them correctly and giving credit where credit was due. Despite my efforts, it is possible that I may sometimes have failed to do this, through ignorance or inadvertence. If so, I can only say that I regret it and am ready to do what I can to set the record straight.

Galbraith had not intended to start such a trend; when it started, he disapproved of it. It seemed to him a regrettable fact that these derivative texts ‘have of recent years attracted more attention than Domesday itself’ (1974, p. 76). The same miscalculation was involved which had carried Douglas off course in the 1930s – a failure to appreciate the difference in value ‘between the evidence of strictly contemporary manuscripts and of [even] slightly later copies’.39 No one can doubt that he was right about that. Though much of the documentation resulting from the survey has been lost, the amazing fact is that a good-sized fraction survives in the original – nothing of B, but roughly one-eighth of C, roughly one-sixth of D, and at least five-sixths (per-

Meanwhile, for as long as I was pursuing my own train of thought, I deliberately refrained from reading anything new. If a book or article was published before 2000, I allowed myself (or, in some cases, compelled myself) to read it; if after that, I did not. When I first imposed this embargo on myself, I hardly realized how long it would have to stay in place; but I still think that it was a sensible decision.

38

In some passages, Galbraith can be seen backing away from the assertion that DB is a specimen of contemporary ‘curial’ script. He says, for instance, that DB ‘is written in a single distinctive set-hand . . . which is not found elsewhere in our surviving materials. One is tempted to see in this script the copy-book hand taught to the scribes of the royal curia, but the hypothesis cannot be verified for lack of comparable evidence of so early a date (Galbraith 1961, p. 4). (This sounds to me as if he had been consulting with Chaplais.) The ellipsis represents a parenthetical remark – ‘possibly even by a single scribe’ – which administers the kiss of death.

40

This echoes an earlier statement to the same effect (1948, p. 100).

41

Or so I thought – until I read Roffe’s (2000) book.

42

Sifting through the dust, one may hope to find a few items which were undeservedly forgotten and ought to be reclaimed. As far as Kent is concerned, I can think of only one item which falls squarely into this category: an unfinished edition of DB-Ke (Larking 1869) which subsequent commentators have conspired to ignore.

39

This is one of the neo-Douglasians recalling her attempt to elicit some response from Galbraith (Harvey 1980, p. 125). The distinction that he was making seemed ‘puritanical’ to her.

9

The survey of the whole of England As my thinking began to take an original turn, it seemed increasingly safe to assume that nobody else was on the same track as me. That assumption, as far as I can judge, was sound enough.

book in which it appears; the other is David Roffe’s (2001) review of his own new book (Roffe 2000). Though it has attracted some favourable comment, not just from Roffe himself, I have to say that the book left me bemused. Roffe’s sense of the priorities seems thoroughly wrong-headed to me. His valuation of the evidence differs from mine to an extent which I would not have imagined to be possible. To note only the most striking contrast, his remarks about the Exeter manuscript are brief and unoriginal (Roffe 2000, pp. 94–8);47 they take up slightly less space than his remarks about the ‘Crowland Domesday’ (pp. 101–5). I was three-quarters of the way through the book before I found a passage which seemed on target to me (below, p. 104); and only one section of it is, to my mind, unquestionably an advance in the right direction. The idea that it might be possible to seriate the DB booklets had been in the air for some time, but Roffe was the person who knuckled down to the job and tried to get the seriation worked out (pp. 191– 211). His analysis is not completely successful, but he is entitled to the credit for making the first attempt.48 Apart from that, it is hard to see what common ground exists. On one point, however, I venture to think that Roffe will agree with me – that compromise is not the way forward. On matters of detail, no doubt, there is room for some give and take; beyond a certain point, it is no more possible than it is desirable to think of splitting the difference. One of us is on the right track, and one of us is not.

The world, however, did not stop turning in the year 2000, and I have recently been trying to catch up with it. Several important developments have taken place within the last few years, and it may be helpful for the reader if I comment on them briefly here. A pair of books had been sitting on my shelf for a considerable length of time before I was ready to read them; but I have finally got round to doing so. The collection of essays edited by Hallam and Bates (2001) includes a number of useful pieces,43 but only one which overlaps with the contents of my own book to any large extent. A long chapter by Frank Thorn and Caroline Thorn (2001) reports on the work that they had been doing, over the previous ten years or so, in collaboration with Michael Gullick. Caroline Thorn’s familiarity with the manuscripts is vastly greater than mine, and I will only say that there is very little here with which I feel at all inclined to disagree.44 Their seriation of the DB booklets (Thorn and Thorn 2001, p. 43) is (or was at the time) incomplete and tentative; but I am pleased to see that it is at least largely compatible with mine. I hope that they will work it out in full, in their forthcoming book, and will realize, as they do so, that Eyton’s conjecture is becoming vacuous.45 I also hope that they will stop stressing the suggestion – originally Baring’s (above, p. 4) – that DB derives immediately from C. In the nature of the case, this proposition is not demonstrably true; to my way of thinking it is presumptively false. The evidence that they adduce (pp. 67–8) is not in the least ‘compelling’; on the contrary, it is all ambiguous.46

By my own reckoning, this book of mine will only be successful if it encourages some readers to look or look again at the primary sources. Though the C text is still hard to get at,49 both D and DB have recently become much more accessible, but only in a qualified sense – more accessible for people who have deep pockets (or a first-rate library just around the corner), less so for the rest of us. The translations of the DB text which appeared originally in the Phillimore edition (1975-86) have been consolidated and released on CD-ROM (Palmer, Palmer and Slater 2000). How far this is a welcome development I have to confess to feeling doubtful. I do not mean to belittle the Phillimore edition, which – above all because the individual volumes are cheap – has served a valuable purpose, and will continue to serve it for some years to come. Thanks to John Morris, the man who initiated this edition (he died in 1977), thanks also to those who continued what he had begun, D and DB were, almost for the first time,50 made

Two of the pieces in Hallam and Bates (2001) are in the nature of book reviews. One (Holt 2001) is a review of the 43 The chapter by Prescott (2001, pp. 180–5) includes some interesting information, new to me, about the Record Commission’s ‘Additamenta’ volume (Ellis 1816). I have made some consequential corrections and additions in chapters 4 and 8. 44 To mention just one point on which I feel competent to speak, I agree with their explanation of the marginal addition in DB-Ke-9rb (Thorn and Thorn 2001, p. 59, ill. 28). (But I would add that neither of the places in question has been satisfactorily identified.)

47 And, if I may say so, not as well informed as they might be: there is no reference to Ker (1977) or Webber (1989).

45

It is disconcerting to see that they express their faith in this conjecture (Thorn and Thorn 2001, p. 42) in language which seems to be a deliberate echo of Galbraith’s (1942, p. 162). For Galbraith it was ‘generally agreed’; for them it is ‘generally accepted’. After almost sixty years, does the conjecture still have no stronger claim on us than that? (By that sort of reasoning, the earth would still be flat and whales would still be fish.)

48 The added footnotes in chapter 2 will indicate how far my own results were anticipated by Roffe. I ought not to have overlooked an earlier publication of his which came close to establishing the sequence for the first five counties (Roffe 1990, pp. 320–1 = 2000, pp. 202–3). 49 Caroline Thorn allows me to say that she is working on a new edition of the Exeter manuscript.

46

As was indicated by Holt (2001, p. 23), the issue is a matter of logic, not of evidence. Because Baring’s theory is (within the limits that they suppose to apply) not demonstrably inadequate, they say that it ought to be preferred to Galbraith’s because it is simpler – more economical – than his. Of course it is simpler in a trivial sense; but is it simpler in an interesting sense, all things considered?

50 The crude facsimile issued by the Ordnance Survey in the 1860s was not kept in print, and copies seem to have dropped out of circulation fairly soon. Whether it was worth resurrecting this facsimile and scanning it onto the Phillimore CD, I am not in a position to decide.

10

Introduction readily available to anyone who wanted to consult them.51 Without the Phillimore edition, the task which I set myself would have been impossible; and I think it only fair to say so. Nevertheless, the edition has rather more than its share of faults. The translations are eccentric, and often somewhat inaccurate; the editorial contribution varies greatly, in quantity and quality, between one volume and another. Every page of my copy of the Kent volume (Morgan 1983) is thick with pencilled corrections and annotations. It is no longer the original edition: it is the record of a one-sided conversation between the editor and me, protracted over twenty years. For my part, I cannot imagine having any use for a copy of this translation in an uncorrected and (more to the point) uncorrectable form. The Phillimore edition, I fear, was not ready to be cast in bronze.

be true or nearly true; but it is far from being true for Kent, where many places have been identified wrongly, or not as rightly as they might be.55 The same photographs taken in 1985 were used for creating a digitized facsimile of both manuscripts, released in January 2003 (Alecto Historical Editions 2003). Again, I am not sure how widely this CD-ROM edition will become available. It does not (as yet) appear in any of the catalogues that I have checked. (For that reason, again, I have not yet seen it myself.) I should not be surprised if it had to overcome some degree of consumer resistance. If I were a librarian who had spent a large sum of money, not many years ago, to buy what was advertised then as the best-possible facsimile of DB, I might hesitate before spending another large sum of money to buy (almost) the same thing again.56 In one respect, this new edition is less good: it cannot replicate the physical properties of the original manuscript. In the quality of the reproduction, however, it is (I have reason to expect) an astonishing improvement.57 The transparencies made in 1985, reproduced by a method which does them justice, are superb; and the photographer ought to be thanked by name by anyone who uses this facsimile. The entire edition is beyond the budget of most private individuals; but the intention is for each county text to be issued separately, and these ‘county editions’ come at a more affordable price.58

Back in 1985, while D and DB were being repaired and rebound, both manuscripts were photographed in their entirety, with a view to the publication of a new facsimile.52 The photographer was Miki Slingsby; he was assisted, he tells me (and asks me to state for the record), by Kerstin Firth-Clarke. The facsimile edition of DB was first published in 1986 (Alecto Historical Editions 1986); a separate edition for each individual county was issued over the next few years (Alecto Historical Editions 1987–92).53 A facsimile of the three D booklets came out at last in 2000 (Alecto Historical Editions 2000). I am not sure that this facsimile of D will become as widely accessible as that of DB; to judge from a few online catalogues which I have checked, not every library which purchased the DB edition has felt obliged to purchase this one too. (For that reason I have not yet had a chance to see it myself.)

These various attempts to make money out of DB – or, to 55

In the Alecto edition of DB-Ke, Williams and Martin (1992, p. 67) give a list of the place-names which they identify differently from Morgan (1983). It seems an impressively short list. The fact is, however, that they are simply repeating almost all of Morgan’s identifications, irrespective of whether they are right or wrong or something in between (and some of the changes which they have made are changes for the worse). I cite a few examples to illustrate the sort of inadequacies which occur. (i) Stoches (8va19) is Stoke, but that is not specific enough: the particular manor in question here is the one which came to be called Malmaynes (TQ 8175). (ii) Berham (9vb35) is Barham, but that name – as Ward (1933) pointed out – was being used in a very loose sense: the particular place in question here is Kingston (TR 1951). (iii) Middeltune (2va46) is Milton (TQ 9065), but that was a huge manor: a score of places which seem to be missing from DB are missing because they are silently included in this paragraph. (iv) Finally one example of an outright error: Boltone (4rb37) is not Boughton Malherbe, correctly identified elsewhere (8rb1); it is Boughton Monchelsea (TQ 7749).

The English translations originally made to accompany the Alecto facsimiles of DB and D have been extracted and collected together to make a separate book (Williams and Martin 2002), already out in paperback (2003). I do not know who is expected to buy this book and not lose interest after the first few pages. It would be a pity, in any case, if the availability of a translation diverted attention from the Latin text.54 It would also be a pity if this book conveyed the impression that everything has been settled, and that nothing is left to be done. For some counties this may (for all I know)

56 The CD-ROM edition includes the introductions commissioned for each volume of the ‘county edition’ (1987–92), which were previously hard to get hold of. (Few libraries which bought the ‘library edition’ saw any need to buy the ‘county edition’ as well.) When I put in a request through the inter-library loan system for a copy of the ‘county edition’ of DB-Ke, the copy which eventually arrived in Clemson came all the way from Boston Spa.

51

Not in their original form, but in the form which was given to them by an eighteenth-century compositor. His name, as far as I can see, went unrecorded. We know who made the copy and checked the proofs. We know who designed the type; we know who manufactured it; but we do not know who did the work of setting it. 52

The history of the Alecto facsimile is recounted by Pearson (2001). Almost twenty years on, I do not see why it should not be said that the DB facsimile was poorly reproduced, and that the apparatus could mostly have been done without. All that was wanted was a good facsimile, as cheap as it could be.

57

Some years ago, I wrote to the people at Alecto to ask whether something had gone wrong with the facsimile of DB-Ke-10r, where one patch of text is illegible (10ra15–18). They were kind enough to send me a fullsize reproduction of the original transparency. As soon as I recovered from my surprise, I wrote to ask whether they could let me have similar reproductions of all the other pages in DB-Ke. When they told me how much they would have to charge per page, I dropped the idea.

53

A collection of essays which accompanied the DB facsimile (Williams and Erskine 1987) has recently been repackaged (Erskine and Williams 2003). The format is different, the title is new, but the contents appear to be (except for the acknowledgments) identical.

58 The last time that I checked (November 2005), four of these county editions were available: D-Ex, DB-Yo (part of booklet DB-YoLi), DBDn (part of booklet DB-DnCo), DB-Wa. The Web site to watch is www.phillimore.co.uk.

54

With a little practice, the script of DB is quite easy to read, and the language is, most of the time, not difficult to understand. (This is business Latin, not the poems of Catullus.)

11

The survey of the whole of England put it bluntly, out of the ignorance of people who do not realize that DB means dreadfully boring – are not much to my taste;59 but they are, I am sure, only a passing phase. It will not be long (if it has not happened already) before somebody starts developing an online edition of DB and D, freely available to everyone. The basis for it will be, I hope, not an English translation but the tightest possible transcription of the Latin text – by clicking on which it will be possible to summon up a translation into English (or some other metalanguage), if that is what one wants, or an extended version of the Latin, or an explanatory note. (I also hope that the DB booklets will be put into the right order.) To say nothing of my incompetence, I do not have the time to undertake this task myself; but I should be happy to contribute, so far as I am capable, so far as Kent is concerned.

59

‘For anyone tracing the history of his or her family, homeland or village [this edition] is invaluable.’ Though no one should expect a publisher’s blurb to tell the plain truth, is this not a touch dishonest?

12

Chapter 2 Seriation of the DB booklets

From the enterprise known to contemporaries (if they could read Latin) as the Descriptio totius Angliae, ‘the survey of the whole of England’, three large batches of documentation survive in the original. As Galbraith (1942) was the first to realize, they represent successive stages in the process of compilation. Here I am concerned with just one of these batches, typologically the latest of the three – a collection of two dozen booklets, written (so it seems safe to say) in 1086–7, containing abbreviated versions of the survey reports for thirty counties. At some uncertain date, these booklets were bound together. The book thus brought into existence (rebound from time to time) has never been out of official custody. It is now in the Public Record Office, with call-number E 31/2. I propose to refer to it here as the Descriptio brevis, DB for short.1

ing, DB reveals more structure. Apart from the inserted slips already mentioned, it consists of 360 paired leaves (i.e. 180 folded sheets) and 19 single leaves, organized into 47 gatherings. Not counting the singletons, most of the gatherings (29 of them) are regular quires comprising eight paired leaves; but gatherings of six leaves (10), ten leaves (6) and four leaves (2) also occur. By and large, the division into gatherings correlates closely with the division into counties. Almost every county begins at the beginning of a gathering; almost every county ends near the end of a gathering, without much space being wasted. It is clear, up to a point, that this result was brought about deliberately, by adjusting the size of the gatherings to the expected quantity of text. These facts became knowable whenever the book was rebound, but nobody was obliged to take any notice of them, except for the binder himself. They could have been recorded in the 1860s; in fact the opportunity went untaken until the 1950s, when DB was repaired and rebound once more, at the instigation of the Deputy Keeper, Sir Hilary Jenkinson.5 A diagram published in 1954, showing exactly how DB is constructed (Jenkinson 1954, app. I), can fairly be said to have put discussion of the book on a solid basis for the first time. The credit for producing this diagram belongs to A. W. Mabbs.6

DB comprises 379 leaves, measuring about 370 mm long by 260 mm wide, plus four inserted slips. All the way through, the text is arranged in two columns, but the number of written lines per column varies greatly, from fewer than forty to more than sixty. The leaves were numbered in January 1660 by one of the Exchequer officials, Edward Fauconberge:2 his numbering included the inserted slips, but did not include the first leaf, originally blank.3 Modern commentators, not wanting to disturb Fauconberge’s foliation, which is otherwise perfectly accurate, have chosen to refer to the first leaf as folio 0.

Given the facts recorded in Mabbs’s diagram, DB can be seen to divide itself objectively into 27 booklets (Table 1, cf. Rumble 1985, pp. 34–5). There are 16 booklets which comprise a single county contained in a single gathering; six booklets which comprise a single county but extend into more than one gathering; and four booklets which comprise two counties, the second county beginning in the same gathering which contains the end of the first. The largest booklet is the one for Yorkshire and Lincolnshire, which runs to nine gatherings. Finally there is a single gathering contain-

As long as DB remains in its binding, the only obvious division is the division into counties.4 Every new county begins a new page; most counties are separated from the preceding county by one or more blank pages – usually only one or two, twice as many as seven. Removed from its bind1

It is sometimes called ‘Domesday Book’ (‘Domesday Book volume 1’, to be more precise). That name is a twelfth-century joke. The name was never very apt, and the joke was never very funny; I suggest we might cease repeating it.

5

3

This is where Fauconberge wrote a note recording what he had done and giving the date as 3 January 1659 (Jenkinson 1954, p. 20; Hallam 1987, p. 146). There is, I take it, no doubt but that he was starting the year in March.

The facts discovered in 1952–3 were put on record in a pamphlet published by the Public Record Office in April 1954. No author’s name appears on the title-page, but the text is obviously an individual (not to say egotistical) production, and I cite the pamphlet as Jenkinson (1954). As the preface acknowledges, however, all the preparatory work had been done by Jenkinson’s staff. (The copy which I have used is a second impression, ‘with corrections’, dated 1960.)

4

6

2

Edward Fauconberge (d. 1679) was chamberlain of the Receipt 1655–60, deputy chamberlain 1660–79 (Sainty 1983, pp. 19, 176).

Some parts of the country fell outside the basic scheme. Rutland (293va– 4ra) is appended to and indexed with Nottinghamshire (280vb); the New Forest (51ra–vb + 50vb) and the Isle of Wight (52rb–4ra) are appended to and indexed with Hampshire (37vb); six hundreds north of the Mersey (269va–70rb) are appended to and indexed with Cheshire (262vb). (In this last case there is no proper index, just a note explaining why in Cheshire no index is needed.)

The diagram showing the construction of DB (there is also a similar diagram for E 31/1, rebound at the same time) is credited to Mabbs in the preface (Jenkinson 1954, p. viii). This is Alfred Walter Mabbs, who joined the PRO as an assistant keeper in 1950 and retired as Keeper in 1982. His diagram is reproduced, with some alteration in the labelling, as Williams and Erskine 1987, app. III. In both versions, it has one defect: it fails to distinguish between singletons and slips.

13

The survey of the whole of England Fauconberge’s foliation

number of gatherings

paired leaves

single inserted leaves slips

[0]–15 16–29 30–36 37–55 56–63 64–74 75–85 86–99 100–125 126–131 132–142 143–153 154–161 162–178 179–188 189–202 203–208 209–218 219–229 230–237 238–245 246–251 252–271 272–279 280–296 297–372 373–382

2 2 1 3 1 1 1 2 3 1 1 1 1 2 1 2 1 1 1 1 1 1 3 1 2 9 1

14 12 6 18 8 10 8 14 24 6 8 10 8 16 10 14 6 8 10 8 8 6 20 8 16 74 10

1

47

360

19

2 2 1 1 1 1

2

2 3 1

2 1

1 2

counties Ke Sx Sy Ha Be Wi Do So DnCo Mx Ht Bu Ox GlWo He Ca Hu Bd Nn Le Wa St ShCh Dy Nm YoLi

Kent Sussex Surrey Hampshire Berkshire Wiltshire Dorset Somerset Devon–Cornwall Middlesex Hertfordshire Buckinghamshire Oxfordshire Gloucestershire–Worcestershire Herefordshire Cambridgeshire Huntingdonshire Bedfordshire Northamptonshire Leicestershire Warwickshire Staffordshire Shropshire–Cheshire Derbyshire Nottinghamshire Yorkshire–Lincolnshire clamores etc.

4

Table 1. Division of DB into booklets. ing a collection of supplementary material relating to these two counties. This gathering is a puzzle by itself, and I do not discuss it further here.7

body had ever thought to ask) that DB was the work of a team of scribes, all trained to write the same type of script. The fact that nobody had ever managed to distinguish the contributions of individual scribes was not allowed to undermine that assumption: on the contrary, it tended to reinforce the notion that DB was produced by a strictly disciplined scriptorium. The credit for realizing that the question needed to be asked – was DB the work of one scribe or many? – seems to belong to Jenkinson; the credit for getting the right answer belongs to Fairbank. There were, it seemed to him, ‘certain weaknesses’ in the script – individual foibles occurring throughout the book – which indicated ‘that a single hand was responsible for the whole’ (Jenkinson 1954, p. 34).

The pamphlet which included Mabbs’s diagram was momentous in another sense too. At Jenkinson’s invitation, the disbound leaves were examined closely by A. J. Fairbank – a civil servant in the Admiralty, but also, more to the point, a distinguished calligrapher and teacher of calligraphy.8 Having taught himself to imitate the script of DB, Fairbank participated in timed experiments designed to estimate how many man-days would be required to write the whole book. (The answer was: in the order of 240 mandays.) He also decided, from his study of the script, that DB was the work of a single scribe throughout.

Unlike Mabbs’s diagram, Fairbank’s suggestion did not command instant assent. It was a statement of opinion, not of fact; and Jenkinson made a point of assuring his readers that they were free to disbelieve it, if they chose.9 His expectation was that most readers would do just that –

That was, at the time, a quite astonishing conclusion. Till then, it seems to have been taken for granted (as far as any7

((The ruling is similar to that used for aspect 3, but not exactly the same. To judge from its spelling of French words (below, p. 141), this quire belongs with aspect 1 or the earlier part of aspect 2.))

9 Fairbank’s opinion drew its weight from his experience (not mentioned by Jenkinson) as a teacher of calligraphy, accustomed to scrutinizing every detail of his pupils’ work. But perhaps it would not be unjust to say that there were a few palaeographers active at the time whose opinion, had they been asked for it, might have carried greater conviction than Fairbank’s.

8 This is Alfred John Fairbank (1895–1982). A festschrift presented to him on his seventieth birthday (Osley 1965) includes a biographical chapter; there is an obituary in The Times, 20 March 1982, p. 8.

14

Seriation of the DB booklets would prefer to continue thinking that the work had been ‘distributed’, shared out among several indistinguishable scribes. After 1954, though Fairbank’s suggestion was not forgotten, it is hard to detect any definite progress until the 1980s, when a spate of important publications (Rumble 1985, 1987, Gullick 1987, Chaplais 1987) carried the discussion very much further forwards.10 Whether to rely on Fairbank’s opinion or not is no longer a pertinent question; but for thirty years it was.

away. A new formula appears, alternating with the old one at first and then replacing it. (An example of this is shown in Fig. 1.) If a change of this kind can be found, it is safe to assume – as safe as we can hope for it to be – that the change occurred during the writing of this particular manuscript. It makes no difference where the new formula came from, nor whether the scribe made the change consciously or not. All that matters is the solid fact that the change did occur. Any booklet which uses the old formula is sure to be earlier than this one; any booklet which uses the new formula is sure to be later than this one.

For historians, and for anybody else who wants to make use of this evidence, more is at issue than merely deciding whether to say ‘scribe’ or ‘scribes’. Interpretation may vary, perhaps to a large extent, depending on whether DB was a collaborative effort or the work of just one man. It is important to know how far we can feel sure of our ground.

There are 26 booklets to be dealt with. A preliminary classification – which is also a partial seriation – is given in Table 2. This classification is based, first, on variations in the ruling – the grid of vertical and horizontal guidelines scored onto each sheet before writing began.11 These variations seem to have gone unnoticed until the 1950s: once again it was Jenkinson’s pamphlet which first drew attention to them. While the sheets were disbound, the number of rulings on each was counted and recorded;12 and the facts discovered were published – not in full detail, leaf by leaf, but in summary, gathering by gathering (Jenkinson 1954, table I).13 I reproduce these data here, simplifying them even further.14

Suppose it is true that DB was written, entirely or almost entirely, by one scribe. Then it follows (subject to certain conditions) that we ought to be able to arrange the constituent booklets into a single sequence corresponding with the order in which they were written. Conversely, if we can arrange the booklets into such a sequence, that will tend to confirm the single-scribe theory. To put the question in archaeological terms, is it possible to seriate the booklets? Nobody would have thought of asking that question before 1954. It arose for the first time when Fairbank’s suggestion was juxtaposed with Mabbs’s diagram. Once the question is asked, however, the answer is clear enough. As this chapter aims to show, it is possible – mostly quite easy – to sort the booklets into a single sequence; and the single-scribe theory is, to some degree, corroborated independently by that fact. Furthermore, the sequence turns out to be very different from the sequence in which the booklets were eventually bound; and that may be, in the long run, the more important result.

Variation in the ruling is not enough by itself to form the basis for an adequate classification. We need more evidence; and for this we have to turn to the text itself. As even a cursory inspection will prove, there is wide variation, between one group of booklets and another, in the organization and wording of a typical entry. Thus the booklets which I classify as aspect 1 resemble one other and disresemble all the rest in some rather obvious features: here and only here, for instance, the first personal name to appear in each entry is the name of the man who held this manor T.R.E. (mean11 In several booklets, however, the scribe disobeyed the horizontal ruling, more or less consistently. Where this happens, the result is always that the written lines are closer together than the ruled lines.

1

12

As far as I can gather, all or most of this work was done by another of the assistant keepers, D. H. Gifford, whose contribution is acknowledged in the preface (Jenkinson 1954, p. viii), but only in general terms. This is Daphne Heloise Gifford (d. 1990).

Seriation would not be possible unless the scribe had varied his procedure, to some significant extent, as he went along. If he had settled on a plan in advance and stuck with it from start to finish, we would have no hope of seriating the booklets. On the whole, the scribe had a definite idea what format he wanted to use and what facts he wanted to include, and he followed this format and reported these facts consistently throughout. But that did not prevent him from altering his plan, in many small and some quite large respects, as the work progressed. Examples are quoted below. For purposes of seriation, the most useful changes are those which occur gradually, over the space of several columns, in the middle of a county. Before our eyes, one formula fades

13 All this evidence was looked at again by Gullick (1987, pp. 94–7). At some points, the results which he reports are at variance with those reported previously by Jenkinson, and I cannot feel sure that Gullick’s results are always closer to the truth. Here I have to speak carefully, because what is say is based only on a study of the facsimile – and the facsimile, of course, did not aim to optimize the visibility of the ruling. It is generally possible to see some trace of the lines, or of the prickings for them, but only rarely possible to see the whole pattern. In the footnotes, I have indicated some of the points which seem to be in need of further investigation. But the seriation does not depend entirely (nor even mainly) on this evidence, and the areas of uncertainty are not alarmingly large. ((Caroline Thorn, who has scrutinized the evidence more closely than ever before, has been kind enough to clarify some details for me. The reader may be sure that all doubts will be dispelled when her findings are published in full; but I have not thought it right to anticipate them here.))

10

Around the same time, the Phillimore edition of DB was brought to completion, and the Alecto facsimile was published. DB itself, disbound in order to be photographed, was rebound once again, as two volumes, in 1985–6.

14

In Jenkinson’s table, ‘unruled’ should be taken to mean that there are two horizontal lines, one each at the top and bottom of the frame (Gullick 1987, p. 95).

15

The survey of the whole of England aspect 1 2 3 4 5 6

ruling 8 8 7 4 4 4

booklets

44 44 50 2 50–59 2

Hu, Dy, Nm, YoLi Mx, Ht, Bu, Ca, Bd Ke, Sx, Sy, Ha, Be GlWo, He, Wa, St, ShCh Wi, Do, Ox, Nn, Le So, DnCo

Table 2. A preliminary classification of the DB booklets. ing tempore regis Eduuardi, ‘in the time of king Edward’). When the evidence supplied by the text is combined with the evidence of the ruling, the classification falls out quite easily.15

YoLi Yo is earlier than Li.18 The formula T’ra ad .. car’, which evolves near the end of Yo, persists halfway through Li. Then it contracts to T’ra .. car’ (Fig. 1), and in that simplified form persists throughout the rest of aspect 1.19

The seriation is worked out, link by link, in the following paragraphs. Some of the relevant transitions are illustrated by diagrams; their format will, I hope, be easy to understand.16 The argument is, by design, as sparse as I can make it. Because it can be proved that Sx is later than Ke, it does not need to be proved that Ke is earlier than Sx. Because it can be proved that Sy is both later than Sx and earlier than Ha, it follows immediately (since Sy is the only county for which both statements hold true) that there is only one possible place in the sequence for it. The reader will have no trouble finding evidence which tends to confirm the sequence given here. At least some of the relevant evidence was known to Jenkinson (1954, pp. 30–2); but he disabled himself from making sense of it, by presuming the order of binding to be original.

Nm Later than Li, earlier than Dy. Dy The formula Ibi n’c in d’nio .. car’ evolves early in Dy, replacing an unstable formula – Ibi m o in d’nio .. car’ or Ibi in d’nio .. car’ – which occurs sporadically in Nm. Evolving here, the new formula persists throughout most of Hu, the last booklet in aspect 1, with only slight variation (but consistency seems to be breaking down near the end). Hu

Later than Dy.

Aspect 2 The transition to aspect 2 is marked by (among other changes) a drastic alteration in the value clause. In aspect 2 the clause looks basically like this: Val’ .. sol’, Quando recep’ .. sol’, T.R.E. .. sol’. Three values are given, not two; and the T.R.E. value is given last, not first.

Aspect 1 In aspects 1–2, the sheets are ruled with 8 vertical lines (a pair of lines on either side of either column) and 44 horizontal lines.17 In aspect 1, by and large, the written lines conform to the ruling; in aspect 2, by and large, they do not. The distinction appears more sharply in the text – in the value clause, for instance, which has two parameters in aspect 1, three in aspect 2 (see below). The usual formula here is: T.R.E. ual’ .. sol’, m o .. sol’.

Mx The Quando recep’ formula can be seen evolving in Mx: et quando, then quando, then Quando. Having evolved, it persists throughout the rest of aspect 2, and (with qualifications) further still. Bd In cases where the Quando recep’ value is the same as the current value, Bd tries three formulas which avoid repeating the numeral (Fig. 2). It starts with Quando recep’ similit’, which is the formula found in Mx; then it shifts towards et tntd’ quando recep’; and then it makes a more drastic shift, telescoping the first two terms to make the formula Val’ et ualuit .. sol’, T.R.E. .. sol’. (This is an echo from the formula Val’ et ualuit semper .. sol’, employed

15

((A similar approach was taken by Roffe (1990, 2000), whose findings, so far as they anticipated mine, are cited in footnotes below.)) 16

Gatherings are shown by square brackets. Paired leaves are numbered, single leaves marked ‘S’, inserted slips omitted. Frequencies are counted column by column: the recto columns appear above the line, the verso columns below it. Each instance is represented by a unit square; for aesthetic reasons, the individual histograms are centred (as in the classic paper by Dethlefsen and Deetz (1966), to which I hope some readers will recognize my indebtedness). All the counting was done from the printed text, the errors in which are very few, and also (more importantly) unsystematic: a small number of sporadic errors, including any made by me, will not affect the pattern. For one diagram (Fig. 3), I have checked each item against the facsimile, without finding a single mistake. Anything visibly added as an afterthought – in the margins, between the lines – is disregarded; thus in this figure I ignore the ualb’ which appears in an entry written across the foot of a page (17v) and count T.R.E. `et post´ ualeb’ (18ra) as an instance of T.R.E. ualeb’.

18 The Yorkshire text deserves very close analysis, because this is where we can see the scribe coming to terms with his task – deciding, for instance, what degree of abbreviation is appropriate. That explains why (but only at first) he shows ‘a surprising taste’ – as Jenkinson (1954, p. 32) put it – ’for writing carucata in full’. The scribe has reached 303va before he decides that caruca ‘plough’ can normally be written car’; he has reached 315rb before he decides that carucata ‘ploughland’ can normally be written car’ t’r˛e, and that his readers can be trusted to tell the difference. (Here and everywhere, it has to be remembered that the scribe is thinking in French and assuming that his readers will do the same: the words he is trying to communicate are carue and caruede.) Many more such examples can be found. For present purposes, however, the only point which needs to be established is that Yo is earlier than Li; and that is obvious enough.

17

The first two YoLi gatherings are anomalous: they were ruled before the scribe had settled on a plan. The distinction made by Gullick (1987, p. 96) between ‘pattern 1a’ and ‘pattern 1b’ seems too slight to be significant.

19 ((The sequence for aspect 1 was worked out by Roffe (1990, pp. 320–1 = 2000, pp. 202–3), except that he did not prove that Hu is later than Dy.))

16

T’ra arabil’

T’ra totid’ car’

T’ra ad totid’ car’

T’ra .. car’

T’ra ad .. car’

Seriation of the DB booklets

LINCOLNSHIRE

349 350 351 352 353 354 355 356

Figure 1. Part of booklet DB-YoLi. Transition from T’ra ad .. car’ to T’ra .. car’. (Instances which replace the numeral with the word totid’ are counted separately (channels 3–4); here the transition starts later but finishes sooner.) when all three values are the same.) The telescoped formula is the one which is used, almost exclusively, in the rest of aspect 2.

comes unstable, the verb being frequently omitted. The oscillation continues into Be. Be Halfway through Be, the formula T.R.E. ualb’ .. sol’ is replaced by Valuit .. sol’.

Bu, Ht, Ca All later than Bd, but I have not been able to seriate them satisfactorily with respect to one another.20

Aspect 4 With the transition to aspect 4, again the ruling changes: the number of vertical lines is reduced to four (that number persists throughout the rest of DB); and horizontal ruling is dispensed with.23 The assessment clause changes to Ibi .. hid˛e geld’.24

Aspect 3 The transition is marked by a new pattern of ruling, with 7 vertical and 50 horizontal lines, and reflected also by changes in the wording. For example, the value clause is turned back to front, the T.R.E. value now being given first. In Ke the clause looks like this: T.R.E. ualeb’ .. sol’, Quando recep’ .. sol’, Modo .. sol’; here and later, Quando recep’ alternates with the simpler phrase et post. With some variation in detail, that basic arrangement persists throughout aspect 3.21 Ke

GlWo This booklet and the next one are in need of detailed analysis. Gl is sure to be earlier than He, and that is the main point; but the textual evidence seems to suggest that Wo is later than He.

Earlier than Sx.22

He Late in He, the formula Val’ et ualuit .. sol’, normal in Gl and before, starts to alternate with Valuit et ual’ .. sol’, normal in the rest of aspect 4.

Sx Early in Sx, the formula T.R.E. ualeb’ .. sol’, normal in Ke, changes to T.R.E. ualb’ .. sol’ (Fig. 3), normal in the rest of aspect 3.

ShCh Early in Sh, one category of information changes place. The number of slaves stops being put after the number of ploughs belonging to the peasants, and starts being put after the number of ploughs belonging to the lord. This new arrangement persists into Ch and throughout the rest of aspect 4.

Sy Later than Sx, earlier than Ha. Ha

Late in Ha, the formula T’c se defd’ pro .. hid’ be-

20

((Having looked at these booklets again, I am now inclined to think that Ca should follow Bd (below, p. 141); but Bu and Ht continue to defeat me.))

St

21

Later than ShCh, earlier than Wa.

23 The first GlWo gathering (162–9) retains the same vertical ruling as aspect 3. According to Jenkinson (1954, p. 26), the sheet at the centre (165 + 166) has only six ruled lines. This seems to be right; but the sheet was pricked for seven lines, like its companions, and there is no need to make a special case of it.

((That Ke is the earliest aspect 3 booklet was recognized by Roffe (2000, p. 207).))

22

The second Ke gathering (8–15) seems to have been made up from the stock of parchment left over from aspect 2. Gullick (1987, p. 96) points out that the sheets have been re-ruled to increase the number of lines. This seems sure to be right: the two systems of parallel lines are distinctly visible on the blank leaf at the end (15), and so are the two sets of prickings.

24 ((The sequence for the first three aspect 4 booklets, GlWo He ShCh, was established by Roffe (2000, pp. 208–9).))

17

Val’ et ualuit .. sol’

et tntd’ quando recep’

Quando recep’: similit’

Quando recep’: .. sol’

The survey of the whole of England

MIDDLESEX

126 127 128 129 130 131 209 BEDFORDSHIRE

210 211 212 213 214 215 S S 218

Figure 2. Booklets DB-Mx and DB-Bd. Oscillation between Quando recep’: similit’ and et tntd’ quando recep’, followed by a transition to Val’ et ualuit .. sol’. Wa Early in Wa, the formula X tenuit et liber homo fuit, normal in St, is replaced by X libere tenuit T.R.E. (Fig. 4).25

in Ox. Ox

Aspect 5 Horizontal ruling is reinstated in aspect 5.26 The number of lines is variable, but 53 or 54 seems to be the norm. Nn

Later than Le, earlier than Wi.

Wi Halfway through Wi, the formula De hac t’ra s’t in d’nio .. hid˛e, normal in Ox, changes to De ea s’t in d’nio .. hid˛e (Fig. 5). The new formula persists into Do and beyond.28

Earlier than Le.

Le Late in Le,27 the formula In d’nio s’t .. car’, normal in Nn and before, is replaced by N’c in d’nio .. car’, normal

Do Later than Wi. Aspect 6

25

Horizontal ruling disappears again.29

((That Wa follows St was recognized by Roffe (2000, p. 209–10).))

So Late in So, the formula X tenuit T.R.E. et geldb’ pro .. hid’, normal in Wi and Do, changes to X teneb’ T.R.E. et geldb’ pro .. hid’ (Fig. 6), normal in DnCo.

26

This is where Gullick’s results diverge the furthest from Jenkinson’s. There are two areas of disagreement. (1) Three of the gatherings in aspect 5 (Le, Wi and Do) are said by Gullick to have six vertical rulings. Looking at the facsimile, I can see no proof of this. I think it possible that the scribe may sometimes have ruled an extra line where he thought he had made a column too narrow at first; but I am not convinced that he ever ruled six lines as a regular policy. (2) Two gatherings (Nn and Ox) are said by Gullick to lack horizontal ruling. It seems to me that some ruling can be discerned in both these gatherings; the last leaf in Nn (229) is quite certainly ruled on the verso, with 4 vertical and 52 horizontal lines.

28

((The sequence Wi Do So DnCo was worked out by Roffe (2000, pp. 207–8). It is too obvious, however, for anyone to claim much credit for discovering or rediscovering it.)) 29 In this negative respect, aspect 6 resembles aspect 4 more than it resembles aspect 5. But negative evidence does not count. According to Jenkinson’s table, part of Do is also unruled; I cannot be sure, from the facsimile, whether that is right or not.

27

The sheet at the centre of this gathering (233 + 234) seems to have been left over from aspect 3 (Gullick 1987, p. 96).

18

TRE: .. sol’

TRE ualb’ .. sol’

TRE ualeb’ .. sol’

TRE ual’ .. sol’

TRE ualuit .. sol’

Seriation of the DB booklets

16 17 18

S U S

S E X

19 20 21 22 23 24 S 26 27 S 29

libere tenuit TRE

tenuit et liber homo fuit

Figure 3. Booklet DB-Sx. Transition from ualeb’ to ualb’.

STAFFORDSHIRE

246 247 248 249 250 251 238 WARWICKSHIRE

239 240 241 242 243 244 245

Figure 4. Booklets DB-St and DB-Wa. Disappearance of one formula coinciding with the appearance of another.

19

De ea s’t in d’nio .. hid˛e

De hac t’ra s’t in d’nio .. hid˛e

De hac t’ra .. hid˛e s’t in d’nio

In d’nio s’t .. hid˛e

The survey of the whole of England

64 65 66 WILTSHIRE

67 68 69 70 71 72 73 S 75 77

DORSET

78 79 80 82 83 84 S

Figure 5. Booklets DB-Wi and DB-Do. Stabilization of word-order in the formula De hac t’ra s’t in d’nio .. hid˛e, followed by a transition from De hac t’ra to De ea. fairly rapid succession, without much pause for thought.) As far as it goes, however, the seriation is robust. I am sure that this sequence is right, and that anyone who thinks of repeating the experiment will arrive at the same result.32

DnCo Later than So. That is the end of the sequence. The conclusions arrived at are summed up in Table 3.30 .

only booklet which, more often than not, replaces iacuit et iacet with iacet et iacuit semper, presumably through interference from the formula Val’ et ualuit semper . . . .) So far as that is true, it suggests that Ca may be either the earliest or the latest of the batch; and since it cannot be the earliest (that place is already occupied by Mx), it may possibly be the latest. But the indications are slight, and I do not regard them as conclusive. ((Though I no longer agree with it, I allow this footnote to stand; at least I was right to say that this evidence is not conclusive. It seems rather to mean that the scribe became slack, while he was writing Ca, but then tightened up again.))

2 It is clear that this table is not the last word on the subject. For a start, the latter part of aspect 2 has still to be sorted out: there are three booklets here which I have not been able to seriate satisfactorily.31 (Presumably they were written in 30

32

Sceptical readers are welcome to try constructing a different seriation. If there are significant resemblances which cut across the sequence that I am proposing, it will be interesting to see how they can best be explained. (There are several possibilities.)

((A modified version of this table will be found in chapter 11))

31

In some respects, Ca seems to stand apart from the rest of aspect 2. (For example, in the formula Hoc m’ iacuit et iacet in d’nio e˛ ccl’˛e . . . , Ca is the

20

tenb’ TRE et geldb’

teneb’ TRE et geldb’

tenuer’ TRE et geldb’

tenb’ TRE et geldb’

teneb’ TRE et geldb’

tenuit TRE et geldb’

Seriation of the DB booklets

86 87

89 90 91

M E

92 93

S O

R S E

T

88

95

94

96 97 98 99

Figure 6. Booklet DB-So. Transition from tenuit to teneb’. (Plural instances are counted separately (channels 4–6); but the distinction becomes invisible when the verb is put into the imperfect tense.) main sequence. In Ke, for example, one of the added entries (4va) uses a formula – Olim xx sol’, Modo ual’ xxx sol’ – which is characteristic of aspect 6. (It starts appearing near the end of So and persists throughout DnCo.) Instances of this kind corroborate the basic sequence (by proving, as this entry does, that aspect 3 is earlier than aspect 6), and they also shed light on the history of the individual booklet (by proving, as this entry does, that Ke was being corrected, or still being corrected, only shortly before the whole project came to an end). The same Olim . . . Modo formula appears in some of the inserted entries in aspect 1 – most conspicuously in Nm. If the scribe had ceased work a few weeks sooner than he did, those additions (so it seems) would not have been made; if he had continued for a few weeks longer, there is no knowing what other additions he might have wanted to make. Because DB as a whole was left unfinished, it cannot be said for certain that any of the constituent booklets was completed, except in an adventitious sense (the scribe stopped making additions in a given booklet because he had stopped work altogether); and in that sense the booklets were all completed simultaneously.

Even if the sequence were fully resolved, it should still be regarded only as a first approximation. The sequence which I am seeking to establish here is the order of inception; I do not suggest that each booklet was completed – absolutely completed – before the next was started. From more detailed analysis it may appear that the scribe sometimes interrupted work on one booklet in order to start another, not going back to finish the first booklet till after he had finished the second.33 Each booklet has a history of its own; and that history will need to be compared with the basic sequence, once that sequence has been adequately worked out. In every booklet (though in some much more than in others) entries occur which are manifestly later than the main text – because they are written in the margins, because they lack rubrication, or because they display some other incongruity. In checking through what he had written, the scribe discovered that he had omitted an entry which ought to have been included (or, rarely, vice versa) and inserted it as best he could (or cancelled it). In some cases, he may have discovered his error almost immediately; in others, perhaps, not till very much later. To some extent it may be possible to seriate such added entries by matching them with the

There is, however, one action taken by the scribe which amounts to a declaration that in his view the booklet is finished, or very nearly so. At some point he decides that the booklet is ready to be rubricated and sets himself to work, highlighting some elements of the text and adding some

33

There is one place (but only one place) where I suspect that this may have happened. If we were seriating counties rather than booklets, I think that He might fall between Gl and Wo.

21

The survey of the whole of England aspect 1 1 1 1 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4 5 5 5 5 5 6 6

binding sequence

17th-century foliation

Yorkshire–Lincolnshire Nottinghamshire Derbyshire Huntingdonshire Middlesex Bedfordshire ? Buckinghamshire ? Hertfordshire ? Cambridgeshire Kent Sussex Surrey Hampshire Berkshire Gloucestershire–Worcestershire Herefordshire Shropshire–Cheshire Staffordshire Warwickshire Northamptonshire Leicestershire Oxfordshire Wiltshire Dorset Somerset Devon–Cornwall

26 25 24 17 10 18 12 11 16 1 2 3 4 5 14 15 23 22 21 19 20 13 6 7 8 9

297–372 280–96 272–9 203–8 126–31 209–18 143–53 132–42 189–202 0–15 16–29 30–6 37–55 56–63 162–78 179–88 252–71 246–51 238–45 219–29 230–7 154–61 64–74 75–85 86–99 100–25

Not seriated: clamores etc.

27

373–82

counties

Table 3. Seriation of the DB booklets. other elements in spaces which he has reserved for them. In the history of every booklet, there are, it seems, only two definite moments: the moment when the scribe starts writing, and the moment when he starts rubricating what he has written. (The ruling of the sheets is a definite moment too; but that precedes the formation of the booklet.) It may be possible to construct an independent seriation of the booklets based solely on variant features of the rubrication. If this can be done, even if only with partial success, the order which emerges ought to be approximately the same as the order of inception; but it need not be exactly the same.

which appears in Nm relates to both counties: the survey of the town of Nottingham (280ra) is followed by the survey of the town of Derby (280rb), and then by an account of the customs of Nottinghamshire and Derbyshire, the two counties being treated together but named in that order (280va). The survey of Derbyshire, therefore, was certainly meant to follow the survey of Nottinghamshire (with its Rutland appendix), completing the order of business implied by the preliminaries; and the fact that Dy begins at the start of a new gathering (with five pages left blank at the end of Nm) does not mean that it forms a new booklet. That reduces the number of booklets to 26 (or to 25, if the gathering which I have ignored is treated as an appendix to YoLi, not as a separate booklet).

3

Second, there are some fairly definite discontinuities. In the early booklets large changes occur, but they do not coincide with the start of a new county. The beginning of Li resembles the end of Yo, the beginning of Nm resembles the end of Li, and so on. But eventually we come to a break: the beginning of Mx is obviously very different from the end of Hu. Subsequent discontinuities are not nearly as sharp as this one, but breaks occur fairly definitely at the beginning of Ke and GlWo, less definitely later on. From the fact that

Despite these uncertainties, we have a clear enough idea of the order in which the booklets were written to draw some conclusions from it. First, at one point the division into booklets needs to be reconsidered. It turns out that Nm and Dy were intended to form a single booklet, even though Dy consists of a separate gathering.34 The preliminary material 34

Uniquely, these two counties were coupled together during the fieldwork phase of the survey. One sheriff was in charge of both; one meeting at least was convened at which both counties were represented.

22

Seriation of the DB booklets START

Yo 













Ch







Nm

Dy



Li

















St





Sh



Le










Nk



Hu











Wo

Nn 

Wa



He

Ca


Bd 







Sk

Ox









Bu 

Gl

Be



Wi 









 









Ha

 





Mx Sy













Dn

Ex














So

Ht 





Do

Ke





Sx





Co





FINISH

Figure 7. The sequence of DB booklets represented as a tour of the country. these discontinuities fade away, it seems obvious that they originated during the writing of DB. The scribe, I suppose, set himself a target: he would complete a certain number of counties (five would be a good number), and then he would take a break. Having reached his target, he stopped work, and spent some time looking over what he had done. Perhaps he discussed it with his colleagues; perhaps he had to submit it to a supervisor. By the time he started work again, he had decided (or had been advised or instructed) to make certain changes in the treatment of the information. Early on, while he was still feeling his way, those changes might be very large ones; later, after he had settled into his task, the changes were relatively slight. As far as I can see, there is no need (and therefore no justification) for any deeper explanation than that.

preceding county. The sequence thus represents an imaginary journey around the country, starting in Yorkshire and finishing in Cornwall (Fig. 7).36 There are some anomalies, however, and possibly they mean that the accidents of the compilation process constrained the scribe to vary the order which he would otherwise have preferred. His headlong descent from Derbyshire through Huntingdonshire into Middlesex seems out of keeping with the overall plan. The explanation may be that after the scribe had finished with Dy the source text that he wanted next was not yet ready; and since he could not afford to wait for it his only other option was to jump ahead. At this juncture, therefore, the order did become haphazard, but only momentarily. Perhaps the same explanation applies to the three counties – Norfolk, Suffolk, Essex – which were (apparently) omitted from his itinerary.37 However we think of explaining them, these anomalies tend to prove that the order in which the booklets were written was not exactly the same as the order in which the scribe would have chosen to write them, if his choice had been unconstrained.

Third, the order is not haphazard – though conceivably it might have been, if the scribe had let it be dictated by the order in which the source texts became available.35 However the compilation process worked, it seems safe to assume that the source texts were not all completed simultaneously; and the scribe must surely have been constrained by that fact, to some degree. But he was not so tightly constrained that he could not think of imposing some order on the booklets; and the order that he imposed is – by and large – a geographical order. As far as he could, he arranged things so that each new county was contiguous with the

By extension, the order in which the booklets were written cannot be assumed to be the same as the order in which the scribe would have wanted to have them bound, when that moment arrived. As far as I can see, there are no indi36

((The crux is the intersection of edges (Be, Gl) and (Ox, Wi), where the sequence loops back over itself: the moment when I understood that was my eureka moment. I am gratified to see that Thorn and Thorn (2001, p. 43) incline towards the same conclusion.))

35

((Though I have shortened this paragraph, which seems very primitive to me now, I have not rewritten it with the advantage of hindsight. At the time, I think, it was an honest attempt to face up to the evidence – without taking anything for granted, without postulating deep explanations for what might be merely surface phenomena.))

37 The scribe may have intended to complete his imaginary tour by writing booklets for Nk, Sk and Ex. Apparently he never did so: but the fact that no such booklets survive does not prove that they never existed.

23

The survey of the whole of England cations as to what the ultimate ordering would have been, if the scribe had completed the manuscript and prepared it for the binder himself.38 He had committed himself on some points – Li had to follow Yo, Wo had to follow Gl, Ch had to follow Sh, Co had to follow Dn – but in most respects he had kept his options open. This does not necessarily mean that he had any positive intention of shuffling the booklets into a different order: he may simply have found that it was generally more convenient, while the text was being written, and perhaps even more so while it was being checked, if each county occupied a booklet by itself. My guess is that he would have made a few transpositions, but not more than a few. There is no certainty, however. For all we know, he may have been intending to rearrange the booklets into some completely different order – or, alternatively, may never have made up his mind. It is disappointing not to know how the components were supposed to be assembled: it means that we cannot form any distinct idea what the scribe would have wanted the finished book to look like. From one point of view, this uncertainty may be welcome. If we knew beyond doubt what ordering had been intended by the scribe, we should have to think of taking the book apart, rearranging the booklets, and renumbering every leaf. Because we do not, we are spared from doing that. The order in which the booklets are bound will have to be left unchanged. We may not like it, but we are not in a position to say that it is wrong (except as far as Dy is concerned).39 On the other hand, anyone working with a facsimile (or with the Phillimore edition) will presumably want to reshuffle the booklets, so as to be able to deal with them in their original order.40 Much will be gained by doing so, nothing lost. Anyone who thinks that it will make no difference has not thought hard enough. Perhaps it might be excessive to say that anything written about DB in ignorance of this seriation was written too soon. But there cannot be many questions which will not need to be reconsidered once the seriation has been thoroughly worked out.

38

Ten gatherings are signed; but the signatures are not original (Gullick 1987, p. 104), and in any case they only serve to keep the YoLi booklet in order. Jenkinson pointed out that the red D in DOUERE (1ra) is, by some margin, the largest coloured initial in the book, and might be taken as a sign that the Ke booklet was meant to come first. This seemed to him ‘a small point’, worth mentioning only in a footnote (Jenkinson 1954, p. 32); I concur. 39

It was certainly a mistake to put the third NmDy gathering in front of the other two: in this respect, without doubt, the order of binding is not the order originally intended.

40

Sooner or later, I suppose, it will have to be decided whether to introduce an alternative foliation, to be used for some purposes in preference to Fauconberge’s.

24

Chapter 3 Reconstruction of the D volumes

The enterprise known to contemporaries as the ‘Survey of the whole of England’ produced a vast quantity – many thousands of leaves – of parchmentwork. Most has been lost; but happily three large batches of documentation, all written (so it seems safe to say) in 1086–7, survive in the original. I call them C, D and DB (Table 4). They exemplify three successive stages in the processing of the information collected during the fieldwork phase of the survey. As Galbraith (1942) was first to realize, the presumption is that the report for every county passed, or was at least intended to pass, through the same three stages.

degree of success, I go on to deal with some consequential questions. In particular, I try to get some grasp on the process by which D and DB evolved, over time, into something quite different from what was originally intended – something which, by around 1180, had begun to be referred to as (this is a joke) the ‘Book of Judgment Day’. A sequence of diagrams (Fig. 8) may help the reader to visualize the fourstage model which I propose.4 If anyone thinks of calling this the cuckoo theory, I shall not take offence.

1

In general appearance, C and D resemble another while DB stands apart. In DB the page-size is nearly twice as large as it is in C and D, and the text is laid out in two columns. Moreover, DB was written by a single scribe throughout,1 whereas C and D were produced by teams of scribes. In the organization of the contents, on the other hand, it is D and DB which are similar and C which is different. In C the basic unit, normally comprising a booklet by itself, is the lands of an individual tenant, extending over as many as two or three counties;2 in D and DB the unit is a single county. These facts are enough to indicate that D belongs between C and DB: in some respects it agrees with C against DB, in others with DB against C. To put it briefly, the D text is a rearrangement of the C text, and the DB text is an abridgement of the D text, put into a larger format.

The three surviving D booklets cover one county each; and that, surely, must have been the regular policy. Its advantages would have been at least as obvious to the scribes concerned as they are to us. It meant that they could start the booklets in whichever order was most convenient, without worrying what the final order would be; it meant that they could work on two or more booklets in parallel, without having to finish one before the next could be started; and it meant that they could easily make additions, if the need arose, at the end of any booklet. (An example of this occurs in D-Ex, where some detailed information relating to the town of Colchester was added at the end.) I think we may safely assume, by and large, that each county occupied a booklet by itself. One exception to this rule, however, is fairly sure to have been made. It seems clear that Nottinghamshire and Derbyshire were treated as a unit at every stage – in the fieldwork phase, when a single meeting was held for both counties (DB-280va), in DB, where one booklet covers both, and presumably in all the intervening stages as well. In DB, some other pairs of counties share a booklet (YoLi, GlWo, ShCh, DnCo), but there is nothing to suggest that the linkage originated during the fieldwork phase, or at any stage prior to DB itself. My guess would be that these pairings were created by the DB scribe; it will do no harm (as far as I can see) to assume that this guess is right. On that view, D would consist of 32 booklets covering 33 counties – one county in each booklet except for D-NmDy.

In the previous chapter, I tried to open up a new perspective on the DB booklets by working out the order in which they were written.3 For the purposes of the present chapter, the sequence does not need to be exactly known. If it can be agreed (as I hope it can) that the order in which the booklets were eventually bound up was very different from their original order, that will be enough. Why the order differs is one of the questions which I hope to answer in this chapter; but my first concern here is with the D booklets, the source from which the DB text was derived. From the evidence of the three surviving booklets, and from the proxy evidence of the C and DB booklets, I begin by trying to form some concrete idea of the size and shape of D. Once that objective has been achieved, with (as it seems to me) an adequate 1

The scribes who worked on D were presumably aware that these booklets were all intended to form parts of a single whole. If that can be taken for granted, it can also be assumed that some measure of uniformity was aimed for and

This statement is almost but not absolutely true (below, note 19).

2 The organization of C is surprising and calls for some explanation; I deal with that question in chapters 4–5. 3 I continue to ignore one gathering (fos. 373–82), the contents of which are anomalous (though the format is the same). It is not clear to me what larger purpose the scribe had in mind when he wrote these pages.

4

It is wrong, no doubt, for the volumes to be shown standing upright: they ought to be flat on their backs. I claim artistic licence.

25

The survey of the whole of England C

a collection of booklets covering Devon, Cornwall and Somerset (C-DnCoSo = Exeter Cathedral Library 3500, fos. 83–494), together with part of a similar collection covering Wiltshire and Dorset (C-WiDo = fos. 25–62)

D

three booklets covering one county each, Essex (D-Ex = PRO, E 31/1, fos. 1–108), Norfolk (D-Nk = fos. 109–280), and Suffolk (DSk = fos. 281–451)

DB

25 booklets covering one or two counties each, 30 counties in total (PRO, E 31/2, fos. 0–372)

leaf of DB (or, equivalently, the number of pages of D corresponding to one page of DB).5 The DB scribe would have understood this notion (or something equivalent to it). He had to know the current value of the compression factor in order to be able to estimate the sizes of the quires that he would need. No direct comparison is possible: there is no overlap between D and DB, no county for which both D and DB survive. For three counties, however, we have the opportunity of comparing C and DB, and that offers us a point to start from. The collection of booklets which I call C-DnCoSo comprises 412 leaves: in DB the same three counties occupy 40 leaves, 14 for DB-So plus 26 for DB-DnCo.6 That gives a compression factor, between C and DB, for these three counties, of 10.3, which, if we have any sense, we will round off without hesitation. Roughly speaking, one page of DB covers as much ground as ten pages of C.

Table 4. Surviving original records of the survey.

achieved. In the case of DB, though the basic design was fixed at the start and could not be altered afterwards, a great deal of experimentation occurred while the work was in progress. In many small and some quite large particulars, the scribe kept changing his plan as he went along. (This is what makes it possible to work out a seriation for the DB booklets.) In the case of D, experimentation of this kind is unlikely to have occurred to any significant extent. The surviving D booklets are generally so similar to the surviving C booklets that the scribes can hardly have found much room for variation, even if they thought of looking for it. To a large extent, the C-like design would have been fixed in advance. The size of the page, the ruling, the disposition of the text – features such as these are unlikely to have varied more than slightly between one county and another.

This calculation is easily made, but does not tell us what we really want to know. There are two complications. First, C-DnCoSo is very wasteful of space.7 There are 75 blank pages here (including 21 leaves which are blank on both sides). By analogy with D-Ex and its companions, it is probable that the D booklets for these three counties would have been less extravagant than C in their consumption of parchment; it is possible also that the number of lines per page would have been increased, from 20 (the usual number in C-DnCoSo) to 24 (the usual number in D-Ex). For these reasons, I would think it safe to assume that the text had already been compressed to a significant extent in passing from C to D, before being compressed again, much more drastically, in passing from D to DB. Correspondingly, the number we want to know, the compression factor between D and DB, would be less than the number we can calculate, the compression factor between C and DB – distinctly less than 10, perhaps as little as 8.

In one respect, however, the D booklets would have varied greatly. Because the quantity of information collected was very much greater for some counties than for others, some booklets must have been very much thicker than others. (Thickness I take to be measured by the number of leaves, or – less accurately but perhaps more palpably – by the number of gatherings.) Without doubt, the thickest booklets were several times as thick as the thinnest ones. Of the three surviving booklets, the thinnest is D-Ex, with 108 leaves, and the thickest (just) is D-Nk, with 172 leaves. The average, for these three, is 150 leaves. If we thought it fair to assume that this average would hold true for all 32 booklets, we might estimate that D as a whole comprised 4800 leaves, more or less. But that seems certain to be an overestimate, because the three survivors are not properly representative. There are many counties which are likely to have generated a thinner booklet than D-Ex, only a few which are likely to have generated a thicker booklet than D-Nk.

Second, the DB scribe increased the compression factor, 5

Stray remarks by Galbraith suggest that in his view the compression factor was something over 10 (Galbraith 1961, p. 32), perhaps as much as 15 (1961, p. 205). These seem to be just guesses, and I feel sure that they err on the high side. Assuming a compression factor of 15 is the same as assuming that D consisted of about 6000 leaves (373 times 15, for the booklets represented by DB, plus 451, for the booklets that survive). At that rate, D-Nk and D-Sk would each represent about one fortieth of the whole, D-Ex about one sixtieth. Since these fractions are plainly too small, the compression factor must have been less than 15 – considerably less, I would say.

6

In two places, C-DnCoSo’s text is interrupted by the loss of a leaf (after fo. 414 and after fo. 421). Collation with DB suggests that six paragraphs have been lost in the first place, one paragraph in the second; so perhaps we should add two or three to the number of leaves in C. It also turns out that there are four chapters in DB-DnCo for which no corresponding C booklet survives. These chapters add up to roughly one page of DB; so perhaps we should deduct half a leaf from the DB total to allow for their existence. (Dividing 415 by 39.5 would give a compression factor of 10.5.) But these adjustments are too small to matter. Above all we want to avoid succumbing to delusions of accuracy.

It is possible, I think, for thickness to be estimated rather more satisfactorily than that, for the individual booklets, and so for D as a whole. The basic idea is to try to arrive at some estimate of what I propose to call the compression factor – the number of leaves of D corresponding to one

7

The reason for that lies in the nature of the compilation process, a topic which I postpone to the following chapter.

26

Reconstruction of the D volumes

Period 1 D

D

D

D

D

D

D

D

D

D

1

2

3

4

5

DB

DB

D

DB

(DB)

Period 2 D 6

Period 3 D

D

D

D

D

D

6

1

2

3

4

5

Period 4 D 6

Figure 8. Changing perceptions of the documentation resulting from the survey. quite deliberately, as his work advanced. That fact has been suspected by previous commentators; when the booklets are restored to the order in which they were written, it becomes instantly obvious. As it happens, the DB booklets which can be compared directly with C, DB-So and DB-DnCo, were the very last booklets to be written. (I doubt whether this is due to chance: there is probably some reason for it.) By this time, the text was being very tightly compressed, with more than 60 lines in each column. In the earliest booklets, by contrast, in DB-YoLi, DB-NmDy and DB-Hu, the text is much more generously spaced out. There are 44 lines per column (less than that, on the earliest leaves of all); blank lines and blank spaces occur quite frequently; the lines being further apart, the script is larger too, with the number of words per line much less than in DB-So and DB-DnCo.8 All told, the quantity of text per page is barely half as much, in the earliest booklets, as it is in the latest ones; and that difference is (or seems to be) solely due to changes in procedure on the part of the DB scribe.

one. To measure the trend precisely would involve a vast amount of work, and no one should think of wasting their time on this until the text becomes available (unextended) in electronic form. Provisionally, it seems reasonable to assume a compression factor varying from something like 5 for the earliest booklets to something like 8 for the latest ones. On that basis I arrive at the results listed in Table 5; the three surviving booklets are included there, recognizable by the blanks in columns 3–4.9 The estimated number of leaves (column 5) is the number of leaves in DB multiplied by the appropriate compression factor; the estimated number of gatherings (column 6) is the estimated number of leaves divided by 8.10 The D booklets cannot be put into any order which is intrinsically correct, so I list them here alphabetically. An independent calculation encourages me to think that these estimates, though crude, are not too far from the truth. Working on the records of the survey relating to Kent (which at one time I thought was the only county that I should be dealing with), I tried estimating the compression factor for DB-Ke by assuming that D-Ke bore a fairly

In consequence, we have to assume that the compression factor increased appreciably (and fairly steadily, it seems) during the compilation of DB. The increase in the number of lines per page, from fewer than 40 to more than 60, is the most obvious sign of this tendency, but not the only

9

In setting the value of the compression factor (column 4), I use the same ad hoc classification as in the previous chapter (Table 2). The value is set at 5 for aspect 1, 6 for aspects 2–3, 7 for aspects 4–5, and 8 for aspect 6.

8

10 In D-Ex and its companions, most of the gatherings are regular quires of eight, and the rest are larger in some cases, smaller in others. The upshot is that the average size of a gathering is very close to 8.

The degree of abbreviation might also be a factor; but this appears to be fairly uniform, except in the early part of DB-YoLi, where the scribe was still feeling his way.

27

The survey of the whole of England booklets in D Bd Be Bu Ca Ch Co Dn Do Ex Gl Ha He Ht Hu Ke Le Li Mx Nk Nn NmDy Ox Sh So St Sk Sy Sx Wa Wi Wo Yo

leaves in DB

compression factor

10 8 11 14 10 6 20 11

6 6 6 6 7 8 8 7

9 19 10 11 6 16 8 37 6

7 6 7 6 5 6 7 5 6

11 25 8 10 14 6

7 5 7 7 8 7

7 14 8 11 8 39

6 6 7 7 7 5

Bedfordshire Berkshire Buckinghamshire Cambridgeshire Cheshire Cornwall Devon Dorset Essex Gloucestershire Hampshire Herefordshire Hertfordshire Huntingdonshire Kent Leicestershire Lincolnshire Middlesex Norfolk Northamptonshire Nottinghamshire–Derbyshire Oxfordshire Shropshire Somerset Staffordshire Suffolk Surrey Sussex Warwickshire Wiltshire Worcestershire Yorkshire totals

373

leaves in D

quires in D

60 48 66 84 70 48 160 77 108 63 114 70 66 30 96 56 185 36 172 77 125 56 70 112 42 171 42 84 56 77 56 195

8 6 8 11 9 6 20 10 14 8 14 9 8 4 12 7 23 5 22 10 16 7 9 14 5 21 5 11 7 10 7 24

2772

350

Table 5. Estimated sizes of the D booklets. (In this and the subsequent tables, all estimated numbers are italicized.) close resemblance to the surviving D booklets. Taking DEx as a proxy for D-Ke, and making myself a proxy for the DB scribe, I chose some specimen entries from the source text and rewrote them as the DB scribe would have done, omitting some categories of information and recasting the rest into the formulas which are normal for DB-Ke. Next I designed a template which, when a deformated copy of DB-Ke was fed into it, would divide the text automatically into columns and pages of roughly the right size; and then I fed the simulated entries into the same template, to see how much space they would take up here, in proportion to the amount of space they take up in D. The results are too subjective to be worth describing in detail, but the conclusion I came to was that rather more than 6 pages of D-Ke, edited and reformated in this way, would be needed to generate one page of DB-Ke.

in size, amounting altogether to roughly 2800 leaves (or roughly 350 gatherings). These estimates do not pretend to be exact; but I think they give a fair idea of the relative sizes of the individual booklets, and of the overall size of the collection.11 Reversing the argument, we can estimate how large DB would have been, if it had been completed (or had survived complete). If the scribe had written booklets for Ex, Nk and Sk, compressing the text of D to the same degree as he did for the adjoining counties (except Hu), our estimate is that these booklets would have consisted of 18, 29 and 29 leaves respectively, making 76 extra leaves. In its 11

My guess would be that these estimates are on the low side. Increasing every compression factor by 0.5 would increase the total to 2964 leaves (369 gatherings), and possibly that may be closer to the mark. At least in round numbers, I think we would be safe in saying that D consisted of roughly 3000 leaves: it seems certain that 4000 would be an overestimate, since that would imply an average compression factor of 9.5, and more or less equally certain that 2000 would be an underestimate, since that would imply an average compression factor of 4.2.

To sum up, it seems that we should visualize D, in its unbound state, as a collection of 32 booklets, varying greatly

28

Reconstruction of the D volumes two rather large blocks.14 A slight jump, from Cheshire to Nottinghamshire, can be taken to mark the dividing line between them. The blocks then consist of 83 and 101 leaves respectively; because the compression factor is relatively low, the D volume corresponding to the latter block would be rather the thinner of the two. That gives us a reconstruction of D bound in six volumes, as set out in Table 6. If it were our sole purpose to equalize the size of the volumes, regardless of which counties go with which, we could do a very much better job of it than this. If we want the arrangement to make some geographical sense, however, we have to be prepared to compromise; and that means that we have to let the volumes vary in size.

finished state, therefore, DB would have had roughly 450 leaves.

2 Three of the booklets belonging to the D collection were put together and bound up to make a book. We know this because the book survives, in the shape of D-ExNkSk. The survival of this volume is sure to be linked, somehow or other, with the fact that these three counties are absent from DB. Given these facts, the presumption is that the other booklets were also bound, at the same time, into volumes resembling D-ExNkSk, and that the failure of these volumes to survive is somehow linked with the fact that the counties they covered are present in DB.

What we are doing here amounts to reenacting the decisions which had to be made by the man who prepared the D booklets for the binder. Before he started, he could see how large the entire collection was; he had a rough idea how large a volume could be before it became too large; and on that basis he set about sorting the booklets into a suitable number of stacks of suitable size, making it a rule that the counties which would share a volume should be geographically connected (i.e. that each should be contiguous with at least one of the others). By trial and error, he worked out a satisfactory arrangement – not necessarily the best possible arrangement (whatever that might mean), but one which seemed good enough to him – and the stacks of booklets were then sent off to be bound.15 There is (as far as I can see) no more to it than that. The arrangement originated at this time, for this practical purpose; except accidentally, it did not coincide with any division that may have existed at an earlier stage.

Suppose for the sake of argument that this is true. The D booklets were divided up into batches of suitable size, and the batches were then each put into order and sent off to be bound. One of the resulting volumes survives; the rest have all been lost. Can we find, in DB, in the arrangement eventually imposed on this collection of booklets,12 any trace of the existence of such volume-sized batches of D booklets? There is no particular reason why we ought to be able to do so, but we might be able to do so, if the circumstances were favourable. The surviving volume, if it was represented in DB, would be represented by a sequence of booklets adding up to roughly 70–80 leaves (possibly more or less than that, depending on the compression factor); so the blocks that we are looking for are blocks of that sort of size.13 One is instantly obvious: its beginning is marked by a jump from Cornwall to Middlesex, its end by a jump from Herefordshire to Cambridgeshire. This block consists of six booklets, starting with DB-Mx and finishing with DB-He, adding up to 63 leaves. The corresponding volume of D would (hypothetically) consist of 413 leaves, a little less than D-ExNkSk. To the south (so to speak) of this first block there are nine DB booklets covering ten counties. If we split them down the middle, between Berkshire and Wiltshire, we have blocks consisting of 64 leaves and 62 leaves respectively; the D volumes would not be so nearly equal, however, because the compression factor is greater for one than the other. To the north of the first block there are ten DB booklets covering 13 counties. They add up to 184 leaves, and the corresponding booklets of D add up to 1050 leaves; so it seems that we should be looking for

Five of the six original volumes are lost, except to the extent that they have left their ghostly outlines in the arrangement of DB. (For this to be true, there must have been some reason for its being true. I return to this point in due course.) One survives, in the shape of D-ExNkSk, and there is no reason not to suppose that this volume was typical of the whole set. In each case, once the booklets had been assembled into their intended sequence, somebody added a colophon, written in red capitals, at the end of what was to be the final booklet. It consisted of one monumental sentence:

14 The only alternative would be to look for three rather small blocks. Yo and Li would form one; the remaining booklets would have to form two blocks, but we should have no clue (so far as I can see) where to make the split between them. Wherever we make it, however, the thinner volume will not have more than 307 leaves, and that, we may suspect, would not have been thought thick enough.

12

When they were bound, the DB booklets were frozen into the following sequence: Ke, Sx, Sy, Ha, Be, Wi, Do, So, DnCo, Mx, Ht, Bu, Ox, GlWo, He, Ca, Hu, Bd, Nn, Le, Wa, St, ShCh, Dy (= NmDy quire 3), Nm (= NmDy quires 1–2), YoLi.

15 If the binder had been asked for his advice, perhaps he might have suggested making eight volumes of about 350 leaves each. One volume would consist of Nk and Sk; Yo and Li would make a second, rather thicker, volume; Co, Dn and So would make a third, rather thinner, volume; without much trouble the remaining booklets could be arranged into fi ve more volumes of similar size, each containing a group of contiguous counties. But that is not what happened. It was decided, we know, to include the Ex booklet in the same volume as Nk and Sk. It seems that the man in charge wanted the bound volumes to be as thick and heavy as the binder could make them: he wanted them to look impressive.

13

We cannot assume that the surviving volume was roughly of average size: for all we know, it may have been either the thickest or the thinnest of the set. Nevertheless, from the fact that the binder included all three of these booklets, not just two of them, it does seem fair to infer that a volume of around 300 leaves was thought to be too thin, and that a volume of around 450 leaves was not thought to be too thick.

29

The survey of the whole of England leaves in DB

volumes of D

booklets in D

leaves in D

quires in D

ExNkSk MxHtBuOxGlWoHe KeSxSyHaBe WiDoSoDnCo NmDyYoLi CaHuBdNnLeWaStShCh

63 64 62 101 83

3 7 5 5 3 9

451 413 384 474 505 545

57 52 48 60 63 70

totals

373

32

2772

350

Table 6. The bound volumes of D reconstructed. rowed, it would leave a space on the shelf; when a volume was returned, it would be put into a space on the shelf; and normally that space would be the same space. But accidental transpositions could easily occur – would be quite likely to occur, for instance, if two volumes had been borrowed at the same time.17 Furthermore, even if the sequence stayed the same, the sense of it would be ambiguous, depending on whether one ran one’s eye along the shelf from left to right or from right to left. What looked like the first volume to one person might look like the last volume to somebody else, and vice versa.18 At this stage, there was no right order – no order which was uniquely better than any alternative order, no order which would need to be restored if it ever got disrupted. There was only the order existing at a given moment, objective so far as it was represented by the sequence of volumes on the shelf, subjective so far as one had to choose which end of the sequence to start from. Eventually (as will appear) it did become necessary for the order of the volumes to be settled, but that was not true at first.

ANNO MILLESIMO OCTOGESIMO SEXTO AB INCARNATIONE DOMINI, VIGESIMO VERO REGNI WILLELMI, FACTA EST ISTA DESCRIPTIO, NON SOLVM PER HOS . . . COMITATVS SED ETIAM PER ALIOS.

Presumably this was the very last thing that was done before the batch of booklets in question was delivered to the binder. In the case of D-ExNkSk, the colophon appears at the end of the Sk booklet: the scribe who wrote it (not one of the scribes who had participated in the writing of these booklets) supplied the number of counties appropriate to this volume, which is TRES (in the other volumes it would vary from QVATTVOR to NOVEM). The man who composed this sentence, and who wrote it or had it written into every volume of D, was addressing himself to posterity.16 He assumed that D would be kept for all time. In that light, it seems fairly certain that these words were written while the king was still alive. If the scribe had been writing after September 1087, he would (we may think) have wanted to make it clear to future generations that the king of whom he spoke was Willelm I, not Willelm II; his silence suggests that it had not yet became necessary for that distinction to be made. For that reason, and in the absence of any evidence to the contrary, I think we may assume that the six volumes were put together and bound during the lifetime of Willelm I. (It is highly unlikely that the king ever saw them, however.) They had already become a book (there was nothing unfamiliar about the notion that a single book might consist of multiple volumes); the word liber was already appropriate, if anyone chose to use it.

3 Having followed the history of the D booklets through to the moment when they were bound and shelved, we need to track back. Perhaps as soon as the first finished D booklets began coming off the assembly line, somebody set about composing an edited and somewhat shortened version of the survey, using a format which would allow the whole text to be contained in a single volume. His work survives in the shape of the DB booklets, which each correspond with one or two D booklets. The entire manuscript, every line of it, was written by this one man.19 His intention was (so

There was no particular reason, during period 1 (which begins when D is bound), for the volumes to be kept in any definite sequence. The few people who ever had access to them would have learnt to recognize them by their individual properties – by small differences in size and shape or colour and texture, and by other accidental features. It was necessary for them to know which counties were in which volume, and to be able to tell at a glance which volume was which; but that is all. Up to a point, the shelved volumes would keep themselves in order. When a volume was bor-

17

Unless the volumes were chained in place; and I am not suggesting that.

18

In fact, it does seem possible that the order got reversed. The DB scribe, we may guess, would have regarded D-NmDyYoLi as the first volume, not the last, and D-WiDoSoDnCo as the last volume, not the second. 19 Except for one four-line entry, almost but not quite at the end of DB-Be (63vb42–5), sandwiched between entries written by the DB scribe. The fact that this entry was written by somebody else was first pointed out by Gullick (1987, p. 103). There is, I think, no doubt but that Gullick was right: compare rege here with rege three lines above and four lines below; note also the F, the sign for -us after b, and the 7-shaped sign for et. The wording is incongruous too. For instance, where this scribe has writen Ibi i car’ in d’nio, the DB scribe would have written In d’nio e’ i car’. This entry proves the existence of a second scribe, active while the DB scribe

16 ((But posterity has sometimes managed to misunderstand him. Perhaps he ought to have said HAEC instead of ISTA and CAETEROS instead of ALIOS. Nevertheless, his meaning is plain enough.))

30

Reconstruction of the D volumes Perhaps there was never a time when they were neglected.22 In Gullick’s view, hundreds of alterations have been made in DB which are not the work of the DB scribe himself. Almost all of them (but not the two mentioned in the following paragraph) are thought by him to have been made by a single corrector. I am not sure how far to rely on that. There are only thirteen places where the corrections amount to six words or more (Gullick 1987, p. 103), and of these only two where they amount to two lines or more. (There is also a four-line entry in column 63vb (above, note 19), integral with the text, which Gullick is inclined to attribute to the same scribe.) Whether there is enough evidence here to characterize an individual hand seems (to me) doubtful; 23 but I do not think it intrinsically unlikely that the work was nearly all done by just one man. The presumption is that at any given moment only one or two people would have been in a position to make corrections and additions in DB; and in these circumstances a single hard-working or long-lived clerk might make the sort of impression on the text which Gullick believes his correcting scribe to have made.

I suppose) that the finished volume, suitably bound, should be presented to the king. The purpose of his task, as he understood it, was to put the results of the survey into a portable form, so that the king and his household officials could consult them, whenever they needed to do so. That intention failed to be achieved. The collection of DB booklets shows some obvious signs of not having been completed to the scribe’s satisfaction. In DB-Do the last entry breaks off unfinished (85ra), perhaps because the scribe was unhappy with what he had written and intended to erase some part of it before trying again. In several booklets, entries inserted as afterthoughts were left unrubricated. In DB-Mx there is no description of London, though there is blank space which looks as if it may have been reserved for that purpose; the same is true for DB-Ha, which lacks a description of Winchester but has the space for it.20 From indications such as these, it seems clear that the scribe, for one reason or another, ceased work before reaching his objective. These facts are well known, and it is, I think, fairly generally agreed what explanation is the likeliest – that the work was interrupted by the death of the king (more precisely, by the arrival of the news of his death). But that is not the only possibility. The scribe may have been promoted; he may have been dismissed; or perhaps it was he who died.

However many scribes are involved, the importance of these corrections lies in the fact that they imply the existence of some parallel record which was considered to be more reliable than DB. This is not true for every single one of them. In column 37vb, for instance, somebody has erased and rewritten three names in the index of tenants for Hampshire;24 but that alteration could have been made, and probably was made, to bring the index into line with the main text. The scribe concerned did not need to refer to any other source, only to DB itself. In column 121ra, on the other hand, where somebody (somebody else) has erased and rewritten half a line of text – Ibi est una hida qu˛e nunquam geld’ (DB-Co-121ra) – he must have got his information from somewhere; and he must have thought that it was more to be trusted than whatever the DB scribe had written here.25 (He must also have thought that it served some useful purpose to make corrections in DB.) We can be sure that there was a time when the clerks who had access to DB had access to D as well; we can be sure that there was a

In any case, it is not enough to explain why the work stopped; we need also to explain why it never started again. Even if the original scribe was no longer available, the task could have been finished by somebody else. At any stage, if the incentive had existed, DB could have been continued and brought to completion. (The missing rubrication could be added tomorrow, if this was thought to be a sensible idea.) Given time to familiarize himself with the DB scribe’s way of working, as it was exemplified by the existing booklets, a good scribe could have done the job well; a mediocre scribe could have done it after a fashion.21 If the new king had ordered the job to be done, somebody could have done it. But nobody ever did. Perhaps the new king was never asked whether he wanted DB to be completed; perhaps he was asked and said no. Nevertheless, the booklets were kept, because there was no particular reason for throwing them away. The DB scribe’s friends and colleagues – and anyone who could appreciate how much work had been invested in them – would feel reluctant to discard them; besides, the king might possibly change his mind. So inertia prevailed, and the DB booklets survived.

22 ((I have rearranged and revised the following paragraphs in the light of some detailed comments from Caroline Thorn (for which I am very grateful), and also in the light of my reading of Thorn and Thorn (2001). But the gist remains the same.)) 23

was active; but I do not think it can be said to prove that the two of them were working in partnership. On the contrary, it was obviously the DB scribe’s intention that he should write the whole manuscript himself; and I would guess that he was extremely annoyed when he discovered (as he did) that somebody else had added an entry while his back was turned.

((Caroline Thorn assures me that the hand is very distinctive and instantly recognizable in the original. (She asks me to point out that the passage reproduced as Thorn and Thorn (2001, ill. 22) is the work of the main scribe: it is not the passage which she was intending to show here.) Though I am willing to be convinced, I am not sure of anything yet except that the passages in question are not the DB scribe’s work. It is not obvious to me that the hand which wrote the intrusive paragraph in DB-Be (Gullick 1987, fig. 14.11 (b)) is the same as the hand which made some additions in DB-Ca (Thorn and Thorn 2001, ill. 23, Gullick 1987, fig. 14.11 (c)) or as the hand or hands which inserted words and phrases between the lines in DB-Ke (Thorn and Thorn 2001, ill. 28), DB-Wi (ill. 25) and elsewhere.))

20

24

The suggestion that space was left for accounts of London and Winchester – which, however, failed to be supplied – seems first to have been made by Maitland (1897, p. 178).

25

((This correction is reproduced by Thorn and Thorn (2001, ill. 12.))

As was noted by Thorn and Thorn (1979), this fact is not reported in CDnCoSo; so we can rule out the possibility that it might have come from there. That the hand is not the DB scribe’s was pointed out by Gullick (1987, p. 104). ((This correction too is reproduced by Thorn and Thorn (2001, ill. 13).))

21

As is proved by the evidence cited in note 19. It cannot be said that the main scribe was ‘irreplaceable’ (Chaplais 1987, p. 77), only that in the event he was not replaced.

31

The survey of the whole of England time when D was considered more authoritative than DB; so we have a ready-made explanation for these corrections. They are the work of scribes who were altering DB to bring it into line with D.

D. Second, the DB scribe had done his best to make things easier for his readers, using capital letters and red ink to accentuate the key words. He had put a great deal of thought and effort into this. The clerks who discovered this new use for DB were not the readers whom he had been expecting to have – but they were the ones who benefited. From time to time, they were told to find out what D might have to say about some topic; and what they did was to search for the information in DB first, where it would be relatively easy to find, and then turn to the corresponding section of D. If there was some discrepancy, the clerk concerned might see it; and then he might think of making the appropriate adjustment in DB, so that the error did not cause confusion in the future. In short, he would have both the opportunity and the motive for making a correction in DB.

The evidence is hard to interpret, however, because we cannot repeat the comparison for ourselves. Nowhere is it possible for us to draw up a list of the discrepancies between D and DB, checking to see how many were put right (by the DB scribe or by a subsequent corrector) and how many were left uncorrected. In counties where some independent evidence exists (i.e. evidence which is approximately contemporary with but not derived from DB), we may be able to identify some defects in the DB text – in Kent, for instance, an important manor belonging to the archbishop is missing26 – but we cannot be sure whose fault they were: they may have been inherited by DB from D (and possibly by D from C).

Once it began to be a habit for D and DB to be used together in this way, it would become important for the sequence of counties to be made the same and kept the same in both. Whatever order the volumes of D were arranged in at this moment became the permanent order. It is not likely that the volumes were actually numbered; but we are now free to number them ourselves, if we wish, without anachronism (Table 7).28 From this moment onwards, that order had to be preserved. As much as before, it was constantly at risk of becoming disrupted. If that happened now, however, somebody would notice, click his teeth, and put things right again. Correspondingly, this is when the DB booklets acquired their present order. As I understand it, the order was dictated by the preexisting order of the D booklets: within each volume, the order that had become fixed when the D booklets were bound; between volumes, the accidental order into which by this time the D volumes had got themselves arranged. On that basis the DB booklets were sorted into order and then sent off to be bound.29 When DB came back from the bindery, it was put on the shelf in some convenient place, so that it could be used alongside D. Period 2 begins at this moment.30

To some extent, perhaps to a large extent, we are dealing with spot corrections.27 The scribe concerned was not looking for mistakes: he just happened to come across one. For some reason of his own, he found himself comparing an entry in D with the corresponding entry in DB. Doing this, he noticed a discrepancy; having noticed it, he decided to put it right. In these circumstances, a conscientious scribe might think that he ought to look of some of the surrounding entries, to see whether similar discrepancies existed there; and thus he might be led into making a series of corrections, until he stopped finding mistakes or lost interest. Small spurts of activity, each prompted by the accidental discovery of some error in DB, might add up over time to produce the observable result. At least it is clear that there ensued some period of time, after the departure of the DB scribe, during which D and DB continued being compared with one another. Sometimes it happened that somebody had both of them in front of him, each open to the same entry. Sometimes a discrepancy was noticed; and sometimes (perhaps not every time) a correction was made in DB – in DB, not in D, because D was assumed to be right. But what would be the point of making sporadic corrections in a manuscript which had never even been finished?

That was the plan; but the plan was not executed fully. One of the D volumes – the one that we have called D6 – was not included in the scheme. In the long run, the consequence was that this volume survived while the other five did not, 28 The numbering is arbitrary, however, in one respect. Rather than putting D-ExNkSk at the end, we might put it at the beginning.

The explanation is, I think, that an unintended use had been discovered for the DB booklets. Those clerks who had access to the volumes of D and were sometimes required to consult them began to use DB as a finding aid. It is obvious enough why this would be a good idea. First, the DB text is more compact: by scanning a single page of DB, one saves oneself the trouble of turning over several leaves in

29 With regard to D-NmDy, it is possible that the beginning of the Dy text coincided with the beginning of a new quire in D, as it does in DB. In other words, it is possible that the transposition of these two counties (above, note 12) originated with the man who arranged the D booklets for the binder, not with the man who did the same for the DB booklets. He, on this view, would have had no choice but to perpetuate the error (short of having the D volume taken apart and rebound). I see no way of deciding whether this was so or not.

26 The manor in question is Teynham, described in a contemporary account of the archbishop’s lands (below, p. 111).

30

For this argument to hold, it has to be permissible to think that the firstever binding of DB occurred at a different and somewhat later date than the first-ever binding of D-ExNkSk. Traces of the original bindings are slight (Gullick 1987, pp. 107, 111), but it seems fair to say that there are some technical differences, as well as some similarities. The crucial fact is the obvious one – the absence from DB of a colophon matching the one that occurs at the end of D-ExNkSk.

27

((Two more spot corrections are reproduced by Thorn and Thorn (2001, ills. 14–15), each, they think, the work of a scribe who occurs only once. I am not sure that the second (DB-Le-235ra) might not be by the same hand which made some additions in DB-Ca (above, note 23), but the first (DB-Co-121rb) is very distinctive.))

32

Reconstruction of the D volumes volumes D1 D2 D3 D4 D5 D6

booklets

KeSxSyHaBe WiDoSoDnCo MxHtBuOxGlWoHe CaHuBdNnLeWaStShCh NmDyYoLi ExNkSk totals

leaves

quires

5 5 7 9 3 3

384 474 413 545 505 451

48 60 52 70 63 57

32

2772

350

Table 7. The bound volumes of D put into the sequence which determined the binding sequence for the DB booklets. and we have to be thankful for that. But we still have to ask how it came about that D6 began to be treated as a special case, and there is no easy answer to that question.

all, and that the information which was only to be found in D (because it had been omitted by the DB scribe) no longer had any value. So the clerks started using DB by itself, not as a guide to D but as a primary source; and this was the beginning of period 3. Figuratively at least, we may think of the five redundant volumes as having been pushed to the far end of the shelf. By this stage they had practically ceased to exist; but some further length of time may have had to elapse before (so to speak) they fell off the end of the shelf, and were finally discarded or destroyed. Period 4 began at that moment.31

At the moment when the DB booklets were being prepared for the binder, it was, apparently, impossible to include a sequence of booklets corresponding with D6. It would have been done if it could have been done, we can feel sure enough of that; so the fact that it was not done implies that it could not be done. Beyond that point, we find ourselves lost in a forest of possibilities, with no means of deciding whether one possibility is likelier than another. At one extreme, it is possible that the three missing booklets never existed: we know that the DB scribe fell short of his target, and the absence of booklets for these three counties may be one more sign of that. Perhaps he desisted, when he did, without having started any one of these booklets. At the opposite extreme, it is possible that the booklets were written, as completely as any of the others, but that later these three became separated from the rest (circumstances in which that might have happened are easy enough to imagine) and could not be retrieved when they were needed. Those are the extremes. A middling possibility might look something like this: the DB scribe wrote a complete booklet for Norfolk, half of a booklet for Suffolk (this being the point at which he stopped work), nothing at all for Essex. Then, when the time for binding the booklets arrived, the man responsible decided to restrict himself to the first five volumes of D and leave D6 alone: unable to deal with it completely, he preferred not to deal with it at all. And so we might go on, thinking of new possibilities, yet never knowing whether we are getting closer to the truth or not. The danger is that we may latch onto one explanation, assume it to be the right one (in spite of there being any number of alternative explanations which would fit the facts equally well), and build larger conclusions on top of it. In the end, it is hard to be sure of anything, except that D6 did begin to be treated differently from the other volumes, and that this happened no later than the time when the DB booklets (those that survive) were bound.

Once DB and D6 became mutually juxtaposed and separated from the other volumes of D, a new book came into existence, consisting of an oddly mismatched pair of volumes. Seen in this new setting, DB looks much the more impressive of the two. It is bigger; it is better-looking; and it covers thirty counties while the small volume covers only three. By now, DB is the authoritative text. Five D volumes have been ousted, and the sixth has been reduced to the status of an appendix. The cuckoo has taken charge.

4 In outline, I think, this scenario is evidently right. The only question which seems doubtful to me is what time-scale ought to be envisaged.32 Was the whole sequence of events completed in a matter of months, as conceivably it might have been? Or was it drawn out over several decades, as I am inclined to suppose? The only large block of text inserted into DB by a later hand (an early twelfth-century hand) is a list of certain lands in Yorkshire. It carries the title: ‘Here is the feod of Rotbert de Bruis, which was given after the book of Winchester was 31

And lasted until 1985. A new period began in 1986, when DB and D6 were rebound respectively as two and three volumes.

Sooner or later, D ceased to be consulted. The clerks were allowed, or allowed themselves, to simplify the process. With the passage of time (and perhaps with a turnover of personnel), it came to be understood that the information wanted could always be found in DB, if it could be found at

32 Unexpectedly, a few remarks by Galbraith hint at a protracted timescale. Even after the compilation of DB, ‘it was only slowly that men forgot the local returns [the D texts], and that the summary in Vol. I [DB] . . . acquired its later prestige’ (Galbraith 1961, p. 185). It is not certain when the D texts ceased to exist, only probable that they had disappeared ‘by the end of the twelfth century’ (p. 210).

33

The survey of the whole of England written’.33 Among historians, this has traditionally been read as a reference to DB itself, but nobody reading the sentence for the first time would read it in that way: it sounds like a reference to some other book. If the man who wrote these words had meant this book (the book which you are holding at this moment), he would presumably have said precisely that – hic liber or iste liber. By naming it, he seems to imply that he is speaking of some other book; and that suggests that the ‘Winchester book’ was D, not DB. But that conclusion could easily be argued away. It is possible, for instance, that this block of text originated as a separate schedule, and was only afterwards copied into DB; and in that case the ‘Winchester book’ might be DB, though it also still might be D. Since the name did not become current, there seems to be little point in arguing over the significance of one ambiguous item of evidence. The crucial question is whether a copy of the same text was included in D5 as well; since we have no hope of answering that question, we do best to say as little as possible.

extracts from DB occur at a very early date, in a manuscript compiled (I think) from a batch of documents which had come into the monks’ possession after the death of archbishop Lanfranc in May 1089.37 There are four of these extracts; I print the last and longest of them (Table 8), so that readers can make the comparison. It is tolerably certain that the extracts came from DB and not from D, because the same passages appear in a longer and less well-organized form in the text which I call xAug – the excerpts made for the monks of Saint Augustine’s.38 Thus we can see how the wording was altered by the DB scribe, and then altered further by the scribe who copied these passages for the archbishop. (He was not particularly interested in Saint Augustine’s affairs, still less in Saint Martin’s of Dover’s.) This evidence seems to prove that it was possible for an outsider, if he had the right connections, to gain access to some of the records kept in Winchester, and to copy out the excerpts that he needed. Without knowing the circumstances, however, we cannot say how it was decided – by whom and for what reason – that the archbishop’s scribe should be given DB to work with, rather than D1.

Outside DB, two lines of investigation are open to us. There are, first, a few well-known allusions to D or DB in documents dating from the time of Willelm II and Henric I. From the abbey of Abingdon, for instance, a copy survives of a notification by queen Mathildis, addressed to the bishop of Lincoln and the sheriff and barons of Oxfordshire, reporting the judgment arrived at in a case concerning the abbey’s manor of Lewknor.34 The assembly which tried the case was what would later have been described as the court of the Exchequer, but that word is not used here. As the queen describes it, the case was heard ‘at Winchester in the treasury’ (apud Wintoniam in thesauro); in the absence of the king overseas (not stated but certainly implied), the court is ‘my lord’s and mine’ (curia domini mei et mea). The abbot, she says, has proved his claim by appealing to ‘the book of the treasury’ (per librum de thesauro) – by that means alone, apparently. This evidence seems sufficient to prove that there was a book available in Winchester which a litigant might ask to have consulted, in the expectation that any information found in the book would be accepted by the court as decisive. But a casual allusion like this one cannot be expected to answer the question which for present purposes is crucial: was the book D3 or DB?35

As these examples indicate, neither line of investigation is sure to lead to any firm conclusions. There are too many unknown variables. In favourable circumstances, by close analysis, it may be possible to detect some useful indication – some hint how far DB had advanced along the shelf, at the moment captured by some dated document, or by some collection of excerpts made for an outsider. But I cannot say that I am sanguine about our prospects. Moreover, we run a risk – the risk of becoming distracted from the main issue by peripheral doubts which we cannot hope to resolve.

5 One other well-known item of evidence leads to a more definite conclusion. The tract ‘Concerning the proper procedure of the Exchequer’ supplies a detailed account of the financial business for which this institution was responsible. (References throughout this section are to Johnson’s (1950) edition.) Written by somebody who had participated in the Exchequer’s proceedings over many years, the tract takes the form of a dialogue between two dummy interlocutors, Master and Pupil. The authorship is attributed (reliably, it seems) to the man who served as treasurer during almost the whole of the reign of Henric II, Ricard the son of bishop Nigel of Ely.39 Invaluable though it is, this

The second line of investigation concerns the numerous derivative texts which were made for the use of one particular church.36 At Canterbury, for instance, some edited 33

Hic est feudum Rotberti de Bruis quod fuit datum postquam liber de Wintonia scriptus fuit (DB-332va).

texts (chapters 8–9), I now think it possible to prove that the source text for xAug was B-Ke. Nevertheless, it can still be argued that the extracts under discussion came from DB, not from D, if we are willing to assume that the passages were copied more or less verbatim from B into C into D, and not edited drastically except by the DB scribe. This seems to me a fair assumption to make, as long as one is frank about it.))

34

Historia monasterii de Abingdon, ed. Stevenson 1858, vol. 2, pp. 116– 17. The chronicler supplies some explanatory remarks (pp. 115–16), but these may largely have been deduced from the document. 35

There may even be a third possibility. If the information needed on this occasion was (as I suppose it was) the number of hides for which Lewknor ought to answer, an epitome of D or DB might have sufficed. Sooner or later, such epitomes were certainly made. The question is: how soon?

37 Canterbury Cathedral Library Lit. E 28, published in facsimile by Douglas (1944). I have more to say about this manuscript later (below, p. 111). 38

PRO, E 164/27, fos. 17r–25r, printed by Ballard (1920). I have more to say about xAug later too (below, p. 123).

36

((This paragraph has been revised. At the time when I wrote it, I was assuming, provisionally, that xAug was derived from D-Ke. But that is not true. Having gained some useful experience in dealing with the Ely

39

34

Bishop Nigel (who died in 1169) had served as treasurer under Henric I.

Reconstruction of the D volumes xAug / PRO E 164/27, fo. 18v

DB-Ke-2ra

Lit. E 28, fo. 5va

Et si extranei mercatores ueniebant in ciuitate et accipiebant hospicium in terra sc’e trinitatis uel sc’i augustini tunc habebant . . . sui prepositi. Sed fuit quidem prepositus nomine Brimannus qui per totam terram ciuitatis accepit omnes consuetudines et teloneum iniuste de quo fecerunt monachi clamorem regi Will’o qui precepit ut inde fuisset . . . ante ep’m baiocens’ et ante Hugonem de mundfort et comitem ow et ricardum filium giseiberti qui eum iurare fecerunt ut de hac re uerum diceret quibus post iusiurandum dixit quod uerum toloneum habebat acceptum per totam ciuitatem sed iniuste de terra Sc’e trinitatis uel Sc’i augustini . . .

Quidam prepositus Brumannus nomine T.R.E. cepit consuetudines de extraneis mercatoribus in terra S’ TRINITATIS et S’ Augustini. Qui postea T.R.W. ante archiep’m Lanfranc’ et ep’m baiocensem recognouit se iniuste accepisse, et sacramento facto iurauit quod ips˛e e˛ ccl’˛e suas consuetudines quietas habuer’ R.E. tempore. Et exinde utreque e˛ ccl’˛e in sua terra habuer’ consuetud’ suas, iudicio baronum regis qui placitum tenuer’.

Quidam pr˛epositus bruman nomine, tempore E. regis c˛epit consuetudines de extraneis mercatoribus in terra sc’˛e trinitatis. Qui postea tempore regis W. ante L. archiep’m et ep’m baiocensem recognouit se iniuste eas accepisse, et sacramento iurauit, quod ipsa e˛ cclesia suas consuetudines quietas habere debeat a diebus antiquis. Et exinde in terra sua quietas habet ipsa e˛ ccl’ia consuetudines suas iudicio baronum regis, qui placitum tenuerunt.

Table 8. Three versions of a passage from the survey of Kent. (The ellipses in the first column denote places where at least one word has apparently been omitted.) to be produced when the Treasury was audited.41

tract is in many ways a puzzling piece of work. We do not know why or when exactly it was written; I do not think we understand what motive Ricard had for placing this imaginary conversation in a particular year – the twenty-third year (1177) of the reign of Henric II. Despite these uncertainties, it seems that Ricard is giving us a straightforward description of the Exchequer machinery as it existed in the late 1170s, by and large ignoring any changes (there were certainly some) which happened after that date.40

Twice a year, however, at Easter and Michaelmas, when the Exchequer met, the Receipt performed the function from which it derived its name. These were the occasions when cash was paid into the Treasury, by sheriffs and other people who had been summoned to appear before the Barons of the Exchequer, and the Receipt was the downstairs office where these payments had to be made. (In the Exchequer itself no money ever changed hands: that would have been utterly improper.) All the same staff were present (including the Nightwatchman). The Usher kept the door, admitting one sheriff (or his agent) at a time. A sample of the money which this sheriff had brought with him was weighed, to make sure that it was of acceptable quality; 42 then the Tellers (of whom on these occasions there were four) counted it, penny by penny. When the count had been completed to their satisfaction, the Deputy Chamberlains had a tally cut. They gave half of it to the sheriff, as his proof that the payment had been made, and retained the countertally (which would be sent upstairs to the Exchequer and reunited with the tally a few days later). At the same time the Treasurer’s Clerk recorded the amount in writing (in a roll which would likewise be sent upstairs and consulted a few days later); he also inscribed the corresponding tally. While the Exchequer was in session, the Receipt might have to issue cash as well, against writs of Liberate authorized by the Justiciar (p. 32); but its chief function, during these periods, was to supply the Exchequer with the information that it needed to audit incoming payments.43

To understand the system of administration within which the records of the survey found a niche, we need to start by looking at the work of the king’s treasury in Winchester – more particularly at the work of the Receipt, the Treasury’s front office. Ricard’s dialogue, when it gets down to detail, begins by explaining what this office did (pp. 8–13). The Receipt was open for business all year round, but most of the time its function was to issue money, not receive it. On this subject Ricard has little to say, but the basic procedure is clear. Somebody possessing the right sort of voucher – a writ of Liberate, properly worded and sealed with the king’s seal – presented himself at the Receipt, gained admission from the Usher, and delivered his writ to the officials who had day-to-day charge of the department, the Treasurer’s Clerk and the two Deputy Chamberlains. All being well, one of the Tellers would be instructed to count out the sum of money specified in the writ. The payee departed with his cash, tipping the Usher two pence on his way out (the only employee he would not have had a chance to meet was the Nightwatchman); and the writ was retained by the Receipt,

41

Very little is known about these general audits of the Treasury, which did not fall within the scope of Ricard’s tract. He refers to them twice incidentally (pp. 24–5, 32).

Ricard took office no later than 1160; he was promoted to the bishopric of London in 1189 and died in 1198, retaining the treasurership till within a few months of his death (Richardson 1928, pp. 162–4).

42 In some circumstances it was also necessary for a sample to be assayed; but that need not concern us here.

40

At one point (pp. 51–2), the author forgets (or lets himself forget) that this conversation is supposed to be taking place in 1177 and makes Master tell Pupil about something which happened in the following year; so the tract must be later than 1178. The prologue, where the author speaks in his own persona, is addressed to Henric II (who died in 1189).

43 The officials in charge of the Receipt were entitled to pay themselves a daily bonus for as long as the Exchequer was in session – five pence for the Treasurer’s Clerk, eight pence each for the Deputy Chamberlains

35

The survey of the whole of England The Exchequer had no continuous existence; it maintained no archive of its own. Any records that needed to be kept (kept for all time) were kept in the Treasury, and some of them had to be available, during the Michaelmas session of the Exchequer, wherever that took place. In theory at least, the Exchequer might meet anywhere – in any town where suitable accommodation could be found – at the Barons’ discretion. Wherever the Exchequer decided to meet, the Receipt had to be there too. Each year, the Exchequer’s proceedings were distilled into a new Great Roll. Some rolls survived, probably the whole series, from the reign of Henric I (p. 42), but there cannot have been much occasion for consulting them (except for somebody like Ricard, interested in the history of his department).44 The more recent rolls, however, would certainly be required. Large parts of the text of each new roll were copied exactly, or with some appropriate adjustment, from the roll of the previous year. The Treasurer and his scribe had to have that roll in front of them while the Exchequer was in session; the officers of the Receipt had to make sure that it was there. Another document which the Treasurer needed to consult repeatedly was something called the Record of Farms: this is where he looked up the total due from each sheriff for the farm of his county (p. 125). Again, it was the Treasurer’s Clerk and his colleagues who were responsible for looking after this book (or roll), and for making sure that it was available on the occasions when it was needed.

wherever the meeting was scheduled to take place. (That decision had been made at the end of the previous session, before the summonses were sent out.) When the moment arrived, the box was opened, and the purse was removed, taken upstairs, and handed over to the Chancellor (or to whoever was deputizing for him). In full view, the Chancellor broke open the sealed purse and took out the seal. At the end of the session, the procedure was reversed: the seal went back into its purse, the purse went back into its box, and the box was taken charge of again by the Receipt officials. The seal had an ‘inseparable companion’ in the shape of a book (pp. 62–3). Ricard indulges himself, as he often does, by inventing a Latin name for it: he calls it the Liber iudiciarius. But the name under which it was commonly known, he tells us, was Domesdei, ‘Judgment Day’ (p. 64).45 As Master describes it to Pupil, this book contains ‘a survey of the whole kingdom’, reporting the value of every manor in the time of king Edward as well as in the time of king Willelm, during whose reign it was compiled (p. 14). From what he says later about the scope and organization of this book (pp. 63–4), there cannot be any doubt but that he is talking about DB (though we have to assume that he is simplifying matters, for Pupil’s benefit, by omitting to mention D6). It is certain, therefore, that in the 1170s DB was kept in the Treasury, subject to the same security measures which applied to the royal seal. Most of the time, DB would have been locked up in a strongbox. No one could get at it until the box had been opened; and for that to happen all three Receipt officials had to be present. The Treasurer’s Clerk had to break the seal on the strap, and the Deputy Chamberlains had each to open one of the locks.

By the 1170s, in short, the officers of the Receipt had in their custody a large collection of documents and other paraphernalia which needed to be kept secure, and which sometimes needed to be transported around the country (pp. 61–2). Some of the objects in question were extremely valuable. To ensure their safety, the Treasury had equipped itself with a number of strongboxes. Each of these strongboxes had two locks, the keys to which were held by the Deputy Chamberlains; each key was unique. For greater security, the boxes were strapped up, after being locked, and sealed by the Treasurer’s Clerk, so that none could be opened without the consent of all three responsible officials (p. 9).

Like the seal, this book was regularly required to be available while the Exchequer was in session. As long as the Barons were dealing with the routine financial transactions which are the subject of Ricard’s tract, there would not have been any occasion for DB to be consulted. But the Exchequer also functioned as a court of law, presided over by the Justiciar. Ricard had some thought of writing a sequel, a dialogue discussing the judicial aspect of the Exchequer’s business (pp. 126–7), but he did not act on this idea, as far as we know. In the tract which he did write, there is only one passage which conveys some sense of the significance attached to DB at the time. What Master tells Pupil – passing on what he says he was told by bishop Henric of Winchester (who died in 1171) – is that the information collected by king Willelm’s commissioners was recorded in a book ‘so that everybody should be content with his own property and not get away with usurping anybody else’s’, ut uidelicet quilibet iure suo contentus alienum non usurpet impune (p. 63). As a statement of the purpose of the survey

Of the objects kept in the Treasury, the most important was the great seal – the duplicate royal seal used only during sessions of the Exchequer, and only for Exchequer business (p. 19). The official responsible to the king for the safety of this seal was the Chancellor. When the Exchequer was not in session (i.e. nearly all the time), the seal was kept in a purse sealed with the Chancellor’s seal; and the purse was kept in one of the strongboxes (the strongest of all, no doubt) looked after by the officers of the Receipt. As Easter or Michaelmas approached, the strongbox was moved to (p. 13). The Usher got nothing (Ricard was not sure why), but the Tellers got three pence each, the Nightwatchman got one penny, and a halfpenny was allowed for a nightlight. (If the Exchequer met in Winchester, however, the Tellers got only two pence, ‘because that is where they are usually recruited from’.)

45

The allusion is obscure, and Master has some difficulty explaining to Pupil why the name is suitable. Books are mentioned frequently in Revelations, but none of the references seems particularly apposite. There is a line in Daniel 7:10 which might perhaps have been quoted by a judge who fancied himself a wit: Iudicium sedit, et libri aperti sunt, ‘the judgment was set, and the books were opened’.

44

But Ricard does mention one departmental crisis during which one of these rolls was produced in evidence (p. 58).

36

Reconstruction of the D volumes that will not do; but it may be a fair description of the use to which DB was being put in the reign of Henric II. Especially in the early years of the reign, there must have been many conflicting claims, arising out of the confusion of the civil war, which demanded adjudication. Ideally, no doubt, what the Justiciar and his colleagues would have liked to have at their disposal was a detailed record of who held what on the day when the first king Henric was alive and dead; in the absence of that, DB had to serve as a substitute. Nobody was allowed to suggest that DB had got things wrong: whatever DB said was assumed to be true, and both parties were required to argue from that premise.46 This, it seems, was the origin of the legal fiction that DB was infallible, and the reason why the book began to be kept, like the royal seal, under the tightest security that the Treasury could devise.

From the 1170s onwards, if historians see some benefit in using the name ‘Domesday Book’, there cannot be any strong objection to their doing so. The name did exist by then, though only as a nickname (it does not begin to occur in formal records till another hundred years later). Further back than that, nothing is clear. We do not know when the name was invented; we do not know when DB and D6 began to be regarded in such a way that the name, once somebody had thought of it, seemed apt. For some length of time (period 2), the D volumes must still have been regarded as a set and still valued more highly than DB. The problem is that we do not know how long this period lasted. One day, perhaps, we shall know. The alterations and additions made in DB by other hands, so far as it is fair to assume that they derive from D, may possibly supply some clue; there are other lines of investigation which may possibly be worth pursuing. Meanwhile, for historians to use the name ‘Domesday Book’ in speaking of any period before the 1170s is to invite misunderstanding at every turn, because it assumes the existence of something which may not (not yet) have existed.48 Until DB and D6 had been paired together and separated from the other D volumes (at the beginning of period 3), there was no such thing as ‘Domesday Book’, no entity to which the name could be applied; and we cannot be sure that this stage had been reached before some time in the reign of Henric II, three generations later than the date of the survey itself.

Even on occasions when DB was out of its box, the clerks who handled it no longer thought of making corrections or additions of any kind. By this time it would have seemed absurd to do so. Given that DB was infallible, how could it possibly stand in need of correction? On the contrary, it needed to be kept exactly as it was; and the rule established itself that no one should ever be permitted to make an alteration in the text, not even a mark in the margin. The rule was applied so strictly (until Arthur Agarde broke it) that nobody even dared to number the leaves.

6 Once DB had been fetishized by the king’s lawyers, once it had begun to be kept under lock and key, the fate of the six D volumes was more or less determined. One of them, because it supplied the main deficiency of DB, the absence of three whole counties, had to be kept with DB; the other five became redundant. Worse, one can see that they might become an embarrassment, because their continued existence would seem to cast a shadow on the status of DB. How could anyone be expected to accept the evidence of a copy – an imperfect copy, omitting much information – if it was known that the original existed too? This does not have to mean that the five unwanted volumes were discarded straight away; but it does suggest that they would have been kept out of sight and no longer made available to outsiders. Their future, or their lack of a future, became a matter which could be decided within the department. Sooner or later, they all ceased to exist; but we are unlikely ever to discover when that happened.47 to Westminster. That move happened in the early 1180s (Brown 1957), perhaps precisely in 1181. (It must, of course, have happened at some definite moment – the moment, say, when the Usher’s wife and children, with their furniture loaded into the cart behind them, arrived at their new home.) Perhaps DB and D6 were taken to London, the other D volumes left behind. But the Record of Farms, which also ceased to exist, survived at least until the 1190s (when it was made redundant by a change in Exchequer procedure).

46

At the very end of the century, two cases are recorded in which one of the litigants ‘puts himself on the Roll of Winchester’, ponit se super rotulum Wintonie. In one of these cases, a dispute over land in Lincolnshire, the man who appeals to this evidence expects it to prove that the land in question has belonged ‘since the conquest’ to the feod which he holds of the king (Curia regis rolls, vol. 1, p. 263); the relevant entry is in DB366va. The other recorded case originated in Suffolk; so by luck we can be sure that this unapt name was applied to both volumes, D6 as well as DB.

47

48

Worse still is ‘Exchequer Domesday’. Historians should surely break the habit of using this name when they are talking about the eleventh century.

One possibly relevant fact is the Treasury’s removal from Winchester

37

Chapter 4 The surviving portion of the C text – Part I

At least since the fifteenth century, the dean and chapter of Exeter have had in their possession a very remarkable manuscript, a collection of original documentation resulting from the great survey – the Descriptio totius Angliae, as it was called at the time – carried out in 1086.1 Whole theories have been built on the assumption that the manuscript originated in or near Exeter (whatever ‘in or near’ might mean), but there is no reason at all for thinking that. On the contrary, the indications are that the manuscript originated in Winchester – more precisely, that it was compiled in the king’s treasury,2 which, at that time, was housed in the castle at Winchester. How and when the manuscript found its way to Exeter is not so easily decided. It was certainly there by around 1500, when it underwent extensive annotatation at the hands of someone with a particular interest in the entries affecting the bishopric and church of Exeter (see below); and there are a few marginal notes earlier than that which show the same preoccupation.3 Knowing that the manuscript was in Exeter in the fifteenth century, we may perhaps feel safe in supposing that it was already there in the twelfth century. But there seems to be little hope of our ever being able to say for certain when exactly it arrived, or what the circumstances were.

leaf 162 should follow leaf 161 is an original fact, decided in 1086 – at the moment when a certain scribe (whom I call beta), after reaching the bottom of 161v, continued his text at the top of 162r. That leaf 161 should follow leaf 160 is a modern fact, the result of a decision made in 1816. Within any given booklet, the order of the leaves is fixed; but the order of the booklets, with respect to one another, is not. We have no clue what order was intended for these booklets by the scribes who wrote them. In fact, we cannot be sure that the scribes had any definite order in mind. Before it was rearranged and rebound in 1816, the manuscript consisted of two volumes of roughly equal size (257 and 274 leaves respectively). There is some reason to think that the book had already been rebound once before, in the mid eighteenth century (see below), and the division into two volumes may have originated then. The order of the contents, fortunately, is not in any doubt, because the leaves were all numbered, in about 1500, with perfect accuracy: i– xii and 1–246 in what was or what became volume 1, 247– 520 in what was or what became volume 2.4 The arrangement recorded by this old foliation was, for most purposes, far from satisfactory.5 In many places, quires had become transposed, or more or less distantly dispersed. To take just one example, the booklet describing the lands of the bishop of Coutances (now fos. 121–53) consists of six quires; and when these quires are put into the right sequence the old foliation runs as follows: 318–21, 322–9, 135–40, 123–30, 131–4, 52–4. At least three things had to happen before the order of these quires could become so badly disrupted, and all these things had to happen before the quires were bound.

Leaving that question in abeyance (I revert to it in chapter 5), we can start by looking at the manuscript itself. We discover, first, that it is not a book in any normal sense. It has been made to look like a book: it has been bound (more than once) and foliated (twice). It has been treated like a book: it has been deposited in a library. And yet it is not really a book. It is a collection of booklets, self-contained units which might have been arranged in any number of different sequences. The existing arrangement is the one imposed when the manuscript was rebound in 1816 (see below); the leaves were numbered (in fact renumbered) accordingly. Though we cannot dispense with this numbering, we have to remember that the arrangement which it codifies is, to a large extent, both arbitrary and recent. That

The person who started restoring the broken connections was the unidentified scholar (the same man responsible for numbering the leaves) who, around 1500, wrote a note at the bottom of 199v (his 393v) and a matching note at the top of 200r (his 428r) pointing out that the text is continuous.6 4

The numbering included every blank leaf; it also included the bookmark (below, note 21), which I do not count.

1

Exeter Cathedral Library MS 3500, the first good description of which was provided by Ker (1977, pp. 800–7). I thank the library staff for supplying me with a microfilm copy, and for answering numerous queries. For information and advice on various points, and for comments on a draft of chapters 4–5, I am grateful to Caroline Thorn and Dr Teresa Webber.

5 6

Not for all purposes, however, as will appear in chapter 5.

At the foot of 199v he wrote: Verte deinceps ad 33 folia et ad hoc signum, making a trefoil-shaped mark; at the top of 200r he made a matching mark and wrote: Vide principium huius ante 33 folia. (He should have counted 34, not 33.) Later, after he had foliated the manuscript, he continued the first note by adding: quod habes infra fo. 428. The dating is from Ker (1977, p. 805): ‘s. xv/xvi or xvi in.’ The same man (so it seems to me) made many other marginal notes (mostly repeating the place-names which occur in the text) and compiled a partial index (534v, formerly a flyleaf at the front of volume 1). He was certainly connected with Exeter; it would be interesting to know who he was. One candidate is mentioned below

2

As seems to have been realized, at least for a moment, by Galbraith (1950, p. 3). The proof comes from the batch 3 geld accounts (see below), which are discussed in chapter 6.

3

On 120v, for example, in what looks to me like an early fifteenth-century hand, there is a note beginning: Nota de domibus Episcopi in Excestre . . . . Above this, Excester was added by the annotator active around 1500.

38

The surviving portion of the C text – Part I But the man who did most of the work was a well-known eighteenth-century antiquary, Charles Lyttelton, whose appointment as dean of Exeter in 1748 gave him the opportunity to examine the manuscript closely (and apparently to have it rebound).7 Wherever he found the text broken off at the end of a quire, he scanned through the manuscript in search of the continuation. In two instances the search failed, because the leaf onto which the text ought to continue has been lost (see below). In every other instance the search succeeded, and Lyttelton recorded the result by writing a pair of notes in the margins of the manuscript – one at the foot of the verso where the text breaks off, the other at the top of the recto where it resumes.8

Caley, visited Exeter to examine the manuscript in September 1811 (note by Barnes, 541r). Acting on instructions from him, Barnes copied the manuscript line for line and page for page,13 rearranged the sheets into a better order, as far as Lyttelton’s notes allowed him to do so,14 and then sent the finished transcript up to London.15 Working with this, Ellis rearranged the sheets still further, putting them into the order which seemed best to him,16 and forwarded the copy to the printers, the firm of Eyre and Strahan; the proofs were sent down to Exeter, where Barnes checked them against the original. Ellis only saw the manuscript once, in April 1816, when he made visited Exeter for the purpose.17 The introduction that he wrote for this volume is dated October 1816, and 1816 is the date which appears on the title page; but copies did not begin to be issued till March 1817 (Condon and Hallam 1984, p. 383).18 Ellis’s introduction was also published separately, in quarto, with a title page dated 1817.

In 1810, the chapter clerk, Ralph Barnes,9 checked through the manuscript leaf by leaf, making sure that it was complete.10 By that time (I suppose) the Record Commission had started to think of printing the text of this manuscript: 11 it was published six years later, with three shorter texts, in a large volume edited by Henry Ellis.

Soon after Ellis’s visit, the manuscript was taken apart, rearranged into (almost) the same order as the printed text, and rebound as a single volume.19 The new foliation seems to have originated when Ellis rearranged and renumbered the sheets of Barnes’s transcript. Later, when Barnes tried to rearrange and renumber the leaves of the original in the same way, he discovered that it was, in two places, physically impossible to do so; and the foliation of the manuscript itself is, at twice as many places, defective for that reason.20

I do not think we know exactly what happened between 1810 and 1816, but the sequence of events seems to have run something like this.12 The commission’s secretary, John (note 10). 7 A letter of Lyttelton’s, quoted by Lloyd and Woodcock (1956, pp. 15– 16), indicates that this was one of the manuscripts which, as he says, he has had ‘new bound and repaired’ since his arrival in Exeter. ((Ker noticed ‘the pinmark from a former binding’ imprinted on several leaves which, before 1816, were at the end of volume 2 (Ker 1977, p. 805, cf. Thorn and Thorn 1985, note Exon. 19, 28–48). This fact, taken together with the negative fact that there is no similar impression on the leaves which came at the end of volume 1 (now fos. 526–8, 1–6, 529), seems to prove (as I ought to have realized sooner) that the medieval binding consisted of a single volume, and that the division into two volumes originated with Lyttelton’s binder, circa 1750.))

Prescott (2001, pp. 180–5).)) 13

Some pencilled dates written into the original by Barnes appear to record the progress of the transcription. ((It turns out that Barnes did not work alone: ‘a local antiquary called Jones’ was associated with him, checking the copy and, later, helping to check the proofs (Prescott 2001, pp. 182–3). I think that this is John (Pike) Jones (1790–1857), but have not yet been able to confirm it.))

8 A draft (dated 1750) of a paper by Lyttelton discussing the significance of the manuscript is now bound in at the back of it (fos. 538–9). His remarks are sensible enough as far as they go; but they could not go far, because his knowledge of the parallel DB text was confined to a couple of excerpts (535r–v). The paper was read to the Society of Antiquaries in January 1756; the book itself was brought to London and exhibited on that occasion (note by Lyttelton, 541r).

14

That Barnes rearranged his transcript before sending it to Ellis seems to be proved by a concordance which he drew up for his own use (Exeter Cathedral Library X 52, for sight of a copy of which I am grateful to Caroline Thorn). As was noted by Ker (1977, p. 800), what Barnes refers to here as ‘the new arranged order’ is not the same as Ellis’s order but represents a step towards it. 15

The complete transcript was in Ellis’s hands before March 1813, when the volume was ordered to be printed (Ellis 1816, p. viii).

9

Ralph Barnes was born in 1781 and died in 1869. I am obliged to Caroline Thorn for investigating this question and letting me know the result.

16

Ellis (1817, p. 2) calls it ‘the most obvious order’, an odd choice of words. The only outright error is a failure to undo one accidental transposition. The leaf now numbered 400 should come between 402 and 403, not between 399 and 401 (Ker 1977, p. 805). Lyttelton had seen this; Barnes missed it, and so did Ellis.

10

In fact, he discovered that leaf 221 (now 347) was missing, ‘cut out and no doubt stolen’ (note by Barnes, 541r). (The excision did not occur till after the leaves had been numbered, i.e. not sooner than circa 1500.) By a stroke of good luck, the missing leaf was found, recognized for what it was, and returned to Exeter in 1824 (Woolmer’s Exeter and Plymouth Gazette, 16 Oct. 1824, col. 3a, reprinted in The Times, 19 Oct. 1824, col. 2d; Botfield (1849, p. 139) has another account of this incident, presumably derived from Barnes.) Somehow or other, the lost leaf had found its way into the archive at Nettlecombe, the Somerset residence of the Trevelyan family: the Trevelyans suspected (I do not know quite why) that the person who had stolen it (or borrowed it and failed to return it) was Edward Willoughby (d. 1508), prebendary and dean of Exeter. The story might be worth exploring further.

17

Presumably James Basire must also have visited Exeter, to trace a page from the manuscript (117r) to illustrate Ellis’s edition.

18

The lost leaf recovered in 1824 (above, note 10) was printed and distributed later, as a half-sheet intended for insertion after page 326. ((This happened, so I gather, in 1828 (Prescott 2001, p. 184).)) 19 The book was rebound in May 1816, as is noted on one of the new f lyleaves. One quire was misplaced: the leaves now numbered 519–25 are bound between 494 and 495 (Ker 1977, p. 800).

11 Condon and Hallam (1984, p. 383) say that the edition ‘had been under active consideration since 1811’; I take it that Barnes’s examination of the manuscript in the previous year was prompted by some preliminary enquiry. 12

20 The irregularities were noted but not explained by Ker (1977, p. 807). There are two missing numbers and two unnumbered blank leaves. As I understand it, the unnumbered leaf after 64 was Ellis’s 74, but could not be moved because it is conjoint with 63; and the unnumbered leaf before

((I have revised this paragraph to take account of the facts reported by

39

The survey of the whole of England In its present form – the form which it owes to Barnes and Barnes’s binder – the manuscript is a single volume comprising 531 leaves.21 There is no facsimile edition; Table 9 gives a list of all the published reproductions known to me, with the scribes who appear identified in the last column.22 (A complete list of scribal stints is given at the end of chapter 5.)

in size, though most comprise 8 leaves or fewer. The largest by far is the booklet describing the lands of the count of Mortain, which runs to 72 leaves (fos. 210–81). At the other extreme, in ten instances a single leaf forms a booklet by itself.26

Except for two gatherings, the leaves are of a standard size (280 165 mm), and the pattern of ruling is uniform, with twenty widely spaced lines (Ker 1977, p. 806).23 In contrast, the collation is exceedingly irregular, so much so that it is difficult to say how many quires there are.24 In places, there is a run of normally constructed quires: one booklet, for instance, consists of four quires, three of 8 and the fourth of 4 leaves (fos. 288–315). But that degree of consistency is exceptional. Most gatherings have fewer than 8 paired leaves; one has 10; one is a giant with 20 (fos. 255–74). In addition, there are many single leaves, 95 in all; often there are several singletons in succession. To judge from all this (and from the intended purpose of the manuscript, so far as we can infer it), the scribes who wrote these gatherings and leaves had no expectation that they would ever be bound; when it was decided, some time later, to turn them into a book, the binder found himself faced with a rather awkward task.25

(1) part of a batch of booklets (C-WiDo) containing a transitional version of the survey text for two counties, Wiltshire and Dorset (fos. 25–62, 530–1);27

Very briefly, the contents are as follows:

(2) a batch of similar booklets (C-DnCoSo), nearly complete, covering three counties, Devon, Cornwall and Somerset (fos. 83–494);28 also one related booklet (CappDnCoSo) containing short versions of some entries for the same three counties (fos. 495–525); (3) four booklets containing geld accounts for Dorset (fos. 17–24), Devon (fos. 65–71), Cornwall (fos. 72–3) and Somerset (fos. 75–82) respectively; (4) three booklets each containing a different version of a geld account for Wiltshire (fos. [1–6 + 529],29 7–12, 13– 16); (5) two small miscellaneous booklets (fos. 63–4, [526–8]).

For the same reasons, it is hard to say precisely how many booklets exist. By my reckoning, there are 72 of them, but other investigators, counting for themselves, might arrive at a slightly larger or smaller total. The booklets vary greatly

From this point on, I deal only with the first two batches, which, by my count, comprise 15 and 48 surviving booklets respectively. Rather than trying to cope with everything at once, I ignore all additions made in C by hands other than those involved in writing the original text. Some of these additions were made at an early stage, and will have to be taken account of in due course (below, pp. 71–9). Nevertheless, as far as I can, I refrain from discussing them here.30

188 was Ellis’s 176, but could not be moved because it is conjoint with 190. 21

I exclude fo. 532, a narrow leaf which seems to be a twelfth-century bookmark; it is inscribed with a partial index (Ker 1977, pp. 806–7). Before 1816, this leaf was between 280 and 281 (old 91 and 93), facing 281r: the old number (92) is on the verso, therefore, and Ker failed to see it. Unlike him, I am not inclined to read any particular significance into the index. But if the hand could be proved to be an Exeter hand, that would be a point of some importance. 22

Batches 1–2 are, in my notation, the C booklets. What they contain, as Galbraith (1942) was first to see, is a transitional version of the survey text, the medium through which the

((I have added the samples reproduced by Thorn and Thorn (2001).))

23

Some marginal additions begin or end at the very edge of the leaf; some are actually written along the edge (such as the note iii hed˛e remanent at the top left corner of 104v), and would have been cut through if the leaves had been trimmed even slightly. Here and there, however, a leaf which protruded too far seems to have been cut down to size. At 434v, for instance, a marginal addition is partly sliced away.

26

On any natural definition, there is no reason why a booklet should not consist of just one leaf, in the extreme case. It would, I concede, not normally be sensible to think of singletons as booklets, but the circumstances here are special.

24

Anyone who wants to get a grasp on the structure of the manuscript will need to make a replica of it from sheets and half-sheets of paper, taking note of the old foliation as well as the new one. The requisite data are all to be found in Ker’s (1977) description; two small corrections are noted above (notes 20 and 21). There is one further puzzle which I cannot solve. According to Ker, fos. 489–94 are a quire of 6; but it is clear from the old foliation (confirmed by Lyttelton’s notes) that the first two leaves had got themselves transposed; and that, for a quire of 6 , is physically impossible. It seems that Ker’s description must be wrong on this point, but I cannot say what the right description would be.

27

That the last two leaves (old fos. 205–6) belong with fo. 62 (old fo. 204) is stated as a fact by Ker (1977, p. 805); though I cannot confirm it, I do not think of doubting it. They contain some extraneous memoranda, written by a hand which I think occurs only here. See also the following note. 28

It is possible that fo. 398 was originally part of C-WiDo (below, p. 44).

29

Here and in batch 5, square brackets distinguish gatherings which differ from the standard format. 30 To simplify the argument further, I will relegate all discussion of CappDnCoSo (mostly in chapter 5) to the footnotes. The entries here are of a different type from those that occur in the main C text. In every other respect, however, Capp is just one more C booklet, compiled at the same time as the rest. Its existence implies that each D text was intended to include an appendix consisting of entries like those collected here.

25

It is possible that the binder rearranged the leaves to some slight extent, to simplify things for himself. For example, one quire of 4 (fos. 437–40) should perhaps be regarded rather as two quires of 2, one inside the other (Ker 1977, p. 805): since the last leaf (fo. 440) is blank, a binder might reasonably think that it would do no harm if he made one quire out of two.

40

The surviving portion of the C text – Part I Page 1v 8r

Wi Wi So So Dn Dn Dn

Co Co Dn Dn Dn So

8v1–9 9r1–6 9r5–18 9r7–12 14r 15r 47r 47r5–6 103r 114v1–4 108r 117r 117r1–17 153v 153v 175r19–v3 181v3–4 245r16–20 313r 316r7–9 356r6–8 436v 438r 531r

Reproduction

Scribes represented

Darlington 1955, opp. p. 180 Darlington 1955, opp. p. 181

ksi rho (1–11), sigma (12–30), tau (30–41), mu (margin) rho tau tau (5–6), sigma (7–18) sigma sigma (1–20), tau (20–8), sigma (28–9) sigma ksi ksi alpha eta beta alpha (1), beta (2–4), alpha (5–20) alpha (1), beta (2–4), alpha (5–17) DB scribe DB scribe bishop of Durham’s scribe alpha zeta delta (1–5), epsilon (5–10), gamma (11–20) beta epsilon DB scribe alpha (1–5), beta (6–12), alpha (13–20) —

Thorn and Thorn 2001, ill. 27 Webber 1989, pl. 3 Ker 1976, pl. III (a) Webber 1989, pl. I Darlington 1955, opp. p. 216 Hallam 1986, pl. 7 Darlington 1955, opp. p. 169 Galbraith 1961, pl. II [c] Bond et al. 1884–94, pl. 70 Webber 1989, pl. 5 Lloyd and Woodcock 1956 Ellis 1816 Thorn and Thorn 2001, ill. 26 Finn 1951, opp. p. 563 Rumble 1985, pl. 3.5 Chaplais 1987, pl. III (a–b) Galbraith 1961, pl. II [d] Webber 1989, pl. 7 Bond et al. 1884–94, pl. 71 Galbraith 1961, pl. II [a] Galbraith 1961, pl. II [b] Finn 1951, opp. p. 562 Darby and Finn 1967 Darlington 1955, opp. p. 217

Table 9. Published reproductions of sample scripts from Exeter Cathedral Library 3500. (Samples of the C text are distinguished by the county code on the left.) initial version B was transformed into the final version D. No part of B survives in the original, but it is clear enough, from the surviving C booklets themselves, that an earlier version of the text did exist, differently arranged from C. As for D, three booklets survive, one for each of three connected counties (PRO E 31/1), and the presumption is, as Galbraith realized, that a version of the text arranged in the same way did originally exist for every county. (Except for these three, what survives instead is an epitome of D, the version I call DB (PRO E 31/2).) The B text was organized cadastrally: within each county there was a section for each hundred; within each hundred there was a subsection for each vill; within each vill there was a paragraph for each manor. Of necessity, this was the frame which had been used for collecting and recording the information in the field; but it was not the frame in which it was desired to have the information permanently recorded. The intention was for the D text to be organized feodally within each county: there was to be a chapter for the king, and then a chapter for each of the king’s men. At top and bottom the frames coincided – counties at the top, individual manors at the bottom – but in between they were quite differently constructed. The C booklets are the instrument which extracted the entries from the cadastral frame and lined them up ready for insertion into the feodal frame. Unless this is understood, the design of the individual booklets will seem

exceedingly strange; and it needs to be remembered that nobody really knew where one booklet ended and the next began until Ker (1977) worked it out in detail. The two batches of C booklets are segregated geographically. Batch 1 covers Wiltshire and Dorset; batch 2 covers Devon, Cornwall and Somerset. The segregation is strictly maintained: there is no leakage of entries between one batch and the other. If a tenant happens to have held land both in a county covered by batch 1 and in a county covered by batch 2, invariably we find that his lands are described in two separate booklets. A man named Willelm de Moion, for example, held land in four of these counties; so for him we find two booklets, one describing his land in Wi and Do (fos. 47–9), the other describing his land in Dn and So (fos. 356–65). The same holds true for five other tenants.31 If we restricted ourselves to internal evidence, it would be uncertain which batch should come first; but the miscellaneous booklets listed above (batch 5) seem to answer that question for us. In particular, one of the scribes who worked on the C booklets (the scribe whom I call mu) was also responsible for writing out a statistical summary of 31 Apart from Willelm de Moion, the tenants represented by a pair of booklets are the abbot of Athelney (fos. 41 and 191–2), the abbot of Tavistock (fos. 42 and 177–81), Roger Arundel (fos. 50–2 and 441–5), Serlo de Burci (fos. 53 and 452–5), and Walter de Clavile (fos. 62 and 388–97).

41

The survey of the whole of England the lands of Glastonbury abbey (527v–8r), which extended over four of these counties (all except Co). For this scribe, we find, the order in which these counties should be listed was WiDoDnSo. Hence it seems clear that for him batch 1 came before batch 2.32 Moreover, the counties represented here (WiDoDnCoSo) were, at a later stage, the last five counties dealt with by the DB scribe (Fig. 7), and the order in which they were dealt with is almost the same (. . . WiDoSoDnCo). It seems that these were the last counties to pass through the machinery, and the survival of the C booklets is doubtless due partly to that.

and the longest consists of four. Changes of appearance occur with some frequency too (there are two at least in the booklet analyzed by Ker), but these are more a matter of opinion, and less obvious in their significance. (They may just mean that the scribe ceased work for a moment in order to sharpen his pen.) Often a change of hand coincides with the start of a new entry, but that is not always the case. Occasionally the scribe who had just completed one entry, assuming that another would follow, continued with the opening words of the next paragraph: N habet unam mansionem quae uocatur . . . , ‘N has a manor which is called . . . ’. If, in the event, the subsequent entry was written by somebody else, a change of hand is visible at this point. But the hand may also change in the middle of a paragraph, for no apparent reason.

From our point of view, it may sometimes be better to start by looking at C-DnCoSo, the batch which is nearly complete; and that is the policy I follow here in trying to understand more clearly how the booklets were compiled. In principle, the intention was to make a separate booklet for each tenant who held his lands directly from the king. (The king himself was dealt with similarly, but in some respects the land which belonged to him demanded special treatment.) A team of scribes scanned through the B text for the given county, locating the entries for the manors held by this man. (How many scribes were at work together will be discussed in chapter 5; how they managed to get the job done without constantly tripping over one another’s heels is explained in an appendix below.) As each pertinent entry was located, it was copied into the C booklet;33 and then the scan was resumed. In the end, when all the pertinent entries had been located and copied, what this booklet contained was a preliminary version of a chapter to be included in the D text.

The surviving C booklets have one other remarkable property: each batch covers more than one county. It is doubtful, I think, whether this would have been true for every batch of C booklets, but it is certainly true for the batches which survive. As far as batch 2 is concerned, what happened was this. After finishing with the first B text, B-Dn, the scribes continued with a second, B-Co, copying the entries from this into the same C booklets as before, so far as that was possible. When they came across a name which they had not encountered previously – a tenant holding land in Co who did not hold land in Dn too – then of course they had to start a new booklet;35 but otherwise they used the same booklets again, marking the start of a new county with a boldly written heading (unless they forgot). Similarly, after finishing with B-Co, the scribes continued with a third B text, B-So, copying the entries from this into the existing booklets, so far as that was possible, not starting a new booklet except when they had no option. Then finally they did call a halt.

Sometimes it happened that several successive entries or even several successive pages were written by the same scribe; most stints, however, were quite short. It is common to find two or three changes of hand in the space of a single page. In one fairly typical stretch of text (376r–9v), comprising 24 entries for one county, Ker (1977) thought that he could recognize the work of eight different scribes, with 14 visible discontinuities where one scribe took over from another:34 most stints here consist of one or two paragraphs,

Why the C scribes adopted the policy of packing two or three county texts into the same collection of booklets is not an easy question to answer. At least we can say that under certain conditions this policy would do no harm. If the D scribes had been treading on the C scribes’ heels, presumably each C text would have had to be forwarded as soon as it was finished. If, on the other hand, the D scribes were lagging behind, the C scribes might have been sure that they could finish a second or even a third text before the D scribes would be ready to deal with the first one. In those circumstances, the packing policy would not have risked causing delay. But I do not see that it had any positive advantage, except that by continuing with the same booklets, rather than starting a new collection, the scribes would have saved some parchment. From the amount of

32

The summaries copied by another scribe on 530v–1r also assume that WiDo should come before DnCoSo. But this scribe, unlike mu, did not (as far as I can see) participate in the writing of the C text; so his opinion carries less weight. This, by the way, is the only scribe of whom it can be positively stated that these five counties were his universe: for him, at least for the time being, there was nothing before Wi, nothing after So. For the C scribes, only the second half of this statement is, arguably, true. 33

In most of the C texts, the scribe would not have copied the entry straight away. Instead, having found it, he would have started scanning backwards, in search of the hundred heading which governed this entry. Then he would have copied the heading; and only then would he have jumped forwards again and copied the entry itself. In the last few C texts, however, the scribes were allowed to simplify the process by dispensing with this backward scan. Apparently it was thought more important to speed things up than to include the hundred headings.

here (376r10), but not a change of hand – and his (8) is a split personality – the first three lines of this paragraph (378v18–20) were written by (7), the last two lines (379r1–2) by (4). Except for these two points, the equations between Ker’s scribes and mine are straightforward (as will appear in chapter 5).

34

My own analysis is slightly different, though the general picture is the same. Needless to say, I should be reluctant to disagree with Ker’s results if it was clear that the analysis had been fully worked out; but I venture to think that he was only aiming for a first approximation. As far as I can judge, Ker’s (1) is the same as his (2) – there is a change of appearance

35 In fact they only had to start one new booklet (fos. 202–9), but the principle holds. Later, while dealing with B-So, they had to start 15 new booklets.

42

The surviving portion of the C text – Part I blank space they left, it is obvious that the C scribes did not feel constrained by any shortage of parchment, nor greatly interested in reducing their consumption of it; even so, they would presumably have wished to avoid unnecessary extravagance. There may be more to this story than that – but if there is, I have failed to see it.

Accidents aside, there is a more important reason why collation of C and DB does not always work out easily. In principle there is a one-one correspondence between C booklets and DB chapters, but in practice there is not.40 If the scribes who worked on C had adhered absolutely to the rule that they should start a separate booklet for every single tenant, the result would have been an embarrassingly large number of one-sheet or one-leaf booklets. In almost every county this problem arose, to a greater or lesser extent; wherever it arose, it was dealt with in much the same way. Beyond what seemed to them a suitable point, the C scribes allowed themselves to stop obeying the rule. Instead they created a few omnibus booklets to accommodate left-over entries – one for churches and priests, one for the king’s serjeants, one for any English tenants who continued to hold some small piece of land directly from the king, and so on. To some extent, the ad hoc arrangements created by the C scribes persisted into D and DB. In detail, however, a fair amount of reorganization took place: some entries extracted from these booklets were turned into separate chapters. It would be hard to say exactly what motives were at work – why this entry was so treated, why that entry was not – but notions of decency or consistency or both seem sure to have played some part. (A bishop should have a chapter to himself, even if the chapter consists of only three or four lines; an abbot who has a chapter to himself in other counties should have a chapter here too.)41 It would also be hard to say how many of these changes were made by the D scribes, and how many by the DB scribe; but we are not in a position to face that question yet.42

Whatever reason the scribes had for proceeding in this way, the outcome is that the completed booklets may contain up to three stretches of text, derived from B-Dn, B-Co and BSo respectively. These stretches, as many as occur, represent preliminary versions of chapters to be included in DDn, D-Co and D-So respectively. Because of the accidents of ownership, it is only to be expected that many of the CDnCoSo booklets will lack one or two of these stretches. In fact, apart from the king, the count of Mortain was the only man who held land in all three counties. Hence only two C booklets (fos. 93–107, 210–81) show the full sequence of counties.36 Otherwise, the pattern is as well developed as it has a chance to be. In every booklet which contains a stretch of text derived from B-Dn, that stretch comes first; and in every booklet which contains a stretch of text derived from B-So, that stretch comes last. There are no inconsistencies. From internal evidence, and from collation with DB, it becomes clear that the surviving text is not quite complete. In two places, the text breaks off at the end of a verso page, and the leaf onto which it should continue is not to be found. Both times it is part of the C-Dn text which is lacking.37 At these points it seems certain that a leaf or more has been lost – accidentally lost, as far as we can tell, before the booklets were safeguarded by being bound. I am not so sure how we should think of explaining the fact that there are four whole chapters in the DB-Dn text for which no matching text exists in C.38 It is possible that booklets containing draft versions of these chapters did originally exist but came to be lost, either through sheer carelessness or because they were borrowed for some purpose and not returned. But it is possible too that these booklets never existed – that some entries were omitted from C, by mistake or for a reason which seemed a good one at the time, and then later transferred directly from B into D. In So similarly, there are two DB chapters for which no corresponding C booklets exist, and again that may possibly mean, but does not necessarily mean, that several C leaves have been lost.39

One of these rearrangements produced, by accident, a crucial piece of evidence, the significance of which seems first to have been noticed by Whale (1905, p. 266). Two tenants, Goscelm and Walter, both holding land in Devon, share a booklet in C (fos. 388–97): uniquely here, strings of entries for manors belonging to one man alternate with strings of entries for manors belonging to another.43 In DB-Dn, howhave been working for the bishop of Durham (Chaplais 1987). (2) No C text survives corresponding to DB chapter 33, but drafts of this and one other chapter were added on blank pages (153v, 436v) at the back of two C booklets, in a hand which does not just resemble DB (Finn 1951) but is actually that of the DB scribe himself (Chaplais 1987). ((These and some other additions to C are discussed in chapter 7.)) 40

Working on the last C text, the scribes occasionally dared to pack two chapters (382v–6r, 437r–9r) or several chapters (446r– 9r) into a single booklet. By this time, with the end already in sight, they could risk breaking the rules.

36

The Capp booklet shows it too, however, and so does a booklet (fos. 63–4) containing lists of hundreds for the same three counties.

41 These examples come from Somerset, for which (excluding the king) there are 30 C booklets and 46 DB chapters. Part of the difference is due to the creation of new chapters, in D or DB or both. For instance, one of the C booklets contains a stretch of text headed ‘Lands which have been given to the saints as alms in Somerset’ (196r–8v). The entries here mostly recur in chapter 16 of DB-So, ‘Clerics holding from the king’, but four have been extracted and turned into separate chapters (11–13 and 15).

37

There is a break after 414v (where DB has six more paragraphs) and another after 421v (where DB has one more paragraph). 38 Chapters 30 (with 10 paragraphs), 31 (with 4), 40 (with 22), and 45 (with 2). The numbering here is that which appears in the index (DB-Dn100ra), not the main text. 39

Chapters 2 (with 4 paragraphs) and 33 (with 2). These are both special cases. To avoid distraction, I do not discuss them here, but the basic facts are as follows. (1) A copy, presumably an emended copy, of the C text corresponding to DB chapter 2 was added on some blank pages (173v–5v) at the back of one of the other C booklets, by a scribe who makes only this one appearance in the manuscript (Ker 1977); it is suggested that he may

42 A means of answering this question was found by Galbraith (1961, pp. 190–3). It is, I think, the right one. ((With regret, I no longer think that (below, pp. 133–4): the question is not decidable.)) 43 It is stated by Finn (1957, p. 75) that Goscelm was Walter’s brother, but I do not know what evidence he had for saying so.

43

The survey of the whole of England ever, each man has a chapter to himself, chapter 25 for Walter, chapter 26 for Goscelm. The crucial point is this, that the C booklet consists of two quires (388–91, 392–7). In DB’s chapter 26, the order of the entries is the same as it is in C; in chapter 25 the order is not the same – but it becomes the same when the C quires are transposed (392–7, 388–91). (The order is precisely the same except for one entry, the first in C, which had to be treated as a special case.) Here we have the strongest possible proof that DB is derived from C, directly or indirectly. If we thought (as Whale did) that DB was copied directly from C, we should take this evidence to mean that the DB scribe, while writing chapter 25, had the quires of his source text out of sequence (but got them back into the right order before writing chapter 26). If we think (as Galbraith did) that DB was copied from a copy of C, i.e. from D, we infer that it was one of the D scribes who had temporarily transposed the C quires. The transposition has to be assumed to make its mark on whichever version of the text derives immediately from C; but on either view this evidence proves that C is ancestral to DB.

remainder of a larger batch. Of the booklets that survive, one alone (fos. 47–9) contains entries for both counties: a single entry for Wiltshire followed by a string of entries for Dorset. The rest relate only to Dorset. Despite that, it is, I think, quite certain that the batch, when complete, would have covered these two counties (these two and only these two) in exactly the same way that the C-DnCoSo booklets cover those three counties. What has happened here (approximately but not exactly) is that the booklets which contained a section of the C-Wi text, regardless of whether they also contained a section of the C-Do text, have (nearly) all been removed from the collection. Since many tenants held land in both counties, the end result was that most of the booklets were removed – and having been removed they were lost. The sub-batch which survives consists (nearly) of the booklets relating to Dorset alone.47 So far as they can be compared, the C-WiDo booklets are identical in type with the C-DnCoSo booklets. They were put together by the same method, with the same purpose in view. Why some survived when most were lost is a question which I propose to come back to later. What proportion has been lost is a question which we can try to answer straight away, though only a very crude estimate is possible. For the three counties covered by C-DnCoSo, there are 40 leaves in DB corresponding to 412 leaves in C.48 For the two counties covered by C-WiDo, there are 20 leaves in DB (not counting two inserted slips). Hence, assuming that the compression factor between C and DB was the same, we would expect there to have been slightly more than 200 leaves in C. Of course this estimate is not to be taken very seriously, but I think it would be fair to say that the surviving 40 leaves are unlikely to represent more than about one-fifth of the original number. A similar argument will give us a very rough idea of the size of C as a whole. I have suggested elsewhere (above, p. 28) that the D text, for all 33 counties covered by the survey, would have amounted to something like 2800 leaves, and a similar calculation leads me to suppose that the C text, complete for every county, would have run to approximately 3500 leaves. By that reckoning, the leaves which survive, 452 in all, represent around one-eighth of the total number.

In Ellis’s rearrangement, this booklet is followed by a single leaf (fo. 398) which raises different questions. It starts with a heading, TERRA GOSCELMI DE ESSICESTRA, of the kind which is normally found at the start of a new booklet or new county, except that here the name of the county is missing. This is followed by a single paragraph (398r2– 7) relating to a place named Herstanahaia; the rest of the leaf is blank.44 In D and DB (in D by inference, in DB as a matter of fact), the matching entry was included under Devon, at the end of chapter 26, and that is why Ellis chose to put the leaf here, at the end of the corresponding C booklet. But it has been suggested – again, I think, originally by Whale (1905, pp. 249–50) – that DB is doubly misleading. Goscelm of Exeter, it seems, was not the same man as the Goscelm whose lands are described in this chapter; 45 and the place in question may be a known place in Dorset, not an otherwise unrecorded place in Devon which happened to have the same name. The evidence of the script seems to confirm Whale’s suggestion: the scribe who wrote this C entry occurs elsewhere writing entries in C-Do, but not for any other county.46 Apparently this leaf was originally part of C-WiDo, but went astray at an early stage – early enough for it to be mistaken for part of C-DnCoSo by one of the scribes who worked on the D-Dn text.

The survival of even this much is a miracle for which we ought to be properly thankful (if only we knew whom we ought to be thankful to). Our gratitude would probably be

The C-WiDo booklets, to turn to them, are only the small

47 In other words, if a tenant held land in Wi as well as Do, or in Wi alone, no booklet ought to survive (though in fact one does). If a tenant held land in Do but not in Wi, the booklet describing his lands, if it ever existed, ought still to exist; if it survives at all, furthermore, it ought to survive complete. Up to a point, these predictions hold true; but it is hard to speak precisely, because the text clearly underwent the same kind of reorganization, in D or DB or both, mentioned above with regard to CDnCoSo. For example, two entries extracted from one of the surviving C booklets (fos. 27–9) recur as a separate chapter (18) in DB-Do. There are some other short chapters too (7, 14, 17, 21–3) of which it would not seem unlikely that they came from an omnibus booklet.

44 Except for an unfinished entry, Eduuardus tenet iii hidas terr˛ e ... , added at 398r9 by a different hand (which I do not recognize). 45

The DB scribe, presumably alerted by some abnormality in D, was doubtful about the identity. If he had been sure, he would have starting this entry by saying Id’ Go. ten’ . . . . Because he was not, he wrote Goscelmus de Execestre ten’ de rege . . . , and so gave himself the option of converting this paragraph into a separate chapter, if that proved to be necessary.

46

The same scribe (whom I call omicron) wrote another isolated entry (41r1–9) which looks very similar to this one. There too he forgot to specify the county – but somebody else added the words in Dorseta before any harm was done.

48 Here I am counting fo. 398 as part of C-DnCoSo, which is how it was treated by D (or by DB), but ignoring the fact that some leaves (two at least) have been lost.

44

The surviving portion of the C text – Part I greater, however, if things could have been so arranged that the counties represented by the surviving C booklets were geographically dispersed. Given the choice, we should (it seems to me) have preferred to have our sample of counties scattered widely across the map – Yorkshire, Norfolk, Sussex, Somerset and Shropshire, or something along those lines. In fact, the sample consists of a block of connected counties, and there is, for that reason, some room for doubt how far we can safely extrapolate. (A similar caveat applies to the surviving D booklets, and that makes things twice as difficult.) Suppose that something is true for all five county texts, or at least for all four of which more than one entry survives: do we infer that it was universally true? Suppose, for instance, that a scribe can be identified whose hand occurs repeatedly in the text for these four counties. (There are in fact two scribes who fit this description, as will appear in chapter 5.) Knowing that, do we think that the same scribe’s hand is likely to have occurred in every single C text? If the counties in question were scattered around the country, presumably we would think that; because they form a connected block, we have to hesitate.49 If readers will agree, as I hope they will, that this question needs to be asked, I am willing to leave the answer in suspense.

reaches the end of B: when he does, this C booklet is complete. Setting it aside (after extracting any unused sheets from the middle of the final quire), the scribe goes back to the beginning of B, decides which tenant he wishes to deal with next, and starts another C booklet. And he repeats this procedure as often as is required. Eventually, when every tenant has been dealt with, each entry from B will have been copied into one of the C booklets.50 Once that stage has been reached, the stack of completed booklets will need to be sorted into their final order; and then the text can be recopied continuously, by the same scribe or by somebody else, into as many quires as may be needed to make up the D booklet. Assuming that the scribe always works as fast as he can, the amount of time the job will take is a fixed quantity. To scan repeatedly through B, to copy each entry from B into C at the appropriate moment – all this will take some definite number of man-hours. By making sure that the scribe does not slack, we can prevent the job from taking longer than it should; but we cannot get it done in a shorter time unless we employ more scribes. If one scribe would take ten days to do the job, it seems that two scribes ought to be capable of doing it in five days – and that five scribes ought to be capable of doing it in two days.51 But that will only be possible if we can find some way of arranging things so that two or more scribes can work side by side without obstructing one another. Each scribe has to be free to scan through the source text, copying whichever entries need to be copied, unhindered by his colleagues.

Appendix How to compile a collection of C booklets It is not obvious how five or six scribes could collaborate in compiling a collection of C booklets without constantly getting in one another’s way. If one of the scribes is copying a certain paragraph from a certain page, what can the other scribes do except wait for him to finish, so that one of them can take a turn? And what is the point of hiring several scribes if only one of them is going to be busy at any given moment?

There is (in the absence of any means of mechanical reproduction) only one way of achieving that result: by taking B to pieces. We have not yet formed any clear idea what B would look like, but uncertainty on that point need not detain us here. Possibly B is a collection of booklets, somewhat resembling C; possibly it is a collection of quires, somewhat resembling D; either way, it can easily be taken apart – into batches of one or more booklets or quires, as the case may be. Because I have to call them something, I call these batches booklets; but I do not mean to say that they were booklets in a strict sense.

If time were no constraint, we could hire one scribe and let him do the whole job himself. Suppose we do just that. We recruit a reliable scribe; we give him the source text, a supply of blank quires, and some fairly simple instructions; and then we leave him to get on with the work. It is clear enough what this scribe has to do. He decides which tenant he wishes to deal with first; then he takes a blank quire and turns it into a C booklet by writing a title at the top of the first page. Now he starts scanning through B, looking for an entry for a manor belonging to this tenant. When he finds such an entry, he copies it into his quire; and then he resumes his scan of the source text. One by one, the relevant entries are found and copied; whenever necessary, the scribe adds a new quire to his booklet. Sooner or later he

With the source text divided into several booklets, we can hope to find work for several scribes at once: each scribe can be given one of the B booklets and one of the C booklets and told to copy the relevant entries (if any occur) from the former into the latter. The job divides itself into a set of such tasks. (What makes the process interestingly unpre50

Given B, we could replicate C, simply by repeating the operation: errors aside, the C booklets which we might make would each comprise exactly the same entries, in exactly the same sequence, as the original booklets. But the process is not reversible. We cannot reconstruct B from C, still less from D or DB. Once the entries have been redistributed among the C booklets, they cannot be put back together in their original order. Too many connections have been broken.

49

We might think it possible, for example, that the plan of the writingoffice resembled a map of the country: one group of connected counties (including these four) was dealt with in the south-west corner of the room, and other groups of counties were dealt with simultaneously in the other corners. On that view, what was true in one corner would not necessarily be true in every corner.

51 I doubt whether this level of efficiency could actually be achieved. Some crude simulations, using a modified version of the program described below, lead me to think that the best we could hope for is that two scribes might get the job done in 5.5 days, five scribes in 3 days.

45

The survey of the whole of England dictable is the fact that we do not know in advance the duration of any given task, though we assume that some tasks will take much longer than others.) Since each C booklet has to be collated with each B booklet, the number of tasks is the number of B booklets (which to some extent we can vary to suit ourselves) multiplied by the number of C booklets (which is fixed, whether or not we know it in advance). Now the problem is to ensure that every task gets done precisely once – that none gets overlooked, and that none by mistake gets done twice. But that is easily achieved. We just need to keep the B booklets in the same order throughout, and let each of the C booklets advance along the line, shaking hands (so to speak) with each of the B booklets in turn.

Now there are two tasks waiting to be dealt with: somebody needs to copy from b1 into c2, and somebody needs to copy from b2 into c1. One of these tasks can be assigned to the scribe who has just ceased work; my program chooses the latter task for him.53 The other can be assigned to a second scribe, s2. Then, once again, we wait for a task to be completed.54 In the event, scribe s2 is the first to finish. Booklet b1 goes back to its place, c2 advances one position,55 and a new blank quire, c3 in embryo, can be put in front of b1. (The booklets being used by s1 are currently out of the picture.)

b1 c3

We have the makings of an algorithm here. Exactly how the algorithm was implemented we cannot expect to know, but I will describe the first few steps in a small-scale simulation, to show how it might have worked.

b2

b3

This time it is scribe s1 who finishes first. Following the usual rule, he replaces b2 and puts c1 one position further right.

b4 c4

All the way through, a C booklet in front of a B booklet will signal the fact that a task awaits attention. (Provided that they understand this simple rule, the scribes can left to manage the work by themselves, without supervision.) At this stage, only one task is presenting itself: somebody needs to copy the relevant entries from b1 into c1. We assign this task to the first available scribe, s1. He takes the booklets he needs, carries them off to his desk, and sets to work. Now we have to wait for him to finish: nothing more can be done till then.

b2 c1

b3

b2 c2

b3 c1

b4

Now there are two tasks awaiting attention, (b2, c2) and (b3, c1). One of these tasks (say the latter) can be given to scribe s1; the other can be assigned to a third scribe, s3. Once these new tasks are started, therefore, three scribes are at work simultaneously: s2 is still copying from b1 into c3, s1 is copying from b3 into c1, and s3 is copying from b2 into c2. The reader will not feel dissatisfied, I hope, if I interrupt the narrative at this point. By now it ought to be clear enough how the C booklets can be made to travel along the table, acquiring the relevant entries from each B booklet in turn, as they pass by. The rest of the run, with three scribes and five C booklets, is summarized below (Table 10).

In time he completes his task.52 Having done so, he returns to the table with the booklets which he has been using. What he does now, obeying the instructions we have given to all our scribes, is this: he puts b1 back where it came from, and c1 one position further right. The position in front of b1 is thus left vacant, and can be filled with a new blank quire, booklet c2 in embryo.

b1 c2

b4

There is only one task to be assigned, task (b1, c3), and s2 is available to deal with it. He carries off the booklets he needs, rejoining s1, who is still busy with his task, (b2, c1). Again we wait for one or other task to be completed.

Suppose that we divide B into four booklets, b1 through b4. Finding some workspace, such as the top of a table, we lay out the booklets side by side on this surface. Providing ourselves with a supply of blank quires, we get the job started by placing one of these quires, booklet c1 in embryo, in front of booklet b1.

b1 c1

b3 c2

53 In my program, a scribe who has just completed a task is always assigned another task as soon as possible. If two tasks are available, one is chosen for him at random. 54 When two or more tasks are in progress simultaneously, my program makes a random choice of the one which gets finished first.

b4

55 As the work develops, it will sometimes happen that a C booklet advances into a position which is already occupied by one or more preceding booklets. If we want the C booklets to be kept in the same order throughout, the newly-arrived booklet must be put at the bottom of the stack; and that is what my program will do, unless it is told otherwise. But that is not necessary. If we prefer, the newly-arrived booklet can be put at the top of the stack; or it can be inserted into the stack at some randomly chosen point. The algorithm will work in any case.

52

The completion of this task marks the end of interval 1. Since each interval ends with the completion of precisely one task, the number of intervals is the same as the number of tasks. These intervals, of course, are not equal. We are keeping count of a sequence of events, not measuring the passage of time.

46

The surviving portion of the C text – Part I Interval 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Scribes s2

s1 (b1, c1) (b2, c1) : (b3, c1) : : (b4, c1) : : : : : : (b4, c2) : (b4, c3) (b4, c4) : : (b4, c5)

(b1, c2) (b1, c3) : (b1, c4) : : : (b1, c5) (b2, c4) : (b2, c5) : : : : : (b3, c5)

s3

(b2, c2) : (b2, c3) : (b3, c2) : : (b3, c3) : (b3, c4) :

Table 10. Simulated compilation of a set of C booklets (three scribes, four B booklets, five C booklets). As we have seen, it takes some time, at the beginning of the run, before the scribes can all be set to work.56 Towards the end, similarly, the supply of tasks tapers off, and scribes start falling idle. Right at the end, only one scribe is left to deal with the final task, copying from b4 into c5. In the middle of the run, however, all three scribes are occupied at once, each with a task of his own. It is uncommon, in fact, for the work to flow quite as smoothly as it does here, but I did not have to interfere with the program to make it produce this result.57 Without further comment, I also print the results of a rather larger run, with four scribes, six B booklets and eight C booklets (Table 11). Readers may find it instructive to try tracing out the trajectory followed by some sample booklets. The entries in booklet c3, for example, were copied successively so: from b1 by s2; from b2 by s1; from b3 by s4; from b4 by s4 again but after a break; from b5 by s3; from b6 by s3 again. All four scribes participated. If every task involved the copying of at least one entry, the completed booklet will show three changes of hand, and possibly one or two changes of appearance as well. In other words, it will look very much like one of the surviving C booklets (except that only one county is represented). 56

At the start of interval 7, we could briefly find employment for a fourth scribe. With only four B booklets, however, there is not enough work to keep four scribes occupied for long. It is probably a good rule of thumb that the number of B booklets should be at least twice as large as the number of scribes to be employed.

57 I set the program to run ten times, intending to choose whichever result was prettiest. The second run came out like this, and I halted the program at that point.

47

The survey of the whole of England

Interval 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Scribes s1 (b1, c1) (b2, c1) : : : (b2, c2) (b2, c3) (b2, c4) : : : (b5, c1) (b6, c1) : : (b2, c5) : (b3, c5) : :

(b1, c7) : : : (b4, c5) : : : : : : : : (b5, c5) : : : (b5, c6) : : (b6, c6) : : (b6, c7) (b6, c8)

s2

(b1, c2) (b1, c3) (b1, c4) (b1, c5) : : : : : : : : (b1, c6) : : : : : : : : : (b2, c6) (b3, c6) : : : : (b2, c7) : : (b3, c7) : : (b3, c8) : :

(b6, c5) :

s3

(b3, c1) : : (b4, c1) : (b4, c2) : : : : : : : (b5, c2) (b6, c2) : (b5, c3) : : (b6, c3) : : : : : (b6, c4) :

s4

(b3, c2) (b3, c3) : : : : (b3, c4) :

(b4, c3) : : : (b4, c4) : : : (b5, c4) : (b1, c8) :

(b2, c8) :

(b4, c6) (b4, c7) : (b4, c8) :

(b5, c7) (b5, c8)

Table 11. Simulated compilation of a set of C booklets (four scribes, six B booklets, eight C booklets).

48

Chapter 5 The surviving portion of the C text – Part II

The C booklets are not a masterpiece of calligraphy. Some of the scribes involved write rather badly; none of them seem to be trying to write very carefully. In different circumstances, most of them could probably have done better work than this; but here they had no reason to make any special effort. These booklets, as I understand it, were not expected to be kept for any length of time. Fairly soon, they would be superseded by the D text; once their usefulness had been exhausted, they would all be thrown away.1 The scribes had no idea that their work would be seen by outsiders, still less by future critics. Their instructions were, I suppose, to write as rapidly as they could, provided only that what they wrote was legible. Judged by that standard, they seem to have performed well enough. They got the job done, and there is scarcely ever any difficulty in reading what they wrote.

to complete the analysis. It seems clear enough why he stopped. He did not expect anyone to believe that a dozen or more scribes had worked on this manuscript; he was not even sure that he could believe it himself. But in fact it is certainly true. Ker (1977) said little about the script beyond commenting that there are ‘many rather poor hands of Norman type’ (1977, p. 807).3 In a footnote, however, he analyzed one sample booklet (above, p. 42), distinguishing eight different scribes in the stretch of text relating to Devon (376r–9v), where Finn would have seen only five. In the whole manuscript, Webber (1989) recognized at least fifteen scribes; I see twenty or more. My own results are tabulated in an appendix to this chapter (below, pp. 56–9). As a matter of policy, I worked out the analysis for myself, in a provisional way, before looking in detail at the conclusions reached previously by others; then I checked through the evidence again, wherever some difference of opinion seemed to arise.4 A fair number of corrections and improvements followed from that, but no large alterations. Table 12 shows how my identifications match up (or seem to match up) with those of some other investigators; the last column refers back to the list of published reproductions given in chapter 4. Despite its prolixity, the analysis is far from exhaustive: it ignores many short insertions by other hands, as well as all marginal additions.5 But there is a limit on how much can usefully be done by any

Changes of hand are very frequent in C – much more so than they are in the surviving D booklets (below, p. 134), by a factor of about 15. A large number of scribes contributed, and their contributions are woven together in a way which has more than once been called bewildering. Analysis is certainly laborious, but I do not know that bewilderment need last for long. On the whole, these scribes are experienced workers, with well-developed individual manners. Each of them writes in his own way; they do not appear to be making any attempt to imitate one another. The changes of hand are generally obvious, often glaringly so. By scanning repeatedly through the manuscript (or, as I have done, through a microfilm copy), jumping forwards and backwards from stint to stint, one soon begins to notice the characteristics which distinguish a particular hand. With enough perseverance, one reaches the point where almost every stint can be assigned to a recognized scribe.

1959, p. 367). 3

Ker’s attention was caught by a point which is, for present purposes, of only incidental interest. One of the scribes represented in the geld accounts for Wiltshire (my sigma) is a scribe whom Ker knew from elsewhere: the same man was also employed, with many others, writing books for Salisbury Cathedral (or, as I would think more likely, for bishop Osmund of Salisbury). Webber (1989, 1992) agreed with the identification made by Ker (1976, 1977); she also suggested that two or three other scribes who worked on Exeter 3500 could be recognized as having worked at Salisbury too. But this evidence does not tend to prove that the Exeter manuscript originated in Salisbury (i.e. in Old Sarum): it does not even raise the possibility. None of the three major scribes is Salisbury-connected; if a few of the minor scribes are, that is scarcely surprising. If I had been given the job of recruiting scribes for the survey, bishop Osmund is one of the first people to whom I would have turned for advice.

An article by Finn (1959) was the first attempt to identify the scribes individually. As far as it went, Finn’s analysis seems to me largely correct, but there are no illustrations, and it is sometimes hard to understand what he meant because the references to the manuscript are not sufficiently precise. Finn thought he could distinguish a dozen different scribes – but he lost heart, at around the time when he started coming across the work of a tenth scribe,2 and failed

4

I am greatly obliged to Dr Teresa Webber, who checked a version of this listing against her notes and discovered a number of errors. She allows me to say that my identifications, by and large, agree very closely with hers. A few points of disagreement are mentioned in the footnotes below.

1 I choose my words carefully here: it was not the intention that the C booklets should be discarded as soon as the D booklets had been finished (below, p. 55).

5

The text has been more or less heavily corrected throughout. On the whole, the corrections seem to have been made by the same scribe who wrote the original text, or by one of the scribes who is known to have been working with him; but some alterations and marginal notes are certainly by alien hands. This evidence needs to be looked at very closely, but I have not made any serious effort to deal with it.

2 The scribe in question is lambda, whose existence Finn doubted simply because he did not seem to have written enough. It was ‘not very likely’, he thought, that a scribe ‘would deal with only some twenty manors’ (Finn

49

The survey of the whole of England Finn (1959)

Williams (1968)

Ker (1977)

Webber (1989)

Reproductions (cf. Table 9)

beta alpha mu

A G

C A

1–2 3 6

A1 A2 A3

108r, 117r, 438r 103r, 117r, 438r 8r

eta omicron ksi

J

B

D2

=F

D

114v — 1v, 47r

epsilon gamma delta kappa

T S F

4 5 7

313r 313r 313r —

zeta

C

C

theta iota lambda

H D ?

= D1

rho sigma tau

E D1

245r — — — 8r 8r, 9r, 14r 8r, 9r, 14r

Table 12. Published identifications of the scribes represented in Exeter Cathedral Library 3500. (Scribe sigma is the ‘Salisbury scribe’ identified by Ker (1976).) I suspect that mu may have been the man in command.8

one person, and I think that I have reached it. Several people will need to have looked at the evidence before a fair measure of consensus can emerge. That is why I publish my results in full, despite the amount of space which they take up. I would rather risk wasting space than risk wasting the time of anyone else who may wish to check this analysis, and (if it passes the test) to build upon it.

The minor scribes, by and large, worked on only one county text each.9 A few small exceptions to this rule would not be disconcerting;10 but some apparent anomalies turn out, when looked at more closely, not to be exceptions after all. (1) One entry occupying a leaf by itself (398r1–7), supposedly part of C-Dn, was written by scribe omicron, whose other stints are confined to C-Do. There is, however, some reason to think that this entry became displaced, in D and therefore in DB, and that it was indeed originally part of omicron’s contribution to C-Do (above, p. 44). (2) One entry in C-Dn (98r15–22) was written by two scribes who otherwise do not occur outside Co. Scribe zeta wrote the first two lines; the rest was written by a very poor scribe whose only other appearance is a seven-page stint of C-Co (259rl– 62r9). The placement of this entry allows us to think that

Three scribes stand out from the crowd, because they contributed to several county texts, not just one or two. Of these three major scribes, two are especially conspicuous, alpha and beta. They each have a highly individual hand; they each wrote a large proportion of the text, much larger than any of the minor scribes who from time to time worked alongside them. Though neither of them wrote the sole surviving Wiltshire entry, it would not seem rash to assume that they wrote large parts of the lost C-WiDo booklets, as they certainly did of the surviving (Do-only) booklets. It was Finn’s (1959, pp. 367–8) suggestion that alpha – his clerk G – had some supervisory role; I am inclined to agree.6 The third scribe, mu, wrote very much less than alpha or beta, but I treat him as one of the major scribes because he worked on every county text. In fact, if we allow ourselves to look beyond C for a moment, mu is the only scribe who can be said for certain to have worked on the record for all five counties covered by the Exeter booklets.7 Though alpha and beta did most of the actual work,

Do, Dn and So. Some additions he made in the margin of 8r are the only published sample of his work (Table 9): here he is writing small, but in his most formal manner. The fancy & is his signature, when it occurs; sometimes he used the 7-shaped sign instead. 8

To speak plainly, I suspect that mu was the treasurer – in which case his name was Henric (DB-Ha-49ra). But this suggestion is hardly worth making until it has been agreed that Exeter 3500 originated in Winchester.

9

6 The best evidence for the pecking order will come, I expect, from a study of the corrections, about which I cannot speak with any assurance.

Finn (1959) seems to have approached this conclusion but then backed away from it. Trying to keep the number of scribes as small as he possibly could, he convinced himself that he could recognize contributions by clerk S in Co as well as Dn, by clerk J in Co as well as Do and So, and (more tentatively) by clerk F in Do as well as Dn. None of these identifications seem justified to me.

7 Scribe mu made important additions to the second version of the Wiltshire geld account; as was said above (pp. 41–2), he also wrote the statistical summary (527v–8r) covering the lands of Glastonbury abbey in Wi,

10 One such anomaly is the final paragraph in Capp-Dn (506v1–5), written by scribe iota, who is otherwise only represented in C-So. Presumably this paragraph was added as an afterthought.

50

The surviving portion of the C text – Part II it may have been an addition, and that is almost certainly what it was – an entry inserted here later, in a convenient space, while work was in progress on C-Co.11 (3) The sole surviving entry for Wi (47r1–11) was written by scribe ksi, who also contributed to C-Do.12 That is odd; but the mere fact that this entry survives marks it as a special case, and one cannot think of arguing anything from it.13

C-Dn, the pattern is very clear, and was recognized well enough by Finn. The minor scribes here are three again: epsilon, gamma, delta. For C-Co, the quantity of text is relatively small, and alpha and beta wrote a larger share of it than usual. There is only one minor scribe who occurs here often enough to be given a name, and that is zeta; but two other scribes write one large stint apiece (259rl–62r9, 263rl–4r20).17 For C-So, again, the same sort of pattern exists, though the number of minor scribes occurring here is (as Finn suspected) four rather than three: theta, iota, eta, lambda.18 However, there is only one booklet (fos. 456– 67) in which the last two scribes appear together – lambda wrote an early stint (463r5–v15), eta wrote two later ones (464r18–22, 465r2–v7) – so it seems quite possible that lambda left the squad, for one reason or another, and that eta (the same scribe who had worked on C-Do previously) was brought in to take his place. On that view, the number of minor scribes at work simultaneously would not have exceeded three.

There is, as far as I can judge, only one minor scribe who wrote more than one entry in more than one county text. The scribe whom I call eta wrote a good share of the entries in C-Do; he also wrote a good share of the entries in C-So. Here I am gratified to discover a large measure of agreement between my results and those of both Finn (1959) and Webber (1989): my eta is roughly the same as Finn’s clerk J, almost exactly the same as Webber’s scribe D2.14 It is not to be thought that there was any prohibition against a minor scribe participating twice, but this seems to me to be the only discoverable instance of a minor scribe actually doing so. For that reason I have looked at the evidence with special care, and (to put the result in a suitably negative form) have failed to find any significant consistent difference between the scribe who worked on C-Do and the scribe who worked on C-So. I am satisfied that they are the same man.

I am not proposing to press this point very far. I do not suggest that there was any rigid rule that a squad should consist of exactly three scribes, and always exactly the same three. Some flexibility would obviously be desirable, and was no doubt permitted. But it does seem clear, generally speaking, that the minor scribes were organized into threeman squads – that three, by and large, was thought to be a suitable number, and that scribes who were accustomed to working together were, by and large, allowed to continue doing so. Again, I am not suggesting that there was a different squad of three scribes for every single county text. On the contrary, I take it that each squad would have been employed successively on several texts, in whatever sequence was dictated by the flow of the work. In the surviving booklets, we see four of these squads each writing itself out of a job by completing the county text which formed its final assignment.19

Table 13 gives a summary of the results listed in the appendix, for the C booklets alone (including Capp).15 Ignoring C-Wi, we do not have to look very hard to see some pattern here. As might be expected, the pattern is most distinct in the two counties – Devon and Somerset – for which the evidence is fullest; but it is discernible elsewhere too, to some extent. Two major scribes, alpha and beta, are jointly represented in all four county texts. In each they are assisted by a different squad of minor scribes, of whom normally there seem to have been three. The details are as follows. For C-Do, only a fragment of the text survives, but there is no reason why the fragment, in this respect, should not be representative. The minor scribes occurring here are three: eta, omicron, ksi.16 For

How many squads there were, and how many scribes alto-

11

The entry relates to Werrington. As Finberg (1944) pointed out, it is clear that this manor was initially surveyed as part of Cornwall but then recorded under Devon. In C we can watch that change taking effect: two of the scribes working on C-Co add this entry to C-Dn, rather than including it in their own text.

were attributed by Webber (1989, p. 12) to a scribe who worked on the Wiltshire geld accounts, the scribe whom I call tau. I do not feel confident that the hand is the same, though it is certainly very similar. The disagreement extends further than that, because the scribe whom I call theta (represented only in C-So) is, in Webber’s judgment, the same as scribe tau, not somebody else. For the moment I leave these questions unresolved, intending to come back to them later when I deal with the geld accounts. ((Some further comments will be found in chapter 6 (below, p. 66).))

12

He also wrote most of the first Wi geld account (1r–3r), to be discussed in chapter 6. 13 What happened, I would guess, is that this entry had to be recopied for some reason (perhaps because it had been included in an omnibus booklet at first), after work had started on C-Do.

17 These stints are the ones attributed by Finn to clerk S and clerk J respectively (above, note 9). (For the latter attribution see Finn 1959, pp. 382– 3.) Webber (1989, p. 12) assigned the second stint to her scribe C, who is otherwise the same as my zeta (Finn’s clerk C). The hand looks different to me.

14

Webber (1989, p. 13) gives a list of the stints she assigns to scribe D2. Not counting two marginal additions (which I do not include, though I agree that they are eta’s work), there are only two small discrepancies between her listing and mine (36v19–20, 374v14–15). The same hand was identified by Webber (1992, pp. 12–13) in several manuscripts from Salisbury.

18

The three scribes who worked on C-So alone are not represented in any published reproduction. Perhaps it may save somebody some time if I note that there is just one page (286v) on which all three hands occur together. 19 A third reservation: I am not suggesting that these squads worked only at headquarters and only on C. I am more than willing to believe that each squad may have spent some of its time working with the commissioners in the field, putting together the B text. ((The role that they would have played is discussed in chapter 10.))

15

It is plain to see that the same teams of scribes who worked on the county texts in C-DnCoSo worked on the corresponding sections of CappDnCoSo too. 16

Two stints in C-Do which I have left unattributed (37v3–8r7, 51r17–v6)

51

The survey of the whole of England Wi

Do

Dn

Co

So

Total

beta alpha mu

17 12 1

93 83 6

7 9 2

67 70 2

184 174 11

eta omicron ksi

15 9 7

16

31 9 8

1

epsilon gamma delta kappa

73 71 36 3

zeta

1

theta iota lambda

1

unattributed Total

1

73 71 36 3 4

5 54 24 10

54 25 10

5

5

3

6

19

66

372

25

249

713

Table 13. Numbers of stints performed by each scribe in each section of the C text, summed from the listing given in the appendix. (Stint 398r1–7 is counted under Do.) gether, is hard to decide.20 Four squads are represented in the surviving portions of C, a fifth perhaps in the batch 4 geld accounts, i.e. the accounts for Wiltshire.21 There are, besides, several stints in C which I have left unattributed. In some cases, the script resembles that of an identified scribe, and might perhaps be attributed to him if we were willing to stretch the definition slightly.22 Still, there are several stints of which I think it can be said with confidence that they were not written by any of the major or minor scribes mentioned above. There are, so to speak, some occasional scribes who make sporadic appearances in C; and possibly these are members of other squads, lending a hand here when they have no work of their own to keep them busy. At the very least, four occasional scribes occur. One of them writes three short stints in C-Dn, just enough to deserve a name (this is the scribe whom I call kappa); two others write one stint each in C-Co (see above); and the fourth – with a small and rather elegant hand – writes one sizable stint in

C-So (430v2–1v9).23 In total, therefore, I think I can recognize twenty different hands: three major scribes, thirteen minor scribes (including the three who occur in the batch 4 geld accounts), and four others (including kappa).

20

When Ellis set about rearranging the text,24 he seems to have assumed that the preexisting sequence made no sense at all. He changed it without recording it. Nobody working from the printed text could tell how much of the arrangement originated with the editor. If the leaves had not been numbered previously, there would be no way to undo what Ellis did; because they had, we can – virtually – put the quires back into the sequence which existed before 1816. If we do that, as Whale (1905) did, we discover that the

There is no order for the C booklets which is absolutely right. On the contrary, one advantage of dividing the text into self-contained units of this kind was to make it possible to sort and shuffle the booklets into different arrangements, depending on the task in hand. The scribes assigned to write D-Dn, for instance, would separate out the booklets that they did not need, and impose some appropriate order on the ones that they did; the scribes assigned to write DCo would resort the booklets and make their own arrangement of the ones which interested them. Given some definite task, one can start deciding which order would be best. Without knowing what the task is, one cannot.

If I had been in charge, perhaps I might have thought of employing eight squads and assigning four counties to each. To make up these squads, therefore, I should have had to recruit a total of 24 scribes. Some strategy not very different from that lies behind the division of labour observable in the surviving booklets.

21

Not counting mu, three scribes worked on the second version of the Wiltshire account (7r–9v), two of whom also wrote the third version (13r– 16r). It is clear, by the way, that the third version is a fair copy of the second version. The order in which the batch 4 booklets were arranged by Ellis (1816) is the right order; Darlington (1955) misread the evidence. ((Anyone who doubts this will find the proof at the end of chapter 6.)) 22

In doubtful cases like this, I have preferred to err on the side of caution. For example, one unattributed stint (373r3–v12) is very similar to iota’s work, and is indeed cited as a specimen of it by Finn (1959, p. 367). It does not look quite right to me; but perhaps the differences might be explained away. (The light was bad, the scribe was using a borrowed pen – some ad hoc conjecture of that sort might be enough.)

23 As far as this fourth scribe is concerned, Dr Webber allows me to say that she too thought that this was his only stint. 24

Some of the work seems to have been done by Barnes (above, p. 39), but he was just doing what Ellis wanted done.

52

The surviving portion of the C text – Part II lets became split between two stacks.27 The booklet for the bishop of Coutances, cited previously (above, p. 38) as an example of the disruption which Lyttelton tried to undo, can be cited again here. It consists of six quires. Quires 3–6 contain portions of the C-So text, so they belong in stack 2, even though quire 3 (fos. 135–40 in the old foliation) starts with the broken-off end of the C-Dn text. Quires 1–2 relate to Dn alone, so they belong in stack 4.28

arrangement was far from being senseless: the booklets had been put into a task-specific order. This order was not meant to be permanent – it only became permanent when the manuscript was bound – and Ellis was, arguably, entitled to impose a new arrangement which would be generally more convenient. By choosing to do this, however, and then by failing to explain what he had done, he made a significant aspect of the evidence entirely invisible in the printed edition. Once Barnes had imposed the same arrangement on the original manuscript, the same aspect became invisible there as well.25

Up to a point, it is clear what the stacking means. Some operation was being performed on each county text in turn, and the C booklets were being sorted into stacks for that purpose. As far as these five counties are concerned, the division into stacks suggests that the intended sequence was this: Wi, then So, then Do, then either Dn or Co. There is one weak link here: we cannot feel sure that So was intended to be dealt with sooner than Do. (We could only be sure of that if we had some guarantee that the order of the stacks remained unaltered until it was permanently fixed when the booklets were bound.) It is certain, however, from the way in which the C-WiDo booklets were divided between stacks 1 and 3, that Wi was intended to be dealt with sooner than Do, and equally certain, from the way in which the C-DnCoSo booklets were divided between stacks 2 and 4, that So was intended to be dealt with sooner than Dn and Co.

The task-specific order was, as I have said, rediscovered by Whale (1905). He did not, as far as I can see, have any idea what it meant; but he did realize that there was a pattern in the evidence, obliterated by Ellis’s rearrangement, which could be recovered with the help of the old foliation. The pattern is not perfect – but it is perfectly obvious (Table 14. Some time before they were bound, the C booklets had been sorted into four stacks, as follows: the C-WiDo quires relating to Wi (including those which also related to Do) were put into stack 1; the C-DnCoSo quires relating to So (including those which also related to Dn or Co or both) were put into stack 2;

Suppose that the operation – whatever it was – had continued running smoothly. The sequel would have been this. When work on Wi was finished, the booklets from stack 1 would have been resorted: those which included some section of C-Do would have been added to stack 3 and the rest (relating to Wi alone) discarded (as all the preceding C booklets already had been). When the moment arrived for work to start on Do, stack 3 would now be ready. Similarly, when work on So was finished, the booklets from stack 2 would have been resorted: those which included some section of C-Dn or C-Co would have been added to stack 4 and the rest (relating to So alone) discarded. Eventually it would have had to be decided whether Dn or Co was to be dealt with first, and a fifth stack would then have been formed accordingly – a stack which, when first created, would consist of the booklets relating to Co alone, if Dn was to be dealt with first, or of the booklets relating to Dn alone, in the opposite case. But that decision had not yet been made, or, if it had, had not yet issued in action.

the C-WiDo quires not in stack 1 were put into stack 3; the C-DnCoSo quires not in stack 2 were put into stack 4. Stack 1 does not survive (in due course I propose to ask why); stacks 2–4 do survive, and in that order, sooner or later, they were bound. The plan is clear, even though its execution was not altogether perfect.26 A few mistakes were tolerable, so it seems. If one or two unwanted leaves were included in stack 2, say, that would cause no trouble. If a few quires which ought to have been included were overlooked, that did not matter greatly: their absence would be detected, once the task for which this stack was intended was under way, and any missing quire could be found at that stage, just by flipping through stack 4. Similarly, it does not seem to have been thought important for the quires to be kept in the right order: on reaching the end of one quire, it would be easy enough to find the next, just by flipping through the rest of the stack.

27

The same sort of split has affected Capp-DnCoSo. This booklet consists of four quires. Quires 2–4 relate to So, so they belong in stack 2, even though quire 2 contains the broken-off end of Capp-Dn and the whole of Capp-Co, as well as the beginning of Capp-So. Quire 1 relates to Dn alone, so it belongs in stack 4. Once again, this booklet proves to have had the same history as the other C booklets. (It also turns out that we do not need to think of any special explanation for the disappearance of Capp-WiDo. That booklet would have been put into stack 1; if we can think of some reason for the loss of this stack, that reason will cover the loss of Capp-WiDo too.)

One result of the sorting was that some of the larger book25

There are two issues here. (Q1) Was Ellis justified in rearranging the text? (Q2) Was Barnes justified in rearranging the original? Briefly put, my answers would be as follows. (A1) Certainly he was; in fact he ought to have rearranged it much more thoroughly than he did. (A2) Probably not.

26

Anomalies are few, and perhaps we might think of explaining them away as accidental displacements, occurring while the booklets remained unbound. But that seems facile to me. I think we had better accept that the sorting was not very carefully done in the first place.

28 This split explains only some of the disruption. In addition, quires 4–5 have got themselves transposed with quire 3, and quire 6 has wandered off by itself.

53

The survey of the whole of England

Stack 2 247–50 ix–xii So 437–40 251–7 1–7 So 430–6 258–60 8 So 375 261–8 9–12 DnSo 371–4 13 ? 398 269–71 * * 14–16 DnSo 196–8 272–5 17–24 DnSo 456–63 280–3 284–7 25 So 154 26 So 116 27 So 193 28–34 DnSo 468–74 35–6 DnSo 286–7 37–40 So 282–5 41–2 So 191–2 43–8 DnSo 376–81 49–51 So 185–7 52–4 So 151–3 55-60 DnSo 382–7 61–5 DnSo 366-70 66–85 CoSo 255–74 86–91, 93 So 275–81 94–101 So 422–9 102–5 So [176], 188-90 106–12 So 169–75 113–20 DnSo 83–90 121–2 So 91–2 123–30 So 139–46 131–4 So 147–50 135–40 DnSo 133–8 141–8 CoSo 99-106 149–56 DnCoSo 503–10 157–64 So 511–18 165–71 So 519–25 172–6 So 441–5 177–80 So 446–9 181–2 So 450–1 183 So 107 184–6 So 353–5 187–9 So 113–15 190–5 DnSo 475–80 196–203 DnSo 356–63 * 204 Do 62 207–12 DnSo 490, 489, 491–4 213–16 So 452–5 217–18 So 364–5 219–26 DnSo 345–52 227–30 So 464–7 231–6 So 155–60

Stack 4

Stack 3 Do Do Do Do WiDo Do Do Do

25–8 29–35 36–8 39–46 47–9 50–3 54–7 58–61

307–10 318–21 322–9 330–7 * 338 339–40 341–5 346–7 348–55 356–65 366–73 374–7 378–9 380–7 388–92 393 394–401 402–9 410–17 * 418–21 422–7 428–9 430–7 438–42 443–50 451–8 459–61 462–9 470–8 479–80 * 481–4 485 486–9 490–1 492 493–4 495–8 499 500–1 502 503–6 507–12 513–16 517–20

Dn Dn Dn DnSo Dn Dn DnCo Dn Dn Co Co Dn Dn Co Co Co Dn Dn Dn DnSo Dn Co Co DnCo Dn Dn DnCo Dn Dn DnSo Dn Dn DnCo Dn Dn Dn Dn Dn Dn Dn Dn Dn Dn Dn

117–20 121–4 125–32 161–8 184 194–5 177–81 182–3 210–17 224–33 234–41 220–3 218–19 247–54 242–6 199 288–95 296–303 304–11 312–15 93–8 200–1 202–9 108–12 495–502 316–23 332–4 324–31 335–6, 408, 337–42 343–4 388–91 392 394–7 420–1 419 409–10 399–402 393 481, 488 407 411–14 482–7 415–18 403–6

Table 14. The quires of C restored to the sequence recorded by the foliation of circa 1500. (The old foliation is given on the left, the new foliation on the right; the counties represented are listed in the middle. Anomalies are marked with a star.)

54

The surviving portion of the C text – Part II Of the things which would have started happening as soon as Wi had been dealt with, none actually happened. Hence it is clear that the operation was interrupted, and that the interruption occurred while work was in progress on Wi.29 The four counties last in the queue were left untouched. We can be sure of that; we can also be sure that no other operation ensued which required a rearrangement of the C booklets. The untouched stacks remained untouched, and this task-specific arrangement – expected to be temporary, and not very carefully worked out – became permanent by default (till Ellis decided to change it).

the same chance of long-term survival as any other book in the Exeter cathedral library. I do not know that we shall ever be able to say who the man was who retrieved these booklets from the Treasury shelves and found a safe home for them elsewhere. If we want to try guessing, there is something to be said for preferring the earliest possibility. The sooner we can extract these booklets from a milieu where their chances of survival are almost nil, the sooner we can insert them into a milieu where their chances are fairly good, the less utterly unlikely it will seem that they still exist. In that abstract sense, an early date for the transfer is more probable than a late one. For what it is worth, my own guess would be that the mystery might cease to be a mystery if we knew just a little more about the early career of Willelm de Warelwast, the king’s clerk rewarded for his services by being made bishop of Exeter in 1107.

The C booklets existed, in the first instance, so that the text could be copied from them into the D booklets. But that was not the operation for which these stacks were set up. If the writing of D had been interrupted as this operation was, D-Wi would have been the last D booklet to be written; and in that case DB-Do, DB-So and DB-DnCo would not exist. Since they do exist, the operation in question has to be different from the writing of D – different from and subsequent to it. Thus it is clear that the C booklets were not discarded as soon as they had served their primary purpose. They needed to be kept, at least a little longer, because they were going to be used again, for some secondary purpose; and the division into stacks was made with this purpose in view. There are two possibilities, not mutually exclusive. The C booklets may have been kept because they were going to be used when D was checked; alternatively or in addition to that, they may have been kept because they were going to be used when DB was checked.30 We cannot hope to decide between these possibilities until we have developed some theory as to how the checking process might have worked.

Appendix Booklets and scribal stints in Exeter Cathedral 3500 If it is read alongside Ker’s (1977) description, the list which follows ought to be self-explanatory; but it may be helpful to clarify a few points first. (i) In numbering the lines, I have followed the ruling: in other words, I have counted blank and erased lines, not just written lines. (ii) I have ignored some of the headings, wherever I am not certain that the heading was written by the same hand that wrote the following entry. (iii) I have ignored all insertions made into the text, into spaces or over erasures, as well as all marginal additions and annotations.

Before the interruption, stack 1 had been removed – taken from the shelf (so to speak) and carried off to some other part of the office, wherever the work was being done for which these booklets were needed. After the interruption, stack 1 was not returned; sooner or later, all these booklets dropped out of existence. Stacks 2–4, still on the shelf, had a marginally better chance of survival, and did indeed survive – long enough for someone to rescue them and carry them away, on a journey which ended, sooner or later, in Exeter. Once there, they became part of a library: they entered an environment in which it was taken for granted that loose quires should be bound and that books should be kept for ever. The collection of booklets became a book, with

The attributions made here all seem secure to me. There are no question marks. Wherever I feel any doubt, I have chosen to err on the side of caution by leaving the stint unattributed.

29

I risk speaking loosely here. To describe it more carefully, the situation is this. The last sort which did happen was the sort which had to be done before work on Wi could start. Because Wi was the first county in the next collection of C booklets to be dealt with, the sort could have been done some time in advance (in the same way that stack 2 has already been separated from stack 4). The first sort which did not happen was the sort which would have had to be done before work on Do could start. The operation was interrupted before that moment arrived. (It is possible, however, that the work did not cease altogether: it may have been continued on some simplified plan – if such a plan is conceivable – which meant that it was no longer necessary for the C booklets to be consulted.) 30

Perhaps we should not neglect a third possibility: that they were going to be used for checking the geld accounts. ((It seems to me now that this is the likeliest explanation (below, pp.131–2).))

55

The survey of the whole of England

C-WiDo King Do 25r1–10 25r11–v19 25v20–8r20 28v1–4

alpha eta beta eta

King (queen Mathildis) Do 29r1–v14 beta 29v14–16 — 29v17–30r10 omicron 30r10–v9 eta King (queen’s knights) Do 31r1–6 alpha 31r7–2r12 omicron Countess of Boulogne Do 33r1–16 alpha 33r16–20 omicron St Peter of Cerne Do 36r1–v1 beta 36v1–11 alpha 36v12–18 eta 37r1–9 alpha 37r9–10 beta 37r11–15 eta 37r16–v2 beta 37v3–8r7 — 38r7–17 ksi 38v1–6 alpha 38v7–20 beta St Peter of Abbotsbury Do 39r1–8 eta 39r9–40v4 beta Abbot of Athelney Do 41r1–9 omicron Abbot of Tavistock Do 42r1–8 ksi 42r8–14 beta St Peter of Milton Do 43r1–v1 beta 43v1–8 eta 43v9–19 alpha 43v20–4v3 beta 44v4–5r10 omicron Willelm de Moion Wi 47r1–11 ksi Do 47r12–8r3 ksi 48r3–20 beta 48v1–9r12 ksi 49r13–20 beta 49v1–16 — Roger Arundel Do 50r1–16 beta 50r16–v3 ksi

50v3–10 50v10–1r17 51r17–v6 51v6–2r8

eta ksi — omicron

Serlo de Burci Do 53r1–15 omicron Hugo son of Grip’s wife Do 54r1–18 alpha 54r19–v7 eta 54v8–5v20 alpha 56r1–6 eta 56r7–17 alpha 56r17–7r6 eta 57r7–15 alpha 57r16–19 beta 57v1–20 alpha 58r1–9 beta 58r10–12 eta 58r12–18 ksi 58r18–v3 omicron 58v4–9r1 beta 59r2–v17 eta 59v17–19 beta 60r1–v16 mu 60v16–1v17 eta 61v18–20 —

C-DnCoSo alpha epsilon alpha gamma beta epsilon beta alpha epsilon alpha beta delta beta alpha epsilon alpha theta alpha theta alpha

King Dn 93r1–4r5 94r6–22 94v1–7 94v8–15 94v16–5r20 95v1–6r3

beta alpha beta epsilon gamma alpha

gamma epsilon beta epsilon alpha epsilon zeta — alpha alpha beta alpha beta eta

King (queen Mathildis) Dn 108r1–v5 beta 108v6–9r12 alpha 109r13–v9 beta 109v10–10r4 gamma 110r5–13 epsilon 110r14–19 alpha 110v1–8 epsilon 110v10–17 gamma 110v18 alpha 110v18–11r7 epsilon 111r7–15 delta Co 111v1–12r18 alpha King (queen Edit) So 113r1–9 beta 113r10–v6 alpha 113v7 beta 113v7–14r15 alpha 114r15–v9 eta 114v13–19 theta 114v20–15r3 beta

Walter de Clavile Do 62r1–v17 eta

King Dn 83r1–7 83r8–19 83v1–17 83v18–4r4 84r5–v4 84v5–20 85r1–8 85r9–15 85r16–v1 85v2–14 85v15–6v12 86v13–7r16 87v1–13 88r1–3 88r3–10 So 88v1–9r1 89r3–10 89r11–90r20 90v1–8 90v8–1v10

96r9–17 96r18–v2 96v3–7r10 97r10–v9 97v10–8r4 98r4–14 98r15–16 98r16–22 Co 99r1–102v14 So 103r1–v15 103v16–5v6 105v7–6r4 106r5–7v9 107v9–12

King (Ulward Wit) So 116r1–17 alpha 116r18–v2 theta Bishop of Exeter Dn 117r2–4 beta 117r5–v14 alpha 117v15–18r3 beta 118r4–11 alpha 118r12–17 beta 118r18–v3 alpha 118v3–10 delta 118v11–19r8 beta 119r9–20r16 gamma 120r17–v2 alpha 120v3–10 gamma 120v11–16 kappa 120v17–20 beta Co 199r1–2 alpha 199r2–13 beta 199r14–201r20 alpha Bishop of Coutances Dn 121r1–v10 epsilon 121v11–2v9 alpha 122v10–3r6 gamma

56

123r6–19 123r20–v11 123v12–15 123v16–4r14 124r15–v12 124v13–20 125r1–18 125r19–8r8 128r9–10 128r10–14 128r14–31r11 131r12–2r10 132r10–20 132v1–17 132v18–3v20 134r2–v6 134v7–19 135r1–6 135r7–11 135r11–16 135r17–v11 135v12–6r8 So 136v1–13 136v14–7r3 137r4–10 137r11–19 139r1–40r19 140r20–1v2 141v2–6 141v7–11 141v12–5v4 145v5–8r3 148r4–52r15

— epsilon gamma beta alpha beta alpha beta gamma epsilon beta alpha gamma beta alpha gamma alpha beta epsilon delta epsilon beta theta alpha beta theta beta iota theta alpha theta eta beta

Bishop Osmund So 154r1–14 theta 154r15–v2 beta Bishop Giso So 156r1–v6 156v6–7r5 157r6–60r20

eta alpha theta

Abbot of Glastonbury Dn 161r1–2 epsilon 161r2–8 delta So 161r8–9 alpha 161r9–19 theta 161v1–5 eta 161v6–12 alpha 161v13–8v19 beta 169r1–70r17 alpha 170r18–v17 beta 170v17–3r5 eta Abbot of Tavistock Dn 177r1–v12 alpha 177v12–19 beta 177v20–8v12 alpha 178v13–9r5 beta 179r6–v5 gamma 179v6–12 epsilon 179v12–16 delta 179v17–80v3 beta

The surviving portion of the C text – Part II Co 180v13–1r13 181r13–v10

zeta alpha

Abbot of Buckfast Dn 182r1–14 epsilon 182r15–19 alpha 182r20–v12 gamma 182v13–3r5 alpha 183r6–v15 beta Abbot of Horton Dn 184r1–v4 alpha St Peter of Bath So 185r1–7r5 eta St Peter of Muchelney So 188r1–7 iota 188r8–9r12 theta 189r13–v3 beta St Peter of Athelney So 191r1–9 iota 191r10–19 theta 191r20–v12 eta 191v13–20 beta Abbess of St Edward’s So 193v1–9 alpha Saints Dn 194r1–10 194r11–v11 194v12–18 194v19–5r7 195r8–12 195r12–v6 195v6–6r1 196r2–3 So 196r10–14 196r15–17 196r17–v8 196v9–13 196v13–7r3 197r4–9 197r10–v5 197v7–20 198r1–11 198r12–19 198v1–2

alpha beta alpha beta gamma epsilon delta beta theta alpha beta alpha beta theta beta alpha theta alpha beta

fos. 199–201 should follow fo. 120 Saints Co 202r1–8v9

alpha

Count of Mortain Dn 210r1–9 epsilon 210r10–12v8 beta 212v8–12 epsilon 212v13–13r19 beta 213r20–v13 epsilon 213v14–14r1 gamma 214r2–6 alpha

214r7–v20 215r1–16v3 216v4–9 216v9–14 216v15–17r16 217r17–v3 217v4–19r7 219r7–v18 219v19–22r9 222r10–3r2 Co 224r2–5v7 225v8–33v20 234r1–43v20 244r1–5v16 247r1–54v20 255r1–8v22 259r1–61r20 261v1–7 261v8–2r9 262r9–v15 263r1–4r20 264v1–5r2 So 265r8–20 265v1–20 266r1–v7 266v8–7v20 268r1–7 268r7–10 268r13–19 268v1–9v7 269v7–9 269v9–71r20 271v1–2v19 272v20–3r6 273r7–5v6 275v6–6v9 276v10–19 277r1–8 277r9–18 277r19–v15 277v16–8r20 278v1–81r18

gamma beta gamma epsilon delta beta alpha epsilon beta delta beta zeta beta zeta beta alpha — alpha — beta — beta alpha theta alpha beta theta alpha theta alpha eta alpha beta alpha eta theta alpha iota alpha iota alpha beta

alpha epsilon gamma epsilon alpha epsilon delta epsilon delta epsilon delta epsilon alpha gamma delta epsilon gamma delta beta beta

Juhel Dn 316r2–12 316r12–18r4 318r4–v3 318v4–19v13 319v14–15 319v15–20r2 320r3–3v18 323v18–4v4 324v4–9 324v9–6r19 326v1–31r17 331r18–v16 331v19–2v23 333r1–4r17 334r17–v4 334v5–6 Co 334v10–16

beta alpha beta epsilon alpha gamma beta delta kappa delta epsilon delta epsilon delta epsilon beta mu

Radulf de Pomerei Dn 335r1–5 gamma 335r6–6r12 alpha 336r13–17 beta 336r18–v17 epsilon 337r1–v5 gamma 337v6–12 alpha 337v13–8r17 gamma 338v1–15 beta 338v15–9v17 delta 339v18–40r15 beta 340r15–19 gamma 340r20–1r7 beta 341r8–v2 epsilon 341v3–12 gamma 341v13–20 beta 342r1–5 gamma 342r6–17 alpha 342r17–v13 beta 342v13–3v7 delta 343v7–15 kappa 343v16–4r1 alpha 344r2–3 beta So 344r4–11 beta 344r12–19 lambda

Count Eustachius So 282r1–v11 beta 282v12–3r7 alpha 283r7–12 theta 283r13–19 alpha Earl Hugo Dn 286r1–13 286r14–20 So 286v1–7 286v7–12 286v13–17 286v18–7r3

298r1–9v15 299v16–301r9 301r10–v2 301v3–8r11 308r11–9v4 309v5–14 309v14–20 310r3–14 310r15–18 310r18–v4 310v4–11v5 311v5–12r9 312r10–15 312r15–v7 312v8–13r5 313r5–10 313r11–14v9 314v10–15r2 315r3–7 So 315r12–v10

alpha beta lambda iota alpha theta

Baldwin the sheriff Dn 288r1–90r9 beta 290r10–1v11 gamma 291v11–14 beta 291v14–4v16 gamma 294v17–7r4 epsilon 297r5–v22 beta

57

Walscin de Dowai Dn 345r1–6 gamma 345r7–13 beta 345r14–v6 alpha 345v6–6v8 gamma 346v9–7r8 alpha 347r9–18 epsilon 347r19–v7 alpha 347v8–14 beta 347v14–9v5 epsilon 349v6–14 beta So 350r1–v10 beta 350v11–19 theta 350v20–1r13 iota 351r14–2r2 alpha 352r2–7 theta 352r8–v2 beta 352v2–5v9 theta Willelm de Moion Dn 356r1–8 epsilon So 356r9–v12 beta 356v13–7r7 alpha 357r7–60v5 beta 360v5–12 lambda 360v13–2v10 theta 362v11–3r20 beta 363v1–5 iota 363v6–13 beta 363v14–4r18 iota 364r19–v17 beta 364v18–5r3 alpha Willelm de Faleise Dn 366r1–v18 mu 367r1–20 alpha 367v1–8v19 beta 368v19–9r6 delta So 369r6–18 beta 369r18–v5 — 369v6–20 beta Alvred de Hispania Dn 371r1–12 epsilon 371r13–18 gamma So 371v1–2r3 alpha 372r4–3r3 beta 373r3–v12 — 373v13–17 beta 373v17–4r6 theta 374r7–v12 beta 374v13–14 iota 374v14–15 eta 374v17–5r4 beta 375r5–11 alpha Odo son of Gamelin Dn 376r1–21 beta 376v1–7 alpha 376v8–11 epsilon 376v12–7r5 alpha 377r6–16 gamma 377r17–v20 alpha 378r1–5 mu

The survey of the whole of England 378r5–20 378v1–14 378v15–18 378v18–20 379r1–2 379r3–14 379r15–20 379v1–16 So 380r1–6

gamma delta gamma delta epsilon gamma beta gamma lambda

Turstin son of Rolf Dn 382r1–9 beta So 382v1–3r5 theta 383r6–4v3 alpha 384v4 beta 384v4–12 alpha 384v13–17 theta Willelm son of Wido So 386r1–19 iota 386r20–1 beta Goscelm and Walter Dn 388r1–9 epsilon 388r10–16 alpha 388r17–v8 epsilon 388v9–9r3 beta 389r4–91r9 gamma 391r10–14 beta 391r15–v11 epsilon 391v12–2r10 beta 392r11–17 alpha 392r18–v15 beta 392v16–3r15 alpha 393r16–5v17 delta 395v18–6r9 epsilon 396r10–v19 delta 397r1–5 alpha 397r5–16 epsilon 397r17–20 alpha Co 397v1–6 beta

?

Goscelm of Exeter 398r1–7 omicron

Willelm Capra Dn 399r1–v11 399v12–401r7 401r7–2v17 402v18–400v4 400v5–403r9 403r10–v2 403v3–12 403v12–4r2 404r2–5r8 405r9–14 405r15–v3 405v4–11 405v12–17 405v18–6r4 406r5–6

beta gamma epsilon gamma beta delta gamma beta delta beta gamma delta gamma delta beta

fo. 400 should follow fo. 402

438r13–v1 438v2–14 438v15–9r4 439r5–14

Tetbald son of Berner Dn 407r1–8r2 alpha 408r3–5 epsilon 408r6–9r1 gamma 409r2–7 epsilon 409r8–v6 beta 409v7–10v4 alpha 410v5–6 beta Ruald Adobed Dn 411r1–12 411r12–14r8 414r9–21 414v1–17 414v17–20 . . .

alpha iota beta theta

Roger Arundel So 441r1–v13 beta 441v14–19 alpha 442r1–21 theta 442v1–6 lambda 442v6–8 beta 442v9–3r8 alpha 443r9–v9 beta 443v9–14 alpha 443v15–20 iota 443v20–4 alpha 444r1–17 beta 444r18–v4 alpha 444v4–5r2 iota 445r3–8 theta 445r9–14 iota

beta epsilon beta alpha delta

Willelm de Poilli Dn 415r2 alpha 415r2–v18 beta 415v19–16v16 gamma 416v17–17r2 beta 417r2–9 delta 417r10–18r4 alpha

Gislebert son of Turald So 446r1–5 theta 446r6–15 alpha Osbern Gifard So 447r1–6 theta 447r7–16 beta Walter Gifard So 447r16–22 beta Alvred de Merleberge So 447v1–8 alpha Radulf de Mortemer So 447v10–18 alpha Arnulf de Hesdinc So 448v1–9 alpha 448v10–9r2 iota

Rotbert de Albemarle Dn 419r1–v2 beta 419v3–20r10 alpha 420r10–v14 beta 420v15–1r8 alpha 421r9–14 beta 421r14–v8 delta 421v9–20 . . . alpha Roger de Corcelle So 422r1–4v6 alpha 424v6–5v6 beta 425v7–6r10 lambda 426r11–13 beta 426r13–v2 lambda 426v2–4 beta 426v5–19 alpha 427r1–4 theta 427r5–8v14 alpha 428v15–20 theta 429r1–13 alpha 429r14–v24 theta 430r1–8 lambda 430r8–21 beta 430v2–1v9 — 431v10–19 theta 431v20–2v15 beta 432v15–20 alpha 433r1–9 beta 433r10–v11 alpha 433v12–4r3 beta 434r4–5r19 alpha 435r20–v3 theta 435v4–12 beta

Matheu de Moretanie So 450r1–15 iota 450r16–17 beta 450r18–v3 theta Serlo de Burci So 452r1–16 452r17–v20 453r1–6 453r7–4r11 454r12–20

iota theta alpha iota theta

French knights Dn 456r1–19 beta 456r20–v3 epsilon 456v4–8 alpha 456v8–19 epsilon 457r1–9 gamma 457r10–15 beta 457r16–v11 alpha 457v12–19 gamma 458r1–20 alpha 458v1–14 gamma 458v15–9r16 alpha 459r17–v8 gamma 459v9–18 delta

Edward the sheriff So 437r1–18 alpha Willelm de Ou So 438r1–5 alpha 438r6–12 beta

58

460r1–5 460r6–v4 460v5–14 460v15–1r9 461r10–v12 461v13–17 461v18–2r5 462r6–13 462r14–v6 462v7–8 So 462v11–3r4 463r5–v15 463v15–22 464r1–12 464r13–18 464r18–22 464v1–8 464v8–14 464v15–5r1 465r2–v7 465v7–11 465v12–6r20 466v1–5 466v6–10 466v11–14 466v15–7r3 467r5–14 467r15–19

alpha beta gamma alpha gamma alpha beta gamma epsilon beta theta lambda alpha iota theta eta beta theta iota eta iota alpha beta alpha beta alpha theta alpha

Nicol and others Dn 468r1–7 gamma 468r8–11 alpha 468r12–v3 beta 468v4–19 gamma 469r1–8 beta 469r9–13 epsilon 469r14–v16 gamma 469v17–70r5 alpha 470r6–19 gamma 470v1–1r20 alpha 471v1–2r3 gamma 472r4–v20 beta 473r1–6 gamma 473r7–12 beta 473r13–17 gamma 473r18–19 beta So 473v1–6 lambda King’s sergeants Dn 475r1–8 beta 475r8–15 epsilon 475r16–v17 gamma 475v18–6r5 delta 476r5–10 gamma 476r11–v4 beta 476v5–17 alpha So 477r1–8r13 beta 478r14–v5 theta 478v6–10 iota 478v11–19 beta 479r1–6 alpha 479r7–10 beta 479r11–17 eta 479r18–v16 beta

The surviving portion of the C text – Part II 479v17–80r18 theta 480r19–v8 beta English thegns Dn 481r1–v17 481v18–2r10 482r11–v19 483r1–2 483r3–12 483r13–v16 483v17–4r2 484r3–v20 485r1–7 486v1–7 486v8–7r8 487r9–13 487r14–v11 487v12–8r1 488r2–15 488r16–9r2 489r3–19 489v1–8 489v9–13 489v14–19 490r1–9 So 490r13–20 490v1–5 490v6–11 490v12–16 490v17–1r4 491r5–16 491r17–v10 491v11–14 491v17–22 492r1–14 492r15–16 492r17–21 492v1–6 492v7–15 492v15–3r1 493r1–v3 493v4–6 493v7–12 493v13–14

gamma epsilon gamma epsilon gamma alpha gamma alpha beta alpha gamma alpha gamma beta gamma beta gamma delta epsilon beta alpha alpha — theta beta alpha theta beta theta beta theta alpha beta alpha iota eta theta alpha theta —

500r2–16 500r16–2v10 502v11–3v18 503v18–4r12 504r12–15 504r15–v11 504v12–19 505r1–6 505r6–v11 505v11–13 505v15–6r11 506r12–13 506r13–19 506r19–20 506v1–5 Co 507r1–v2 507v3–8r16 508r16–v3 So 508v6–11 508v11–10v20 511r1–24v12 524v13–15 524v16–5r6

Geld accounts (batch 3)

mu epsilon alpha epsilon — epsilon alpha delta alpha beta epsilon alpha beta — iota zeta mu alpha alpha mu eta — mu

alpha mu alpha

Dn 65r1–9r20 69v1–2 69v2–71r5

alpha beta alpha

Co 72r1–3r15

alpha

So 75r1–82r16 82r17–v20

alpha beta

Geld accounts (batch 4) Wi 1r1–v40 2r1–8 2r8–3r8 3r8–35 3v1–3

ksi — ksi — —

Wi 7r1–20 7v1–31 8r1–11 8r12–30 8r30–41 8v1–41 9r1–6 9r7–v10 9v11–24 Do 11r2–12v10

rho sigma rho sigma tau rho tau sigma mu beta

Wi 13r1–14r20 14r20–8 14r28–16r26

sigma tau sigma

Other booklets (batch 5) Dn 63r1–10 63r13–21

Capp-DnCoSo Dn 495r1–19 495r19–v6 495v7–13 495v14–17 495v18–6r13 496r13–v17 496v18–19 497r1–11 497r11–15 497r15–v12 497v12–15 497v15–21 497v21–3 498r1–13 498r13–9v20 499v21–500r1

Do 17r1–24r6 24r7–9 24r9–21

alpha mu gamma epsilon beta epsilon alpha epsilon — epsilon alpha epsilon mu epsilon mu epsilon

Co 63v1–4 63v7–10 So 63v13–21 64r1 64r15–v14

So 526v1–7r19 4 527v1–8r16 Co 528v1–8

59

alpha alpha + beta + others alpha alpha + beta — — alpha + beta + eta alpha mu —

Chapter 6 The geld accounts associated with the C text

Exeter Cathedral Library 3500 is a collection of booklets (Table 15), written in the spring and summer of 1086, which all relate, more or less closely, to the enterprise known as ‘the survey of the whole of England’. The bulk of the collection (batches 1–2) consists of booklets containing a version of the survey report – the C text, as I call it – more primitive than the versions represented by the surviving D booklets (PRO E 31/1) and the surviving DB booklets (PRO E 31/2). The C booklets have already been discussed, in chapters 4–5; here I aim to deal with two smaller batches of booklets which are found associated with them. These booklets contain accounts, for five counties in southwestern England, of the land-tax known as geld. When Henry Ellis printed the text of this manuscript, one of the points which he stressed in his introduction was the tight connection between the survey and the particular levy of geld to which these accounts relate.1 Ellis was vague about the date of the survey; he failed to see that the geld accounts are not typologically uniform; but his emphasis was in the right place. No one now doubts that the survey was carried out in the year 1086; no one should doubt – what was proved by Galbraith (1950) – that the geld accounts date from that same year.

entries where the scribe wants to be more explicit, we discover what this means: ‘the king has’ becomes ‘the king has in his treasury at Winchester’ (70v–1r, for example). Thus it is perfectly clear that these accounts were drawn up in the Treasury, after the arrival of the money. I hasten to say that this is not an original conclusion. Round stated it first: ‘I am tempted to believe that these geld rolls in the form in which we now have them were compiled at Winchester after the close of Easter 1084, by the body which was the germ of the future Exchequer’ (Round 1888, p. 91). The year is wrong – but apart from that this statement seems sound to me (as long as it is restricted to batch 3). Galbraith agreed with Round. In the (1950) paper which established the date of these booklets, he began with some general remarks about their nature. Having said what they are not (the records of a ‘geld inquest’), he says what they are: ‘They are in fact accounts, some of which at least were drawn up in the treasury at Winchester’ (1950, p. 3).3 That is why, in his (1961) book, he goes so far as to speak of the geld accounts as ‘unique survivals of William I’s administration at work on the humdrum task of ordinary business’ (Galbraith 1961, p. 88). A few pages later, they have ceased to be quite so humdrum: now they are ‘the record of an exceptional administrative activity’ (1961, p. 92), ‘the record of some exceptional, even unprecedented, inquiry’ (1961, p. 96). But both descriptions may be valid, up to a point. These geld accounts are special; but perhaps they are not very different from the sort of records compiled by the Treasury as a matter of routine.

All seven surviving accounts are organized cadastrally: a booklet for each county, a paragraph for each hundred. The larger batch (batch 3) comprises four booklets, one for each of four counties: Dorset, Devon, Cornwall, Somerset. They are the work of three scribes (though largely the work of just one, scribe alpha), exactly the same three scribes who were involved in the writing of the C booklets for all of the same four counties (above, p. 50). The geld accounts are important, in this respect, because they prove that these scribes were employed in the Treasury – permanently employed there, so it seems. To put it briefly (I discuss the evidence in more detail below), there is a statement in every paragraph summing up the proceeds of the tax. In the batch 3 accounts (the batch 4 accounts are different), this statement generally takes the following form: from a certain number of hides ‘the king has’ a certain number of pence.2 From

That Round and Galbraith agree is no proof that they are both right; but in this instance I am satisfied that they are. Indeed, nobody would ever have thought twice about it, were it not for the fact that the Exeter manuscript has been in Exeter for as long as it has any recorded history. Yet there is nothing to prove or make it probable that the manuscript originated in Exeter; and the batch 3 geld accounts are the proof that it did not. It originated in Winchester. How it got to Exeter is a question which needs to be asked, but I do not know that we can have much hope of answering it. The

1

‘Certain it is that the Record itself bears evidence that the tax was raised at the time of the Survey: that it was connected with it: and that, at least in the Western Counties, it was collected by the same Commissioners’ (Ellis 1816, p. xi = 1817, p. 6). The last statement is too strong. The commissioners were not involved in collecting the geld, only in collecting the arrears; and even that is not demonstrably true except for one county.

virgate’, I write 92.875 hides. Rather than ‘twenty-seven pounds, seventeen shillings and three pence’, I write 6687 pence. 3

The implied exception (some but perhaps not all) is for the Wiltshire accounts (batch 4), these being the ones ‘which are palaeographically most distinct from the rest of the book’ (1950, p. 6). Unlike Round, Galbraith had seen the manuscript: the dean and chapter of Exeter loaned it to the Bodleian Library for some period, so that Galbraith could work on it (1950, p. 1, note).

2 The geld is assessed in hides, virgates and acres; payments are counted in pounds, shillings and pence. For simplicity, I state assessments in hides, payments in pence throughout. Rather than ‘ninety-three hides less half a

60

The geld accounts 1

Part of a collection of booklets (C-WiDo) containing a transitional version of the survey text for two counties, Wiltshire and Dorset (fos. 25– 62, 530–1)

2

A collection of similar booklets (C-DnCoSo), nearly complete, covering three counties, Devon, Cornwall and Somerset (fos. 83–494); also one related booklet (Capp-DnCoSo) containing short versions of some entries for the same three counties (fos. 495–525)

3

Four booklets containing geld accounts for Dorset (fos. 17–24), Devon (fos. 65–71), Cornwall (fos. 72–3) and Somerset (fos. 75–82) respectively

4

Three booklets each containing a different version of a geld account for Wiltshire (fos. 1–6 + 529, 7–12, 13–16)

5

Two small miscellaneous booklets (fos. 63–4, 526–8)

been paid on the spot. The entries here tell us this: that from a certain number of hides ‘there have been paid to the king’ a certain number of pence.7 The Wiltshire accounts are the record of payments made locally – not just of payments made in the regular course of events, but also, in a few instances, of payments which were overdue but have now been ‘recovered’. Few though they are, the entries which refer to the recovery of overdue sums are of great interest, and I discuss them in detail below. Unfortunately, when Darlington (1955) edited the Wiltshire accounts, he misunderstood these entries, and in consequence misread the relationship between the three different versions. Establishing the sequence is a technical matter, and I deal with it in an appendix (below, p. 69). But I will say at once that the order imposed by Ellis (1816) is correct. There is no mystery about it: the mystery is how Darlington could get it wrong.8

1

Table 15. Batches of booklets contained in Exeter Cathedral Library 3500.

The batch 3 booklets – the ch accounts, as I propose to refer to them – are very largely the work of a single scribe, the scribe whom I call alpha (Table 16).9 A few lines in chDo are contributed by scribe mu, a few in ch-Dn by scribe beta. Only in ch-So does somebody other than alpha write a large portion of the text: rather more than a page at the end is beta’s work. Even then, the Somerset account is incomplete.10 For this county alone, there also survives an assortment of memoranda relating to the geld account, written by scribe alpha in a separate booklet (526v–7r). It seems likely that memoranda such as these were intended to be thrown away, once the account had been finalized, and their survival, for just this one county, can be seen as another indication that ch-So was never fully finished.

fact remains that these geld accounts are – just as Galbraith said – the only surviving records of the eleventh-century Treasury, the oldest surviving records of any government department.4 The Wiltshire accounts (batch 4) are different. As Galbraith (1950, p. 6) was aware, these booklets – the first one especially, because of its abnormal format – stand apart from the main batch of geld accounts.5 For a start, there are three of them, and already that makes them peculiar. They differ, next, in that only one of the three major scribes (scribe mu) is represented here: he made numerous additions in the margins of the second booklet (very carefully marking the point of insertion for each) and added one whole paragraph at the end.6 These additions aside, the Wiltshire accounts are the work of four minor scribes, two of whom occur only here. In one crucial respect, moreover, the language differs too. These accounts do not tell us how much money has arrived in the Treasury; they tell us how much money has

The presumption is that the survival of these four booklets is fortuitous, and that originally a booklet of the same type existed for every county where the geld had been collected. 7

The first paragraph is exceptional: the formula here is the same as one found in the first half of the Dorset account, ‘and for .. hides the king has .. pence’. From the second paragraph onwards, however, this alternative formula is used. (The only other exception is the paragraph added at the end of the second booklet by scribe mu, which, as might be expected, reverts to Treasury language.)

4

There are, to my knowledge, no published reproductions which show any part of the batch 3 accounts, and that is to be regretted. But a page of the C text written entirely by scribe alpha was reproduced by Bond, Thompson and Warner (1884–94, pl. 70), and the geld accounts look fairly similar to that.

8

In fact, as historians of Wiltshire may know, the error did not originate with Darlington; but he made himself answerable for it.

9

This is the scribe whom Finn (1959) called clerk G, in token of the G which he sometimes uses as a shorthand notation for ‘geld’.

5

In the batch 3 booklets, as in the C booklets, the ruling is for 20 lines (Ker 1977, p. 806). The first Wiltshire booklet has much smaller leaves, ruled for 40 lines. The other two booklets resemble the batch 3 booklets in size and ruling; but the ruling is disregarded (except on 7r), and the number of written lines varies between 29 and 41.

10

Scribe beta began a new paragraph one line up from the bottom of 82v, which is the last page of a quire. He wrote two sentences – and then he stopped, with a quarter of the last line still vacant. At this point, presumably, he was about to make himself a new quire, so that he could continue and complete the text; but that did not happen. Some time later, a memorandum was scribbled in (probably by alpha) at the foot of this page, beginning in the space at the end of the last line and overflowing into the margin. (Below this again, something was apparently written and then erased.) By this time, it seems, events had moved on, and there was no longer any intention of completing the account.

6 Some of the blank pages at the end of this booklet (11r–12v) were used by scribe beta for writing a section (or a draft of a section) of the C text for Dorset. It is possible, as Ker (1977, p. 804) observed, that this quire has been turned inside out: beta’s contribution, now at the back, may originally have been at the front. There are complications here which it may not be safe to ignore, but I propose to take the risk.

61

The survey of the whole of England ery hide which does not belong to the king himself is, in principle, required to pay geld. In a few cases, a manor belonging to a bishop or church may have been exempted altogether, through a gesture of extraordinary generosity on the part of some king; in a case like this it may not even be possible to say how many hides there are, because there are no geldable units to be counted. For fiscal purposes, a manor which enjoys this special status is treated as if it still belongs to the king. Apart from rare cases of that kind, there is only one exception which the Treasury will allow. If the man who holds a manor in domain holds it directly from the king, it is understood that he is entitled to retain a share of the tax proportional to the number of hides which he holds in domain.

Batch 3 ch-Do

17r1–24r6 24r7–9 24r9–21

alpha mu alpha

ch-Dn

65r1–9r20 69v1–2 69v2–71r5

alpha beta alpha

ch-Co

72r1–3r15

alpha

ch-So

75r1–82r16 82r17–v20

alpha beta

Batch 4 g2-Wi

1r1–v40 2r1–8 2r8–3r8 3r8–v3

ksi tau ksi tau

g3-Wi

7r1–20 7v1–31 8r1–11 8r12–30 8r30–41 8v1–41 9r1–6 9r7–v10 9v11–24

rho sigma rho sigma tau rho tau sigma mu

g4-Wi

13r1–14r20 14r20–8 14r28–16r26

sigma tau sigma

Though the language is awkward,14 the meaning is simple enough. In order to claim this partial exemption, you have to satisfy two conditions. First, you have to be – at this moment – personally in possession of the manor. If you have given the manor in feod to one of your men, or if you have leased it to somebody, you lose the entitlement. Second, you have to hold this manor – this particular manor – directly from the king: in other words, you have to be the king’s baron with respect to this manor.15 If you hold this manor from some third party, again you lose the entitlement. If both conditions are met, you are allowed to claim a deduction for as many hides as are reckoned to form the domain – not for the hides which are held by your villains, only for those which are yours in the narrowest sense. You cannot choose your own number, or vary it to suit yourself: some recognized number exists, and that is the number which you are entitled to claim.16 Suppose, for example, that you own a manor rated at three hides, and that one of these hides is reckoned to be domain. Your agent, the reeve of this manor, collects three hides’ worth of geld (216 pence), passes on two-thirds of it (144 pence) to the taxcollectors, and keeps the rest (72 pence) for you.

Table 16. Booklets and scribal stints in batches 3–4. I would guess that all of them were alpha’s work – but there is no way to prove it (or disprove it).11 Apart from the four which still exist, some excerpts from a fifth were copied by scribe mu (more or less verbatim, it seems) into one of the batch 4 booklets (see below); the rest have all been lost.12

If we look for a hundred in which everyone has followed the rules (and we have to look quite hard), the account that we find will record the facts something like this:17

From the surviving ch booklets we can form a fairly clear idea how the system of taxation worked at the time.13 Ev-

shire, earlier than the ch booklets. Unlike them, it is written in English; it does all its sums in hides, never converting to pence. 14

In ch-Do the scribe seems to be trying to clarify matters by making a distinction between dominium and dominicatus, but these words are used interchangeably in the other accounts.

11

I suggested before (above, p. 50) that scribe mu may have been the Treasurer. Now I suggest that scribe alpha may have been the Treasurer’s Clerk, the official in charge of the Receipt.

15 Or, in the language of a later age, you have to be holding of the king in chief, as far as this manor is concerned.

12

With some reluctance, I list the ch booklets in the order which is right for the C text; but that order may be wrong here, and may be misleading too. I can see no satisfactory evidence for working out a seriation, only some doubtful indications that ch-Do may be the earliest of the batch and that ch-So may be the latest. But even if the sequence DoDnCoSo is right, there is nothing to prove that these booklets were written directly one after another. They could be the accidental survivors from a broken sequence .. Do.. Dn.. Co.. So.. (where .. denotes some number of lost booklets). In other words, the fact that these four booklets survive may be due to the fact that these were the last four counties to be dealt with at a later stage by the scribes compiling the C text (below, p. 68); it need not imply that these were the last four counties dealt with at the time by scribe alpha.

16 This is not just a matter of probability, as Galbraith (1950, p. 5) seems to say. The geld accounts are the proof of it. In one of the Wiltshire hundreds, the abbess of Shaftesbury has claimed for 24 hides; but ‘six of these hides were villains’ land on the day when king Eadward was alive and dead’ (2v). Therefore the allowance is reduced to 18 hides, and the abbess’s reeve is charged with withholding the geld from the other six hides (15v). There are similar entries in the same hundred for the abbot of Glastonbury and the abbot of Winchester. 17

This is based on the account for one of the Dorset hundreds (ch-Do-22v– 3r, ed. Williams 1968, p. 142). I have simplified it slightly and made sure that the arithmetic works out. The reader who wants to design a spreadsheet may like to take this for a model.

13

The only other available evidence is a twelfth-century copy (ed. Robertson 1956, pp. 230–7) of some version of a geld account for Northampton-

62

The geld accounts the future may again belong, to one of the king’s barons, such a manor should be made to pay its quota of geld, deducting only the normal allowance for whatever proportion of the land is held in domain. Manors which fall under this heading are regularly mentioned in the geld accounts: the king is listed among the barons claiming some deduction. Usually the phrase ‘the king’s barons’ is replaced by ‘the king and his barons’; usually the king is put at the top of the list. But these are mere niceties. In practice, the king is being treated as one of his own barons, as in the following passage:19

In Haselora hundred there are 64.25 hides. From it the king has 2475 pence for 34.375 hides. The king’s barons own 29.875 hides of it in domain. Of this domain: Hugo’s wife owns 3 hides Walter de Clavilla, 3.625 Roger de Bello monte, 6.5 Durand the carpenter, 0.25 the abbess of Saint Edward, 1 the abbot of Cerne, 3.125 Alvuric the hunter, 0.5 Ascitil de Carisburgo, 2.375 Rotbert son of Radulf, 2 Rotbert son of Gerold, 5 Eddric the reeve, 1 the abbot of Saint-Wandrille, 1 the count of Mortain, 0.5

The king’s barons own 7.375 hides of it in domain. Of this domain: Rainbold the priest owns 2 hides Willelm de Braiosa, 2 Bolo the priest, 2.875 the king, of Herold’s land, 0.5

Thirteen people have claimed deductions here, and the allowances add up to 29.875 hides; the Treasury has received 2475 pence, a sum equivalent to 34.375 hides; the account thus covers a total of 64.25 hides. That is the number of geldable hides in this hundred. If no deductions were allowed, the geld would raise a total of 4626 pence; in fact, the king gets not much more than half of the amount due, as far as this hundred is concerned, and the rest is shared out among his barons.

Usually some note is added, as it is here, to explain why this land belongs to the king for the time being; where the note is missing, we can be sure that the scribe meant to include it but forgot.

In a perfect world, a world in which everything worked to the Treasury’s satisfaction, the geld account for every hundred would look like the one for Haselora hundred quoted in full above. It would contain that information and no more (though the order of the statements might vary). But the real world is far from perfect. In almost every case, the paragraph continues with a string of entries explaining why, in the Treasury’s opinion, the account is defective as it stands.20 From our point of view, the geld accounts derive most of their interest from entries of this kind; but it needs to be realized that they are, quite often, difficult to interpret. They are briefly worded. The scribe who wrote them understood the system, and expected his readers to understand it too. Like a note saying ‘Herold’s land’, these entries assume some nexus of background knowledge which we may or may not possess. We know who Herold was, and why his lands were now in the king’s possession; but we are ignorant of many things which anyone reading these accounts was supposed to be aware of in advance.

In some hundreds, the king himself owns land. This makes a difference, but not very much of a difference, to the form of the geld account. The Treasury makes, or tries to make, a sharp distinction between two groups of manors. One group consists of the manors which belong to king Willelm because they used to belong to king Eadward – the manors which the king has inherited from his predecessor, just as king Eadward had inherited them from his, through becoming king.18 Any manor which falls under this heading is entirely exempt from geld. It has to make other payments, no doubt, but it does not have to pay geld. If the Treasury officials had been asked why this should be so, it is not clear what reply they would have given. But presumably some notion existed that it would be absurd for the king, as lord of the manor, to pay tax to himself, as king. He would merely be transferring money from one hand to the other. In any case, the upshot is that manors of this group do not appear in the geld accounts at all. The second group consists of the manors which happen to have fallen into the king’s hands for one reason or another – through confiscation, through the previous owner’s death, or through any accident which has the same effect. At least in the Treasury’s view of the case, these manors have to be treated differently from those which the king has inherited from his predecessor. They do not belong to the king as lord: they belong to him as overlord (for as long as he chooses to keep them). Because it used to belong, and in

19 Based on part of the account for Bochena hundred in Dorset (23v–4r, ed. Williams 1968, p. 146). 20 In ch-So most paragraphs end with a statement of the sum outstanding. As in the other accounts, the scribe gives a list of entries telling us what has gone wrong; but then, in this county alone, he does the arithmetic for us. For example: ‘For 1 hide which Osbern holds from the bishop of Coutances the king does not have the geld; for 2 hides which Ori holds from the bishop of Coutances the king does not have the geld; for 1.25 hides which Engeler holds from Arnulf de Hesdinc the king does not have the geld; for 5.3125 hides from which the gatherers acknowledge receiving the money the king does not have the geld. From this hundred there are still in arrears 688.5 pence which are owed to the king of his geld’ (ch-So75v). The scribe adds up four items and silently converts the total from hides to pence.

18

In the language of a later age, these are the manors which form the ancient domain of the crown.

63

The survey of the whole of England From the glimpses that we get of it failing to work, the system seems to have been supposed to work like this. In each hundred, four men (or some such number) act as collectors of the tax. They are called by various names, but we shall probably not be misunderstood if we choose to call them the ‘gatherers’.21 In each county, two men (or some such number) act as the Treasury’s representatives. These men are called the ‘bearers’, because they are responsible for carrying the money to Winchester. The geld was collected, not all at once, but in two (presumably equal) instalments, at two set dates. Every penny which is paid, at either term, changes hands three times. The reeve of each manor delivers the money he owes to the local gatherers; 22 the gatherers deliver the money from their hundred to the bearers; and the bearers deliver the money from their county to the Treasury. Some of these transactions, probably all of them, are recorded by means of tallies.23 In addition, the bearers are expected to submit some form of written record, hiring a scribe for the purpose (at the king’s expense).24

Walter de Clavilla – this was concealed from the king – the king has not had the geld’ (ch-Dn-66v).26 What the scribe is usually trying to tell us is that the rules governing the payment of geld have been misapplied. It is often hard to know exactly what he means,27 but most of the time he seems to be harping on one of two themes: that someone has claimed a deduction for domain to which he is not entitled,28 or that land which has lapsed to the king has been treated as if it were part of the king’s inheritance. Then again, there may be hiccups in the collection process. Payments may be late. We are told, for instance, that 144 pence ‘which ought to have been paid at the first term were not paid until the last’ (ch-Do-17r), or that certain sums were paid ‘after the appointed terms’ or ‘after Easter’.29 Payments may be made on time but in the wrong place. In Dorset especially this is a common problem. For example, we are told twice that Hugo de Portu has paid geld in the wrong hundred: the sum is deducted from one account (24r) and added onto another (20r). It does not appear that any action is going to be taken against Hugo and others who have made the same mistake:30 the Treasury just wants to get the facts straight. But some of the gatherers are certainly in trouble, in part because they have accepted payments which they ought not to have accepted.

That all sounds fairly straightforward. The system has been tried and tested over many years. How can anything go wrong? The answer is, we discover, that things can and do go wrong, at almost every point. We are told, over and over again, that from a certain number of hides ‘the king has not had the geld’.25 This is not the gatherers’ fault. They are there to receive the money, when it is handed to them: they are not expected actively to seek it out. The onus is on the man who possesses the land. If he fails to pay on time, he is guilty of withholding the geld; if he is found to own some undeclared assets, he is guilty of something worse. He is guilty of concealment: he has failed to inform the king of something which it was his duty to report. This seems to have happened very rarely. In all four accounts there is only one entry which makes an explicit accusation of concealment: ‘For half a hide which Godefrid holds from

In three of the Somerset hundreds, small sums received by the gatherers have gone astray, and the gatherers have not been able to justify the deficit.31 In many Devon hundreds, 26

In C and DB the word ‘concealed’ occurs more frequently, but still in only a very small proportion of cases. (The word ‘discovered’ carries the same meaning: we do not know that a hide has been concealed until we have discovered it.)

27 To make sense of these entries, one has to try to match them up with corresponding entries in C (or in DB, where C is missing). Because there are hardly any place-names in ch and no hundred headings in C, that proves to be very difficult. How much progress can be made varies from county to county. The reader who wants to pursue this line of enquiry should probably start with Dorset, where the evidence was analyzed very thoroughly by Williams (1968).

21

In Devon and Somerset they are usually called the fegadri, a feebly latinized form of their vernacular name, which also occurs in translation as congregatores huius pecuniae (ch-Do-17v, 18r, 22v). The English word, not otherwise attested, would be *feohgadrere.

28

An example which might be quoted here is an entry in the Somerset account: ‘For 5.75 hides which Sanson holds from the bishop of Bayeux the king has not had the geld’ (ch-So-80v). The men of the bishop of Bayeux are in a peculiar position, because the bishop himself has been held in prison for the last few years. Even so, ‘the land of the bishop of Bayeux’ is still theoretically the bishop’s, not the king’s. His barons have not become the king’s barons. What has happened here is this. The manor in question is assessed at 8 hides (C-So-467r, DB-So-87vb). Sanson has claimed a deduction for that part which is counted as domain: the Treasury scribe is saying that Sanson’s claim is invalid, and that he should be made to hand over the money which he has kept. Sanson’s case has attracted some discussion, most of which neglected a crucial fact knowable only to someone consulting C-So in the original (Thorn and Thorn 1980, p. 314, Chaplais 1987, p. 69).

22

It is assumed, of course, that the reeve will recoup himself by collecting the individual contributions – three pence here, six pence there – which go to make up the total; but that is his business, and does not concern the Treasury. 23 The word for tally is dica. Twice the scribe writes in dicis, running the words together; Williams (1968, pp. 126, 127) reads indicis, but that does not make sense. The same word recurs in a text describing the organization of the king’s household in the 1130s: the marshal, we are told, debet habere dicas de donis et liberationibus que fiunt de Thesauro regis (ed. Johnson 1950, p. 134). 24

The Northamptonshire account (above, note 13) should probably be read as the report drawn up by the bearers. On this view, the most striking formula – ne com nan peni of, ‘there came not a penny from’ – will be the approved way of notifying the Treasury that a problem has arisen. As far as the bearers are aware, there is no reason why the geld should not be paid; but the man who owns the land has refused to hand over the money.

29 The total given at the end of ch-Do (24r), 99705.5 pence, does not include these late payments; nor does it include 180 pence which Rotbert de Oilleio ‘withheld till after Easter’, and which, as is noted above the line, ‘the king still does not have’ (19v).

25

In the Dorset account the scribe tries to distinguish between land from which the king has ‘not’ or ‘never’ had geld. By ‘not’ he means ‘not this year’: he says so in the first two paragraphs (17r–v) but does not bother to repeat it after that. By ‘never’ he means that ‘king Willelm has never’ had the geld (17r), i.e. that payment has lapsed since the time of king Eadward.

30

In the Dorset account as a whole there are fifteen entries of this kind. They do not all pair off as neatly as these two.

31

In Abediccha hundred a sum of about 7 pence is missing, and ‘the gatherers could not give us an explanation’ (de quibus fegadri non poterant

64

The geld accounts the gatherers have quite deliberately kept a small share of the proceeds (usually one hide’s worth) for themselves: we are told repeatedly that the king does not have this money because ‘the gatherers claim that by custom they ought to have it’. The language here seems neutral – but sometimes there is a clear hint of disapproval. The gatherers have ‘withheld’ the money; the custom is only a custom ‘according to what they say’ (69r). In two Dorset hundreds, the gatherers are in surplus, not just because they have ‘received money which they ought not to have received’ (i.e. which ought to have been paid in some other hundred), but also because the number of pence recorded on their tallies is ‘in excess of the number of hides’ (17v, 18r). We should not take this to mean that the gatherers are being accused of extortion. The point seems rather to be that if these gatherers, on this occasion, have collected more money than expected, something must have gone wrong on a previous occasion, and some retrospective inquiry should be made.

geld’ started out with 123371 pence, but some of the money has disappeared along the way, and only 122160 pence have actually arrived.34 There is a discrepancy of 1211 pence. In part, the bearers can explain this to the Treasury’s satisfaction. They claim a round sum of 480 pence for their subsistence; they claim 116 pence for incidental expenses – hiring pack-horses and a scribe, buying packing-cases and wax. The Treasury allows these deductions. But that still leaves a deficit of 615 pence. Of this money ‘the king has not had a penny’, and the bearers ‘have not been able to render an account’.35 Before they can depart from Winchester, the bearers have to pledge ‘that they will pay this money to the king’s commissioners’ (526v–7r). It is clear that this memorandum was not drawn up till after the money had been delivered to the Treasury. The man who wrote it, scribe alpha, must have been a Treasury scribe.

The bearers are not just responsible for seeing that the money is safely transported to Winchester. They are expected to make sure that the gatherers have done their job properly; if they come across any problems, they are expected to report them to the Treasury.32 This means that they have to check the gatherers’ counter-tallies, verifying that the total is the same as the number of pence which the gatherers have handed over. But that is not all. The bearers already know, or think that they know, the number of geldable hides in any hundred. By subtracting from this the number of hides that do not need to pay, and the number that have failed to pay, they can calculate the number of hides which ought to have paid their quota, and therefore the number of pence which ought to have passed through the gatherers’ hands. If the sums do not agree, they insist on knowing why. The bearers, it seems certain, arrive with a written list – a list which gives the name of each hundred and the number of geldable hides believed to be there. That list, I take it, was supplied to them by the Treasury, and was based on the accounts for the geld preceding this one.33

Yet how can a Treasury scribe, sitting at his desk in Winchester, possibly know all this? By interrogating the bearers, he can discover lapses in the collection process. But how can he know that half a hide in Devon has been concealed, or that a man in Somerset has falsely claimed a deduction with respect to domain, or that some land in Dorset has been mistakenly treated as part of the king’s inheritance? How can he have knowledge of things which apparently could only be discovered on the spot? Up to a point, the answer is obvious. As Galbraith (1950) realized, the geld accounts incorporate new information resulting from the survey. Working through some preexisting accounts (of the kind which would have been drawn up regularly whenever geld was paid), scribe alpha checks them against some version of the survey report, in search of any discrepancies; and the accounts that survive, the ch booklets, are the record of what he finds. In the majority of hundreds, he discovers that some money is still due. That money will need to be collected; those people who are guilty of misconduct will have to answer for it. Scribe alpha compiled these accounts so that they could be made available (presumably in the form of fair copies) to those agents of the king who would be chasing up arrears and punishing defaulters.36 We know who those agents were: when the bearers of the geld from Somerset were found to be in deficit, they were not allowed to go home until they had pledged that they would pay the money to the ‘king’s com-

One of the memoranda relating to the Somerset account gives us our only glimpse of the bearers arriving in Winchester and settling up with the Treasury. The ‘bearers of the reddere nobis rationem, ch-So-81v). The word ‘us’ might be taken to imply that the gatherers have been interrogated by the Treasury; I think it means that the Treasury scribe is quoting from the bearers’ account. 32

In the Devon account, the last paragraph ends with a note about the geld from 3.3125 hides which has somehow gone astray. ‘The gatherers say that they received the money and handed it over to Willelm the usher and Radulf de Pomario, who were supposed to carry the geld to the king’s treasury in Winchester’ (ch-Dn-71r). It is not their fault, the gatherers are saying, that the money has failed to arrive. This passage thus gives us the names of the bearers for Devon; it seems also to imply the existence of some line of communication which does not pass through them. In this one instance, the gatherers have been given a chance to speak for themselves. Either they have been ordered up to Winchester; or somebody has been asked to make enquiries on the spot.

34 Presumably these numbers would come from the bearers’ tallies. At this stage, I suppose, the bearers would have a bundle of counter-tallies recording the payments made by the gatherers to them, and also a single tally recording the payment made by them to the Treasury. 35

The language here seems to echo the Northamptonshire geld account (above, note 13) as well as anticipating the twelfth-century Exchequer rolls. 36

My guess would be that in the case of Somerset the fair copy (ch2) had to be begun before the draft (ch1) could be completed (above, note 10). For lack of time, ch1 was cut short, and the remaining paragraphs were entered directly into ch2. After that, the flow was reversed, and scribe beta copied a stretch of text from ch2 into ch1.

33

Probably the geld, at a variable rate, was collected every year. The phrase ‘this year’ (above, note 25) is suggestive; but in the absence of any contrasting phrase like ‘last year’ it cannot be thought conclusive.

65

The survey of the whole of England missioners’ – that is, the commissioners who would shortly be carrying out the survey of Somerset.

relatively easy for him to identify the corresponding entries (much easier than it is for us, working with ch on one side, C or DB on the other); but some guesswork, it seems, must also be involved.

But we seem to have stumbled into a contradiction. At first we were saying that the ch accounts were not drawn up till after the results of the survey became available; now we are saying that the survey was not carried out till after the ch accounts had been compiled. There is no contradiction. From a contemporary description of the workings of the survey, written not by some humble cloister-bound chronicler but by the bishop of Hereford (Stevenson 1907), we know that the survey was a two-stage operation. In stage one a team of local men (this is implied) put together a preliminary version of the survey report (of the B text, as I call it) for their county. In stage two a team of outsiders (this is stated explicitly) made sure that the local men had done their job honestly and approved the final version of the survey report. The B text would thus have existed in two forms: a provisional version B1 and a final version B2.37 Here we have the solution: ch is later than B1 but earlier than B2. Information can flow from B into ch; from ch it can flow back into B.

From the geld accounts, we can fill in some parts of the story which bishop Rotbert did not think worth mentioning. The draft version of the survey report was sent up to the Treasury when finished, not just for safekeeping, but so that it could be worked over in detail (and in a great hurry, one would guess) by the Treasury officials. When the second team of commissioners were ready to start their work, they were given the B1 text which they had to verify; and they were also given a copy of the corresponding ch text, so that they would know what action was expected from them by the Treasury. And thus we can understand why bishop Rotbert ends his description of the survey by deploring the outbreaks of violence which ensued, ‘arising from the collection of the king’s money’. After Easter, usually, the geld was over and done with; but that was not the case this year. Across the country, in hundred after hundred, people found out that more money was demanded from them. Not everyone took the news calmly.

More precisely, the process that I envisage would work like this. The scribe who compiles the ch text for each county is working with two source texts: the provisional version of the survey report, B1, and a version of the geld account, g1. In any normal year, g1 would be regarded as the final version; but this is not a normal year. An opportunity exists such as never existed before to work out exactly how much money is being lost to the king, and scribe alpha makes the most of it. For each hundred in turn, he can, if he wishes, compile from B1 an ideal geld account. He can find the total number of geldable hides (if it is not given) by adding up all the assessments recorded here. He can make a list of the manors which are held in domain directly from the king, noting for each the name of the baron and the number of hides in domain. By adding up those numbers, he can find the number of hides for which deductions are allowed. Then, by subtraction, he can find the number of hides which have to pay, and so, by multiplication, the number of pence which this hundred owes to the king. Though I do not say that he actually bothers to do this, in principle that is what is happening: the scribe is comparing g1 with an imaginary geld account derived from B1. Thus, when he finds a deduction allowed in g1 which is proved by B1 to be unjustified, he omits from ch the entry that he found in g1 and replaces it with an entry of his own at the end of the paragraph, saying that ‘the king has not had the geld’ from the land in question.38 Because B1 is organized cadastrally, it is

2 With the Wiltshire accounts I propose to deal as briefly as possible, concentrating on the features in which they differ from the ch accounts. Reproductions have been published in some number (Table 17).39 The sequence was worked out correctly by Ellis (1816); anyone doubtful of that should consult the appendix (below, p. 69) before reading further. For reasons which will appear, I refer to these booklets as g2-Wi, g3-Wi and g4-Wi. Not counting scribe mu (who made some important additions in g3), four scribes are represented here (Table 16). Two of them, rho and sigma,40 occur nowhere else in this manuscript; the other two, tau and ksi, wrote portions of the C text for Dorset.41 It seems fair to say, without pressing the distinction too far, that booklets g2 and g4 are fair copies, each largely the work of one of Bayeux the king has not had the geld’ (ch-So-80v). In due course, the second team of commissioners investigate the matter. They confirm that the manor is held by Sanson from the bishop of Bayeux, not directly from the king; they also report that there are 5 hides in domain here (and 0.75 hides elsewhere, in a small manor added to the large one). These are the facts reported in B2-So (and ultimately in DB-So). In the circumstances, the second fact is immaterial – but the commissioners record it anyway, because the question has come up. 39

((I have added the sample reproduced by Thorn and Thorn (2001).))

40

Sigma is the scribe whose hand has been recognized by Ker (1976, 1977) and Webber (1989, 1992) in several manuscripts from Salisbury.

37

Whether B2 was a new manuscript or the old manuscript with an added layer of annotation is a question which had probably better not be asked, because there seems to be no hope of answering it. ((But I disregard my advice and come back to this question later (below, pp. 122–3).))

41 To scribe tau I assign two stints here and two stints in C-Do (37v3– 8r7, 51r17–v6) which I previously left unattributed. Having looked at the evidence again, I feel fairly sure (though still not quite as confident as I should like) that these stints are tau’s work. In that I come round to agreeing with Webber (1989); but I still hold to the view that the scribe whom I call theta (occurring only in C-So) is a different individual from tau. As for scribe ksi, there are several stints in C which I attribute to him (38r7–17, 42r1–8, 47r1–8r3, 48v1–9r12, 50r16–v3, 50v10–1r17, 58r12– 18); they include the sole surviving entry for Wiltshire (47r1–11).

38

Suppose, for example, that he finds in g1-So, among the deductions claimed in Meleborna hundred, ‘Sanson 5.75 hides’. Consulting B1-So, he discovers that the only manor in this hundred held by Sanson is not held directly from the king; so he disallows the deduction, and replaces g1’s entry with this: ‘From 5.75 hides which Sanson holds from the bishop

66

The geld accounts scribe, while booklet g3 is a working draft, a collaborative effort by a team of scribes. In g3 we see g2’s text being put into a new shape, ready to be copied out again as g4. Again the presumption is (I do not flinch from this) that a similar set of booklets existed for every county where the geld had been collected. That this particular set is the only one to survive is a fact for which we shall have to seek some explanation.

72 pence’ (3r). In the Wiltshire accounts we catch occasional glimpses like this of two teams of investigators, both in pursuit of geld which for one reason or another has failed to reach the Treasury. Several times we hear of ‘Walter and his colleagues’. Twice we hear of ‘the bishop and his colleagues’ (2r), ‘bishop Willelm and his colleagues’ (1v).43 In one hundred, both teams have been active: 44 From those who collected the geld, Walter and his colleagues recovered 63 pence, not counting 105 pence which the bishop and his colleagues have found (2r).

Booklet g2-Wi. Before it became the starting-point for the compilation process which resulted in g4, g2 was the culmination of an earlier compilation process. Before we look forward, we have to try looking back. In the small size of the leaves and the closely ruled lines, g2 is very obviously different, not just from g3 and g4, but from all the rest of the collection. At first sight, it is a quire of 6. But Ker (1977, p. 804) pointed out that a blank leaf which ought to follow these became separated from them when the manuscript was rebound in 1816. So it seems rather to be a quire of 8, the first leaf of which has been cut away and lost. (I suggest below that this missing leaf may not have been blank.)

From Ellis onwards, historians have identified these investigators as the commissioners carrying out the survey of Wiltshire; but it was only after the publication of bishop Rotbert’s description of the survey that the significance of the entries could be fully understood. Bishop Rotbert speaks of two successive teams of commissioners: here, in one county, we see them both in action. The point to be stressed, however, is the fact that we find no entries resembling these in any of the ch accounts. In g2, conversely, we find no entries of the kind which in ch appear to have originated with scribe alpha – those entries in which he points out cases where ‘the king has not had the geld’ which apparently he ought to have had. If we look more closely, however, we discover that questions of the kind which were raised by scribe alpha are being answered here. In Selchelai hundred, for example, there is a long list of the deductions claimed by the king’s barons. Three of these entries are flagged by codes in the margin:

The text, by and large, is similar to what we would have expected to find in ch-Wi, if ch-Wi had survived. It is organized in the same way, hundred by hundred; the information that is given here is mostly the same that is given in the ch booklets. For a hundred where no complications have arisen, the account will look something like this: In Thornegrava hundred there are 113 hides. The barons own 46.875 hides of it in domain. Willelm de Ou, 5 hides the abbot of Glastonbury, 32 Hernolf, 4.875 Hunfrid de Insula, 5 For 66.125 hides there have been paid to the king 4761 pence (2v).

Rotbert 2 hides which he holds from Willelm de Braiosa, . . . Gislebert Gibart 2.625 hides which he holds from the abbot of Glastonbury, . . . Edward the sheriff 0.5 hides which :::::: his predecessor holds at rent (2r).

One difference between this and the ch accounts has already been noted: the sum which has been ‘paid to the king’ is the sum paid to the gatherers on the spot, not the sum eventually paid into the Treasury. That becomes clear from an entry like this one:

These are not idle remarks. They are the answers to questions posed in ch-Wi. The meaning is: ‘Yes, you are right, these deductions should not have been allowed – in the first two cases because the man who has claimed the deduction does not hold directly from the king, in the third case be-

For 83 hides there have been paid to the king 5976 pence. Of this sum, those who collected the geld have withheld 324 pence till now (2v–3r).

43 There were two bishops named Willelm at the time, the bishops of Durham and Elmham. In the light of other evidence, it is tolerably certain that the man in question here is the bishop of Durham. But in either case bishop Willelm is an outsider, holding no land in Wiltshire.

In this hundred the gatherers have received payment for 83 hides, but the sum that they have forwarded is only 5652 pence. (This is the sum which ch-Wi would tell us ‘the king has’ in his treasury.) A special code in the margin, equivalent to ‘N.B.’, draws attention to this entry.42

44

In this paragraph, atypically, the sum said to have been paid to the king ‘at the appointed terms’, 6519 pence, is the sum which the Treasury acknowledges receiving on time; the two sums recovered later bring the total to 6687 pence, exactly right for 92.875 hides. The corresponding entry in g3 is: ‘and the collectors of the geld withheld 168 pence’ (8r). In g4 the scribe miscopies the number of hides as 93.125 (he writes dim’ uirg’ instead of dim’ uirg’ minus) and is then confounded by his own error. Checking the arithmetic, he decides that the account is short by 18 pence more than g3 says – and promptly assumes that the gatherers are to blame. So the entry turns into this: ‘The four collectors of the geld withheld 186 pence’ (14v). (Later, the missing word minus was inserted above the line, but the misbegotten calculation was not put right. One hopes that the gatherers were not forced to pay for scribe sigma’s clerical error.) Just from comparing the different versions of this entry, it is clear that Ellis’s ordering must be right.

The same entry goes on to say that of the sum withheld by the gatherers ‘Walter and his colleagues have recovered 42

In fact the code is d.m., to be construed as dignum memoria, ‘something worth remembering’ (Ker 1976, p. 26). It occurs frequently in g2 (32 instances), occasionally in g3 (3 instances).

67

The survey of the whole of England Page 1v 8r 8v1–9 9r1–6 9r5–18 9r7–12 14r 15r

Reproduction

Scribes represented

Darlington 1955, opp. p. 180 Darlington 1955, opp. p. 181

ksi rho (1–11), sigma (12–30), tau (30–41), mu (margin) rho tau tau (5–6), sigma (7–18) sigma sigma (1–20), tau (20–8), sigma (28–9) sigma

Thorn and Thorn 2001, ill. 27 Webber 1989, pl. 3 Ker 1976, pl. III (a) Webber 1989, pl. 1 Darlington 1955, opp. p. 216 Ha11am 1986, pl. 7

Table 17. Published reproductions of sample scripts from the batch 4 booklets. cause he does not hold in domain.’45 Accordingly, if we follow these entries forward into g4, we find that they are removed from the list of deductions, put at the end of the paragraph, and transformed into outright accusations:

Wiltshire. Finally the draft is corrected by scribe tau, who makes many small alterations and additions (perhaps also derived from ch-Wi, though I do not insist on that). In the fair copy, g4-Wi, the scribe does some editing as he goes along (especially when he reaches the paragraph added by mu), but in general does not diverge very far from the draft. The last paragraph, apparently left till last because it was problematic, shows a few substantive corrections. They are the work of scribe tau, who, in all three of these booklets, seems to have been responsible for finishing off the text.

Rodbert de Braiosa has withheld the geld of 2 hides, Gislebert Gibard has withheld the geld of 2.625 hides, a certain thegn of Edward the sheriff’s has withheld the geld of 0.5 hides (15r).

Thus booklet g2-Wi belongs to a later stage in the proceedings than the ch booklets. Scribe alpha, when he wrote those booklets, was looking ahead to the time when the second team of commissioners would do its work: that work has now been done. What we have here, I suppose, is a copy taken from the original geld account, g1-Wi, incorporating all the corrections, additions and annotations which had accrued to it by this time. In a manner of speaking (but not a manner of speaking which I would recommend), this booklet seems to be the sole surviving specimen of the ‘original returns’ – the documentation submitted to the Treasury at the conclusion of the fieldwork phase of the survey. On that view, the scribe who wrote most of it, ksi, was one of the scribes who accompanied the second team of commissioners – bishop Willelm and his colleagues – during their visit to Wiltshire.

By the time that scribe mu intervenes, g3 is certainly in the Treasury; probably it has been there since its inception. Booklet g4, we may infer, was written in the Treasury too. But why, for what purpose, was it written? If we try following the evolution of individual entries, the overall trend is clear. The application of the rules becomes increasingly strict, and the language becomes increasingly explicit. Any geld not paid on time is said to have been ‘withheld’, and some particular person is accused of having withheld it. This booklet thus becomes something more than a geld account: it is also a bill of indictment. In drawing up this text, the Treasury anticipates a further visitation of the county, by a team of emissaries who will be empowered not just to investigate but to do justice. That was the plan; but the fact that g4 still survives, and still keeps company with g2 and g3, is a fairly strong hint that the plan was not followed through with, at least as far as Wiltshire is concerned.

Booklets g3-Wi and g4-Wi. Starting with g2, three scribes set about constructing a new version of the text. Booklet g3-Wi is their working draft. Though simplified and rearranged to some extent, the text is mostly derived from g2. But it includes a number of entries, all relating to the king’s land, which are not to be found in the source text. It seems a likely guess (but only a guess) that these entries were derived from a separate list of the king’s land on g2’s missing first leaf. Next, the draft is checked by scribe mu, who inserts in the margins a series of entries, again all relating to the king’s land, which he wants to have included in the text. He also adds one whole paragraph at the end. The wording of this final paragraph agrees in every respect with that of the ch booklets, and we can feel fairly sure that mu’s additions were all derived from the lost ch booklet for

3 Why some of the ch booklets survive is an easy question, provided that we are content with an easy answer. They survive for the same reason (the same concatenation of reasons) that some of the C booklets survive. Like the C text, the ch booklets were part of the contribution made by the Treasury scribes to the compilation process. When the C booklets were sorted into stacks (above, p. 53), the ch booklets were sorted too: with respect to the stacks which survive in fossilized form, ch-So is at the top of stack 2, ch-Do at the bottom of stack 3, ch-Co at the top and ch-Dn just below the top of stack 4. This is all more or less as it should be. If the C text does not survive (as is true for all counties up to and including Wiltshire), the ch booklet does not

45

In this last instance the man who used to own the land remains in possession of it, but now he is Edward’s tenant. (His name has been erased in g2; in DB he is called Azor (DB-Wi-69vb).) It is he, not Edward, who in g4 is accused of withholding the geld.

68

The geld accounts survive either. If the C text does survive (as is true for all counties from Dorset onwards, though for Dorset only in a qualified sense), the ch booklet survives too.

and that C was copied from B. To a large extent in B, to a lesser extent in A and C, it is necessary to distinguish (as far as this can be done) between the initial and the final state of the text. In such a case I write A + > B, meaning that B was copied from A, but not until after A had been altered to A +.

Why the Wiltshire booklets survive is a harder question, not to be answered all at once. We should probably start by asking why it is that not just one but three different versions survive. To that the best answer will be, I suggest, that some checking process was intended for which all three would be needed.46 This explanation will only work if we are willing to accept that the intention was not fulfilled: once the checking had been done, it would no longer have been necessary for all three versions to be kept together (or for the first two versions to be kept at all, as far as one can tell). But I see no difficulty here. The survival of all three versions can be seen as another sign – like the fossilized stacks of C booklets – that the compilation process was brought to an unexpected halt before it had quite been completed.

In Table 18, I print four versions of the text for one small hundred.47 If readers wish to experiment with different sequences, of course they are welcome to do so. (The only constraint is that B must be placed somewhere before B +.) But it seems fairly obvious to me that the sequence is right as it stands. There are two corrections in A +: a few words have been erased in one place, two words inserted in another. Both corrections are carried forward into B. The differences appearing in B are the omission of the number of hides in domain and the inclusion of an entry for 4 hides of the king’s land.48 Probably the number of hides in the hundred was increased by 4 accordingly, but the text is indistinct at this point. In B + three corrections are made: the number of hides in domain is recalculated and inserted (by scribe tau),49 an entry is inserted (by scribe mu) for 5 hides of Herold’s land which has not paid its quota, and the total number of hides in the hundred is adjusted upwards (probably by scribe tau).50 These corrections are carried forward into C.51 The only important difference appearing in C affects the entry for 2 hides of land belonging to a man named Gunter. Already in A we are told that Gunter has been allowed a deduction for these 2 hides,52 but that he is not entitled to it: Turold gave these hides to his niece (Gunter’s wife, by implication), so Gunter holds from Turold, not from the king. In C, finally, this meaning is made explicit: ‘From two hides Gunter has withheld the geld.’

This theory will explain why the survival of one version might entail the survival of all three; but it does not explain why any one of them does actually survive. To that the answer is obvious, up to a point. There is some complementarity at work here. These Wiltshire accounts survive because the Treasury’s own account does not: they survive because they were substituted for it. So the question to ask is why ch-Wi does not survive. Again the answer is obvious, up to a point: for the same reason that the C text for Wiltshire does not survive. Shortly before the work was interrupted, all the C booklets relating to Wiltshire were separated out and removed, to be used for some purpose which was clear enough at the time, even if we find it hard to see the sense of it now. At the same time, with the same purpose in view, the ch booklet for Wiltshire was removed; and this parallel batch of accounts was put in its place, for the time being. They became, temporarily, part of stack 3, i.e. the stack of C-Wido booklets not currently in use. (The Treasury, it seems, would need to have some sort of record available, in case any queries arose.) There was (I assume) nothing new about this: the same thing would have happened repeatedly before, for every county that preceded Wiltshire in the sequence. But then, while Wiltshire was being dealt with, the work suddenly stopped, and the temporary arrangements existing at that moment achieved an unintended permanence; in fact they still exist.

In a few places, two or more entries have been added together and turned into a single entry. As evidence for A > B, we can find two entries in A: Ricardus de terra Alberici ii hid’ & dim’, De terra eiusdem A. iii hid’ & dim’, . . . (2r), corresponding with one entry in B: 47 The reader who would like to see another example will find three versions of the account for Calne hundred printed in parallel by Birch (1887, p. 58). He understood the relationships between them correctly (p. 57). 48 For this addition we have to find some explanation (or else we shall have to reverse the sequence, assuming that the entry was omitted from A rather than added in B). It is one of a series of similar additions in B, all derived, I suggest, from a lost section of the A text (a section which occupied the missing leaf at the front of this booklet), entries from which were woven into the B text.

Appendix Sequencing the Wiltshire accounts

49

Exclusively within this appendix, I will refer to the three Wiltshire accounts as A, B and C, the notation used by Darlington (1955). The sequence that I aim to establish is A > B > C, where the > signs denote that B was copied from A

He includes the king’s 4 hides, but does not count Gunter’s 2 hides.

50

The new total, 60 hides, includes the 4 hides which first appear in B and the 5 hides which first appear in B +. 51 Instead of a shorthand note, ‘Barons 22.5 hides’, C has a properly worded sentence; but a copyist with any sense could construct this sentence for himself, modelling it on the parallel statement in one of the other paragraphs.

46

In normal circumstances, after making a copy, one checks it against the exemplar. But if one plans to make a copy of the copy, it is more sensible to delay the checking until one can check this third manuscript against the first one.

52 But in fact 12 pence has been paid: hence the extra third of a half of a hide and the extra shilling appearing in the last sentence.

69

The survey of the whole of England that textual evidence is ever absolutely conclusive; very often some awkward facts turn up which have to be explained away. But here we have good evidence in favour of the sequence A > B > C, and we are not going to abandon that conclusion unless it clashes with some countervailing evidence of even greater cogency. That is not the case. There is other evidence – changes in the order of the paragraphs, for instance – which tends towards the same conclusion as this. There is, as far as I can see, no evidence which points unequivocally in the opposite direction.

A +: In hund’ de Wrde s’t l & i hid’. De his h’nt barones in d’nio xx hid’ & dim’ :::: :::: :::: :::: Inde h’t Turoldus vi hid’ (& dim’), & Gunterus duas hid’ quas turoldus dedit nepti su˛e, de his ii hid’ redditi s’t xii d’, Rotbertus filius Rolui vii hid’, Grinboldus v hid’, & pro xxx hid’ & dim’ & tercia parte dim’ hid’ reddit˛e s’t regi ix li’ & iiii sol’. B: In hund’ de Worda s’t .... hid’. De his h’t rex iiii hid’ in d’nio, Turoldus vi hid’ 7 dim’, Gunterus ii hid’ quas Turoldus dedit nepti su˛e, de his ii hid’ redditi s’t regi xii den’, Rodbertus filius Roulf vii hid’, Grimboldus v hid’, 7 pro xxx hid’ 7 dim’ 7 tertia parte dimidi˛e hid’ reddit˛e s’t regi ix lib’ 7 iiii sol’.

How, then, did Darlington (1955) arrive at a different result? He was vague about B’s position in the scheme of things, but sure that C was earlier than A. That was implied, he thought, by a pair of entries like these:

B +: In hund’ de Worda s’t (lx) hid’. (Barones xxii h’ 7 dim’.) De his h’t rex iiii hid’ in d’nio, Turoldus vi hid’ 7 dim’, Gunterus ii hid’ quas Turoldus dedit nepti su˛e, de his ii hid’ redditi s’t regi xii den’, Rodbertus filius Roulf vii hid’, Grimboldus v hid’, 7 pro xxx hid’ 7 dim’ 7 tertia parte dimidi˛e hid’ reddit˛e s’t regi ix lib’ 7 iiii sol’, (& pro v hid’ quas uill’ tenent de terra heroldi n’ h’t rex ghildum.)

C: The collectors withheld one penny (13r). A: Those who collected the geld have now paid one penny which was left over (1r).

or these:

C: In hund’ de Worda s’t lx hid’. De his h’nt Baron’ xxii hid’ 7 dim’ in d’nio. Inde h’t Rex iiii hid’ in d’nio, Turoldus vi hid’ 7 dim’, Rodbertus f’ Roulfi vii hid’, Grimbaldus v hid’, 7 pro xxx hid’ 7 dim’ 7 tertia parte dim’ hide reddite s’t regi ix lib’ 7 iiii sol’. De v hid’ quas ten’ uillani de terra haroldi n’ h’t rex geldum. De ii hid’ retinuit Gunterus geldum preter xii d’.

C: The collectors withheld 120 pence (15r). A: Those who collected the geld withheld 120 pence, and now this sum has been paid (2v).

In C we are told that a certain sum has been withheld; in A we are told that this sum has been paid; therefore A is describing a later state of affairs than C. That is Darlington’s argument – the beginning and the end of it, as far as I can see.

Table 18. Four versions of the text for one Wiltshire hundred (A from 3r, B and B + from 9r, C from 15v). Ricardus de terra Alberici vi hid’, . . . (8r).

But that is not the only way in which this evidence can be read. As I read it, the fact that the money has been paid does not alter the fact that the money was withheld in the first place – not paid when it should have been paid. Now that it has the money, the Treasury can cease to concern itself with the details; but somebody has committed an offence, and that cannot be forgotten.

Similarly, as evidence for B > C, we can find six entries in B: E. uicecomes vi hid’ 7 i uirga, Osbernus gifard iiii hid’, . . . Eduuardus uic’ viii hid’ dim’ uirga minus, . . . Eduardus uic’ iii hid’, . . . Osbernus gifard i hid’, . . . Eduuardus uic’ d’ hid’, . . . (8r), corresponding with two entries in C: E. uicecomes xvii hid’ 7 dim’ 7 dim’ uirg’, Osbernus gifard v hid’, . . . (14r). This is very strong evidence. It is easy to see, in the second case, how C could derive from B, by way of a little arithmetic (which we can check) on the part of the copyist; it is impossible to see how B could derive from C – impossible, that is, unless we are willing to suppose that B also had access to some lost source, more detailed than C. There are times, it is true, when one finds oneself obliged to resort to some ad hoc explanation of this kind.53 I am not sure 53 I just did so myself (above, note 48), to account for the presence of an entry in B which is not to be found in A.

70

Chapter 7 Some alien interventions in the C text

is only one more feature of the manuscript which I have any thought of discussing. I hope to deal with scribe mu’s statistical summaries (527v–8r) at some later date.1

The principal component of Exeter Cathedral Library 3500, adding up to 452 out of a total of 531 leaves, is a portion (perhaps about one eighth) of that version of the ‘Survey of the whole of England’ which I refer to as C. If C survived complete, subsequent versions of the survey text would be of no more than incidental interest. It is only because most of C has been lost that these subsequent versions – D and DB (PRO E 31/1–2) – have value. Yet C, for various reasons (partly because it ran away to Exeter), has not received the attention that it deserves. I discussed some aspects of the evidence in chapters 4–5; but there I deliberately restricted myself to the primary text – the text as it was written in the first instance, before anyone had second thoughts – ignoring the numerous corrections and additions which were made to it, some perhaps almost immediately, some perhaps not till after some lapse of time. Alterations of one kind or another, large or small, occur throughout the text. All of them are potentially significant, but some are more obviously significant than others.

1 The first addition to be discussed is the stretch of text – corresponding to a whole chapter in D and DB – which describes the lands of the bishop of Winchester in Somerset (173v–5v). Of the three counties covered by the CDnCoSo booklets, Somerset is the only one in which the bishop owned property; so this stretch of text is the only one that we expect to find. Its abnormal character is not instantly obvious. In substance (except for the final sentence), the description conforms with the rest of the C text. It gives the usual information, in the usual arrangement, in the usual form of words. From reading Ellis’s (1816) edition, one would not suspect (until one reached the last few lines) that there was anything odd about it. Only the fact that it starts on a verso page might perhaps seem strange.

In this chapter I look at a few of the additions that came to be made in C. Though differing in nature, they have two common properties: they were made by scribes who had played no part in writing the primary text, and they tell us something about the uses that this manuscript was put to, after it had left the hands of its creators. To be specific, the additions in question are these: (1) a last-minute revision of part of the C text, made by a scribe who was acting on orders from the king, mediated by the bishop of Durham; (2) a number of memoranda written into C by the scribes who were writing the D text; and (3) two trial passages written on blank pages of C by the DB scribe. I discuss them in this order, because that is the order in which I take them to have been written. Each section of this chapter, however, is self-contained: the conclusions arrived at in one section do not depend on those arrived at in the others. The chronological implications are spelt out separately in the concluding section.

In the original, however, as Ker (1977, p. 806) was first to remark, this stretch of text, in two respects, is conspicuously different from the rest of C. First, it was written by a scribe who makes only this one appearance in the manuscript. Though Ker qualified his statement (‘does not seem to occur elsewhere’), the qualification can be dispensed with. The hand is very distinctive, and this stint is certainly unique (Chaplais 1987, p. 75). This scribe is, for instance, the only one who has the tiresome habit, when he is writing numerals, of elongating alternate minims, if there are more than two of them.2 Thus he writes ılı for iii, ılıl for iiii, vılı for viii, Lılıl for liiii. Less obtrusively done, this might be a good idea; but here it becomes an eyesore. One other detail worth noting is the fact that a red initial near the foot of 173v is the only splash of colour in the manuscript (Ker 1977, p. 807). The C scribes, knowing

Readers familiar with the existing literature will not have been surprised by the numbered statements in the previous paragraph. They may or may not agree with them; they will not think them novel. I do not dispute that. The evidence that I shall cite has all been cited before. The interpretations that I shall propose are largely unoriginal; in fact, they are fairly obvious. In particular, a paper by Chaplais (1987) covers all the same ground that I shall be covering here. Nevertheless, on all these points it seems to me that there is more to be said, and I aim to say some of it. After this, there

1 2

((They are dealt with in chapter 9 (below, pp. 104–6).))

Chaplais (1987, p. 75) pointed out this feature of the script; he also pointed out that the DB scribe occasionally does the same thing. (But he only does it for decorative effect, when he is writing in red: this scribe does it all the time.) In some other features too, such as the treatment of round d, there is a distinct resemblance between this and the DB scribe’s hand, pronounced enough to suggest to Chaplais that the former scribe might have been trained by the latter. (His other proposition, that this training took place in Durham, needs to be considered separately. It derives from an assumed dichotomy – ’If [a scribe] was not attached to a royal department, he must have belonged to an ecclesiastical scriptorium’ (Chaplais 1987, p. 72) – which I consider false. These are not the only possibilities.)

71

The survey of the whole of England that these booklets would soon be discarded, did not think of adding decoration; this scribe had different ideas. As Ker summed it up, the text written by this scribe, unlike the rest of C, ‘appears to be a fair copy’ (Ker 1977, p. 805) – more explicitly, as Chaplais put it, ‘a fair copy of an earlier version which [was about to be] discarded’ (Chaplais 1987, p. 75).

bishop of Durham, in conversation or by letter. The bishop of Durham, we may imagine, replies regretfully but firmly: he would help if he could, but is bound by his instructions from the king. He has no choice but to forward to the treasury the facts that have been reported by the local jurors, as they appear in the B text. One word from the king, however, would alter the case. If the king orders him to make the correction, then of course he will be delighted to oblige his brother bishop by making it. Time is short, but it may still be possible to put things right before the final text (i.e. the D text) is written up. If Walkelin will speak to the king, when they all meet in Salisbury, perhaps the problem can be resolved to everyone’s satisfaction. Walkelin acts on this advice. At Salisbury, with the bishop of Durham at his side, he buttonholes the king, explains the facts to him, and appeals for him to intervene. Thus prompted, the king agrees that he remembers granting these lands to Saint Peter and bishop Walkelin;7 and he commands the bishop of Durham ‘to write this same grant of his in the records’ (ut hanc ipsam concessionem suam in breuibus scriberet).8

Second, this stretch of text does not occupy a booklet by itself: it was written on some of the blank pages at the back of a booklet (fos. 161–75) which already contained the stretch of text describing someone else’s lands – the abbot of Glastonbury’s lands in Devon (161r1–8) and in Somerset (161r8–173r5). That description is a normal portion of the C text, written by scribes who worked on it regularly (above, p. 56). The booklet had no anomalous features until it was chosen to accommodate this addition;3 and it was chosen for no reason, as far as one can see, except that it happened to contain enough blank space. But we can take the story further than that. The last few lines that this man wrote are – and were intended to be – an explanation of the circumstances in which this addition came to be made. The sentence in question comes at the end of a paragraph describing two small manors, Lidiarda and Lega, the status of which has evidently been the subject of some disagreement.4 On the day when king Edward was alive and dead, both manors belonged to a nameless thegn – a free man by definition, but one who enjoyed the specific freedom of being ‘able to betake himself to any lord’ (potuit ire ad quemlibet dominum).5 Now they belonged to two men: Wulward had two hides worth 480 pence at Lidiarda, and Alward had half a hide worth 60 pence at Lega. This, it seems, was the statement of the facts reported by the local jurors. But there were certain dues arising from these lands which had always been paid to the (much larger) manor of Taunton, and it was arguable, therefore, that these two places ought to be counted among the dependencies of that manor. This, clearly, was the view of the case preferred by the bishop of Winchester, who happened to be the owner of Taunton. Put into feodal terms, the question at issue was whether the men who now possessed these lands held them from the bishop or directly from the king.

The addition in the C text was made in consequence of that command from the king. It was written, presumably, not by the bishop himself, but by a trustworthy scribe sent to Winchester for the purpose (escorted there perhaps by bishop Walkelin, to make sure that he did not get lost). Arriving at the treasury, this scribe would presumably have presented a letter from the bishop of Durham, ordering that he be given access to the booklets which contained the C text for Somerset. After a suitable amount of grumbling, the treasury officials complied; and the bishop of Durham’s scribe set himself to work. He had brought his pen and ink with him (red ink as well as black), but he did not have his own parchment. So he looked for a booklet which happened to end with an adequate number of blank pages, and, having found one, copied out this stretch of text, ending with the sentence which authorized him to make this alteration.9 De his terris semper iacuerunt consuetudines et seruitium in Tantone, et rex W(illelmus) concessit istas terras habendas sancto Petro et Walchelino episcopo, sicut ipse recognouit apud Sarisberiam audiente episcopo Dunelmensi, cui (who had not thought it necessary to resign from the bishopric of Winchester when he became archbishop). It is said, but only vaguely said, that Walkelin was related to the king. He died in 1098.

The bishop – whose name was Walkelin – sprang into action.6 Perhaps he began by expressing his annoyance to the

7

As the reader will either know or guess, Saint Peter was the patron of Walkelin’s cathedral church.

3

But it did acquire one odd feature when (possibly well into the twelfth century) somebody took it into his head to use the blank space on 173r for adding a statistical summary of the abbot’s manors in Somerset (173r6– 17). This appears to be a loose copy of the corresponding paragraph in the summary compiled by scribe mu (528r). (Some of the variants might suggest that the relationship was not so simple, but in the nature of the case they cannot amount to proof.) It is not clear who made this addition, or when, or why; nor is it clear that we should care.

8

Not in breuibus suis, as this passage is sometimes quoted; just in breuibus. Used collectively like this, the term breves or brevia (which in French would be bries) seems to mean some assemblage of official records. In the singular (normally neuter) it often means a writ; but it can also mean an itemized list, or something of that sort. (In 1168, for example, the itinerant justices who had just visited Kent submitted to the treasury lists of payments to be demanded individually from at least 90 persons in Milton, at least 75 persons in Canterbury; and these lists are referred to as br’ for breue in the singular, breuia in the plural.) A ‘brief’ should be short; if it cannot be short, at least it should be concise and businesslike.

4 They are identified by Thorn and Thorn (1980) as Lydeard St Lawrence (about 12 km north-west of Taunton) and Leigh (in the same parish). 5 This is a legal formula. What meaning it had in the real world is a question that has often been discussed. I have nothing to add. 6

9

This passage (175r20–v3) is reproduced by Chaplais (1987, pl. III, a–b), together with the corresponding passage in DB-So (pl. III, c).

Walkelin was elected in 1070, after the deposition of archbishop Stigand

72

Alien interventions precepit ut hanc ipsam concessionem suam in breuibus scriberet.

then, the bishop of Durham has been in attendance, helping the king to make up his mind, issuing the orders, verbal or written, which put the king’s decision into effect.12 And the records which were being written in 1086 come back into the picture now: the bishop of Durham can assure the king that the facts of the matter are stated there – meaning, presumably, in D-So.

From here this sentence was copied word for word into DSo, and from there it was copied word for word into DB-So (87va). It is clear, therefore, that the especial importance of this passage was recognized both by the D scribes (or at least by the one who worked on this stretch of text) and by the DB scribe; and the presumption is that similar passages elsewhere, if they had existed in C, would be found in D or DB, whichever survives. That none are to be found tends to prove that none existed. Bishop Walkelin, it seems, was the only man who managed to get some portion of the C text rewritten.

2 Various layers of annotation have accrued around the edges of C. In this section I discuss an assortment of memoranda – fourteen in all – which can be identified as the work of the scribes who were writing the D text, using C as their exemplar. These memoranda are short, relatively few, and scattered through the manuscript. It requires some leap of imagination before one can see that they are interconnected. They are linked together, to some extent (into groups of three and eleven items respectively), by similarities in the wording. But the basic property they share is the fact that they make no sense except on the assumption that the scribes who wrote them were writing a copy of C.

Making sure that the facts were stated correctly in the written record was a hollow success in itself; it was the facts existing on the ground which had to be clarified. Wulward and Alward, supported by the local jurors, had denied the bishop’s claim that they held their land from him, and the bishop wanted their denial contradicted. We are not told, but are expected to assume, that some order went out notifying the shire court that the bishop’s claim had been vindicated by the king.10

How many hands are represented I have not been able to decide. It was Chaplais’s opinion that most of these notes were written in a single hand contemporary with C, but that three of them, ‘probably all in one single hand’, were significantly later than C, by a margin of ‘several decades’ (Chaplais 1987, pp. 66–7). It seemed to him, therefore, that C was copied twice, firstly soon after it was written, but then again later, ‘some time in the twelfth century’. For my part, I cannot believe that the evidence is strong enough to carry this extra weight. The dating of the script is not a subject on which I expect my opinion to count for much, but it seems to me that we know rather little about the kinds of script that were used in the late eleventh century for writing ephemeral documents (documents which, at the moment when they were written, were not intended to be kept for very long) and might also be used for casual notes like these.13 Besides, it seems hard to imagine any circumstances in which somebody might have wanted to make a copy of C (or of C-Dn, or of some part of C-Dn) in the twelfth century, when the information was already long out of date, and when copies of extracts from the official records (D or DB) were obtainable from the treasury.14

We know nothing of the sequel, except for one interesting fact. Some years later, a writ was issued by Willelm II reinforcing that previous message (Galbraith 1920, p. 388).11 The shire court is to know, says the king, that he has been made aware, by the bishop of Durham and by his records (per breues meos), that his father granted Lidiarda and Lega to the church of Saint Peter of Winchester and bishop Walkelin; and he now grants the same manors to the church and bishop in perpetuity. W(illelmus) rex Anglorum I(ohanni) episcopo et W(illelmo) Capre uicecomiti et omnibus fidelibus suis Francis et Anglis de Sumerseta salutem. Sciatis me recognouisse per Willelmum Dunelmensem episcopum et per breues meos quod pater meus concessit Lidiard[am] et Legam ecclesie sancti Petri de Wintonia et Walkelino episcopo, et ego similiter eadem maneria concedo predicte ecclesie et episcopo in perpetuum habere. T(estibus) episcopo Dunelm(ensi) et Ragnulf(o) capellano. There is a hint here that the shire court had thought it doubtful whether the grant by Willelm I was meant to last for all time, or whether it was in the nature of a temporary injunction; the new writ removes that doubt.

12

Is it worth asking whether the second man who witnesses this writ, Rannulf the chaplain, had also had some part to play on the previous occasion? Was he the man sent to Winchester?

Events have repeated themselves. Just as in 1086, the bishop of Winchester has asked the king to act. Now as

13 Of the scribes who worked on C, there are only two who seem so accustomed to writing such documents that they have developed what I would regard as an informal manner: mu, who sometimes writes informally, and alpha, who always does. Is it not possible that a sample of alpha’s work, taken out of context, might also be thought to date from the early twelfth century? There was a time when DB itself was thought by some ‘very possibly’ to date from about 1130 (Johnson and Jenkinson 1915, p. 94).

10 Somebody reading between the lines, probably the DB scribe (but possibly one of the D scribes), took this to be the effect: Modo tenent de episcopo Wluuardus et Aluuardus per concessionem regis W(illelmi) (DBSo-87va).

14 Chaplais suggested that this copy was made, possibly for official purposes, ‘but more probably for the private use of some religious establishment’ (Chaplais 1987, p. 67). I think the suggestion would have to be made more specific than that before it could carry conviction.

11

The bounds on the date of this writ are the bishop of Durham’s return from exile in 1091 and his death on 2 January 1096. (But a date in the latter half of 1088 is perhaps not absolutely out of the question.)

73

The survey of the whole of England There are several aspects of the palaeographical evidence which need to be looked at again, by someone with an eye which (unlike mine) has been properly trained for the task.15 For the moment, however, I am inclined to assume that these memoranda were the product of one operation, not two. If that is right, the interpretation will not be greatly affected (as far as I can see) by our knowing exactly which notes were written by exactly which scribe.

engaged on a fair copy’.18 For Galbraith (1942), that comment took on a more specific significance. It seemed to imply that the DB scribe would have worked from a fair copy of C, not from C itself. In other words, it seemed to imply the existence, for the counties in question here,19 of a version of the survey text different from both C and DB, comparable rather with D-ExNkSk. On this point, in my view, Galbraith was exactly right.20

With respect to their wording, the memoranda divide themselves into two groups. The first group consists of three notes in which the word scripsit occurs, with the implicit meaning that what has been written is part of a copy of C. (Some part has already been written; some other part has not.) All three of these notes are associated with C-Dn.

The evidence is most transparent in the case of note (iii). The original text (414r1–8) was written by scribe epsilon; there is nothing extraordinary about it. Into this text – in the middle of a line, in the middle of a sentence – a different hand inserts a sign, with a matching note in the margin. The message is addressed to all scribes other than R . It tells them that R has copied the text as far as the word uocatur, and therefore that whoever continues the work should start with the word after this, Hanecheforda. In the absence of this note they would not know that; now they do. What happened, one would guess, is that R arrived at the end of a quire and downed his pen at that point. Either he or his supervisor thought it necessary to make this memorandum, because otherwise the scribe who would be writing the next quire might not be sure where to start. Whatever the precise circumstances may have been, it is clear that this note has nothing to do with the writing of C, and (if anyone thinks this a possibility worth considering) nothing to do with the writing of DB. The copy that was being made was a verbatim transcript of C, or something not very far removed from that.

(i) In the booklet (fos. 316–34) describing the lands of Judhel (corresponding to chapter 17 in DB-Dn),16 along the lower edge of the first page (316r) this note has been added: Hoc scripsit Ricardus, ‘Ricard has written this’. (ii) In the booklet (fos. 399–406) describing the lands of Willelm Capra (corresponding to chapter 19 in DB), along the lower edge of the last page (406v) this has been added: Hic debet esse hoc quod Iordan scripsit, ‘What Jordan has written should be here’.17 (iii) In the (incomplete) booklet (fos. 411–14) describing the lands of Ruald Adobed (corresponding to chapter 35 in DB), in the outer margin of the seventh page (414r) this has written as note appears: Usque huc scripsit R, ‘R far as this’. The man who wrote this note inserted a paragraph sign into the text to make it clear exactly what he meant: the sign is placed between the words uocatur and Hanecheforda (414r4).

These notes, in short, are traces of the activity of a team of scribes (two of whom even have names) who are making a fair copy of C-Dn. Since there are only three of these notes (and none at all for any of the other counties), it seems that they were generally not required: most of the time, the transcription proceeded without leaving any trace in the exemplar. Once in a while, however, somebody saw some risk of something going wrong, and made a note in the exemplar in order to stop that from happening. On just these few occasions, it was necessary for the other scribes to know

Notes (i) and (iii) were cited by Ellis (1817, p. 2); he assumed that they were made by the ‘different persons’ (several scribes) employed in the writing of C. They were cited again, and interpreted correctly for the first time, by the author of the comments on two facsimiles of pages from the Exeter manuscript which were published by the Palaeographical Society (Bond, Thompson and Warner 1884–94, plates 70–1). This author pointed out that these notes are plainly not connected with the writing of C itself; he deduced that they were ‘probably the memoranda of persons

18 ‘At the foot of f. 316 is the memorandum “hoc scripsit Ricardus”; and on the margin of f. 414, in another hand, “usque huc scripsit R.” As both these notes are in different hands from those of the text, it is evident that they cannot refer to the compilation of the present MS., but are probably the memoranda of persons engaged on a fair copy’ (Bond et al. 1884–94, letterpress to plates 70–1). This comment was quoted repeatedly but never quite accurately by Galbraith; so I reproduce it here in full.

15 Once it is understood that the C text and the D text were both compiled centrally, a question arises which did not arise before. Can any of the scribes who made additions in the surviving portion of the C text be identifi ed with scribes who worked on the surviving portion of the D text (Rumble 1987)? If the answer turns out to be yes, that will prove the point that C and D were produced in the same place. If the answer turns out to be no, that will tell us something (I am not sure what) about the arrangements involved in the production of the D text. Either way, we need to know.

19

Strictly speaking, the conclusion is valid only for Devon; how far to generalize from that is a separate issue.

20 The pity is that he immediately coupled this idea with two other notions which seem to me thoroughly wrong – that the production of C and the production of D were decentralized operations, and that D existed for no purpose except to serve as a source text for DB. Both of these errors have had pernicious consequences. The second has caused Chaplais (1987, p. 66) and others to doubt the existence of D. Why waste time making a fair copy? Why not let the DB scribe work directly from C? This line of argument seems to me to start from a false premise, the idea that D was optional, and I do not propose to discuss it. But I would ask the reader to bear it in mind that one cannot disprove the existence of D by collating C and DB, no matter how minutely one does it.

16 In the same booklet, along the outer edge of 317r, somebody wrote the word probatio, ‘trial’ (Ellis 1817, p. 2). I have no explanation for this. 17

This note seems first to have been cited by Thorn and Thorn (1985, note Exon. 17, 5), but they did not decipher it fully. It is difficult to read in the microfilm copy; I print it as it was printed by Chaplais (1987, p. 67).

74

Alien interventions what one scribe had already done, so that they did not go astray. Whether or not they were all written by the same hand, these notes should presumably be read as instructions from the man in charge, choreographing the work of his subordinates; and they make good sense if they are read in that way. The second group of memoranda, eleven in all, is characterized by the use of the word consummatum, ‘finished’. In nine cases, the words Consummatum est are written across a blank page occurring at the back of a booklet.21 Most of these messages, perhaps all of them, were the work of just one man, but the style varies from a fairly neat minuscule to an untidy display script. The same two words appear on the blank recto (155r) at the front of another booklet; 22 but here they were, originally, part of a much longer message, the rest of which has been erased.23 Finally, in one place the note Consummatum est usque huc is written in the margin against the end of a stretch of text relating to Devon, before the beginning of a stretch relating to Somerset (490r).24 More precisely it comes at the point which coincides with the end of the final chapter of DB-Dn. Its meaning seems clear enough: ‘The end of D-Dn, but not yet the end of D.’ Most of these messages are associated with the Somerset text,25 and we can work things out a little more carefully for this county.26 If we are willing to assume that the or21

The pages in question are: 209v, 370v, 387v, 449v, 451v, 455v, 467v, 474v, 494v. The microfilm copy that I have used omits 455v, probably because the photographer took it to be blank. That the note Consummatum est appears on this page is stated by Ker (1977, p. 804) and by Thorn and Thorn (1985, note Exon. 17, 5), who, however, add the comment ‘perhaps erased’. Like the photographer, Whale (1905, p. 264) and Finn (1957, p. 49) seem both to have overlooked it. ((I forgot to check this point later, when I had a chance to consult the original.)) 22

This booklet (fos. 155–60) is irregular (Ker 1977, p. 805). It began as a quire of four leaves (fos. 156–9), with the text starting at the top of 156r (‘Land of bishop Giso in Somerset’). The scribe (theta) who came to the middle of the quire continued across the opening, from 157v onto 158r, as if thinking that two more leaves would be suffi cient; but then he found that he needed more space, and so wrapped an extra sheet around the outside of the quire. Thus the booklet now begins with a blank leaf (fo. 155), onto the recto of which this note was written. (There is also a caption on this page, GISONIS Ep’i.)

C booklets

DB-So chapters

83–92 93–107 113–15 116 154 121–53 155–60 185–7 161–75 188–90 191–2 193 194–8 282–5 286–7 210–81 288–315 422–36 441–5 345–55 356–65 366–70 335–44 — 371–5 382–7 452–5 376–81 437–40 446–9 468–74 450–1 456–67 475–80 481–94

1 1 1 1 3 5 6 7 2, 8 9 10 14 11, 12, 13, 15, 16 17 18 19 20 21 22 24 25 27 30 33 35 28, 36 37 38 26, 40 23, 29, 34, 39, 41, 42 43 44 4, 31, 32, 45 45, 46 47

notes

155r

370v

387v 455v 449v 474v 451v 467v 494v

Table 19. Annotations made in C-So by a scribe at work on D-So.

der of the chapters in D-So was, approximately, the same as it is in DB-So, we can arrange the C booklets relating to So into a sequence which matches them up, approximately, with D (Table 19). (In some cases, a single C booklet corresponds with more than one chapter of DB; in such cases, it is the highest-numbered chapter which counts, because the C booklet would not have been finished with until the corresponding chapter in D had been written.) With the evidence presented in this form, it becomes obvious enough that the scribe who wrote these messages in C was the man who was writing the latter part of D-So.

23

The final word is decipherable as Wite, and presumably the note in its original form referred to fo. 116 (‘Land which used to be Ulward Wite’s in Somerset’). It said, I suppose, something to this effect: Finished except that some decision needs to be made with regard to the lands of Ulward Wite. Those lands in DB are entered at the end of chapter 1. 24

According to Thorn and Thorn (1985, note Exon. 17, 5), this marginal note was ‘definitely written . . . by the scribe of the surrounding entries’, i.e. by the scribe whom I call alpha. In my opinion the hand is certainly not his.

25 The exception is 209v, the last page of a booklet which relates to Cornwall alone; it corresponds with DB-Co’s chapter 4. There are hints – the fact that the C booklet was written entirely by alpha, the fact that in DB this chapter is transposed with chapter 3 – suggesting some irregularity about this stretch of text. The C booklet, I suspect, is a substitute, rewritten or recompiled by alpha after the original booklet had been rejected. It is not impossible that the corresponding chapter in D was, for some accidental reason, the last part of D-Co to be written.

order was correlated quite tightly with DB’s. For reasons explained in chapter 11 (below, pp.133–4), I do not think that now. But some loose correlation can safely be assumed – chapters which occur towards the end of DB-So are likely to have occurred towards the end of D-So too – and that is sufficient for the purposes of the present argument. In particular, the last chapter in DB is likely to have been the last chapter in D as well.))

26

((This and the following paragraphs have been revised. At the time when I first wrote them, I thought it reasonable to assume that D’s chapter

75

The survey of the whole of England the C scribes were expecting would be made, namely D. (At the very least we can be sure that it was written by a team of scribes, like the surviving D booklets, and that it was something close to a verbatim transcript of C.) Second, it seems tolerably certain that D-So was the last of the D booklets to be written. It was the DB scribe who changed the order of business here, by dealing with Somerset before he dealt with Devon and Cornwall. For the D scribes as for the C scribes, Somerset represented the end of the line. As work on Somerset neared completion, the celebrations could begin.

As far as I can see, these messages did not serve any serious purpose: the scribe was just amusing himself.27 There is one particular moment which we can visualize. Having finished the final chapter of D-So (or, to put it more cautiously, the chapter corresponding with the final chapter in DB-So, probably the final chapter in D-So too), the scribe turned to the back of the C booklet that he had been copying from and scrawled his parting words across the foot of the page: C O m S v m A t v’

e’ .

The moment when that message was written – the scribe by this time was forgetting how to spell – marked the end of the entire compilation process, as far as the D text was concerned. There may have been some checking still to be done; the DB scribe still had his own task to complete. For this scribe, however, the agony was over.28

3 Two passages, respectively seven and eleven lines long, written on pages left blank by the C scribes (153v, 436v), are the work of the DB scribe. A note in Ellis’s introduction draws attention to these entries,30 but says nothing of their possible significance. The credit for making them known belongs to R. Welldon Finn, who published a short paper on the subject, illustrated with good photographs of the passages in the Exeter manuscript and of the corresponding passages in DB (Finn 1951).31 Rather sadly, though Finn understood that this evidence meant something important, he did not even come close to understanding what that something might be. In this article, as in several books which he wrote later, on and around the subject of DB, he was labouring under false assumptions from which he never managed to break free.32 It makes no difference that these were Galbraith’s assumptions, not his own. The end result was the same.

If D-So survived, I infer, the last section of it would prove to have been written, all or mostly, by the scribe who recorded his progress by writing these gleeful messages in the C booklets. Towards the end the script would probably deteriorate, and spelling mistakes would multiply. The same scribe was apparently involved in the writing of D-Dn, on the evidence of 490r, and of D-Co, on the evidence of 209v. If those booklets survived, his hand would probably be represented there too. In D-Dn, furthermore, the end of the chapter describing the lands of Judhel ought to coincide with the end of a scribal stint (the name of the scribe being Ricard); the start of another chapter, the one which followed the chapter for Willelm Capra, ought to coincide with the start of a quire written by a different scribe (whose name was Jordan); another quire (written by R ) would end about two-thirds of the way through the chapter describing the lands of Ruald Adobed, with the word uocatur,29 and the next quire would be somebody else’s work. It seems that the writing of D-Dn was not a straightforward business, and we might expect that irregular quires and blank pages would be the proof of that.

He took it for granted, first, that DB was the work of a plurality of scribes.33 On this view, what DB represents is the style of script and the technical terminology cultivated inside a government department. The scribe who wrote these passages in the Exeter manuscript was someone who had been trained in that department; but there was no way of knowing whether he had also been involved in the writing of DB.34 Finn was trying to fit this evidence into the

These are empty predictions, because there is not the slightest chance of them being put to the test. Even so, I think we are entitled to hold on to two points. First, whatever predictions we think we can make are not fulfilled by DB. The manuscript which was being written when these notes were added to C was not DB; it was the copy of C which

30 ‘The hand-writing and colour of the ink of pages 153 b. and 436 b. are distinct from the rest of the Manuscript’ (Ellis 1817, p. 2, note 1). 31

Some parts of this paper sound to me as if they were written or rewritten by Galbraith; but there is no note acknowledging help from anyone.

32 Here and elsewhere, I have tried to give Finn any credit that he deserves, but I cannot say that I learnt much from reading his books. There are some useful ideas; but they never seem to be fully thought out, and often they are so allusively presented that it is hard to understand what he really means. In a word, the problem is that he let himself be bullied by Galbraith.

27

The C scribes too, when they saw the end approaching, became a little light-headed. While they were working on C-So, they entertained themselves by competing to see who could make the fanciest paragraph signs. The phenomenon was noted by Finn (1959, p. 364), but he does not seem to have asked himself what it might mean.

33

It is, we are told, ‘obvious that a considerable number of clerks were employed upon the production of the exchequer Domesday’ (Finn 1951, p. 561). There was no exchequer at the time. Not to quibble, we can impute to the treasury, which did exist, the characteristics which Finn imputed to the exchequer.

28 Perhaps it needs to be said (I have not seen it said before) that Consummatum est was the scribe’s idea of a joke. These were the last two words spoken by Christ from the cross (John 19:30). The previous word, I would guess, was on the scribe’s mind as well: Sitio, ‘I need a drink’.

34

These passages, he says, were ‘obviously produced by a royal curial scribe’ (Finn 1951, p. 562). That is fair enough; but the rest of the sentence is gratuitous: ‘and possibly by the one who wrote the corresponding entries in the exchequer Domesday’. Whether this scribe wrote the corresponding

29

Comparing this quire with Ricard’s stint would tell us whether R was the same man or not. In the absence of D, we are never going to know that.

76

Alien interventions overall interpretation proposed by Galbraith (1942). That interpretation is based on the assumption that DB was a departmental product; it collapses if one takes the view that DB was the work of an individual scribe. Unlike Galbraith, Finn was at least consistent in this respect. He never gave up the idea that DB was a collaborative effort. Second, he assumed, again like Galbraith, that the Exeter manuscript originated in Exeter. This assumption never had anything in its favour beyond its supposed coherence with the first assumption. A manuscript produced in a government department would have used the same sort of script and the same sort of language as DB; because the Exeter manuscript does not, it must have been produced somewhere else; 35 so why not assume (to keep things as simple as possible) that it was produced in Exeter?36 Reading Finn’s books, I do not find that he ever doubted this, or saw that it might be doubted.

a sequence of formulas adding up to a standard paragraph; the last few paragraphs that he has written will remind him what this pattern looks like, if he needs reminding. In his source text, D, there is a paragraph (also conforming to a pattern, though not the same as his) which contains the relevant items of information (and some irrelevant items too). To compose his next paragraph, he inserts the facts that he takes from D into the template that he is carrying in his head. As he composes, he writes. This is a risky procedure, but the DB scribe was not short of self-confidence. Nor was he an out-and-out perfectionist. He expected to make some mistakes; he expected to have to mar the appearance of his manuscript by making some corrections. Up to a point, he was willing to take the risk of writing something that he might regret having written. But sometimes the risk was higher than he could tolerate. Perhaps the next paragraph posed some particular problem; perhaps he had been away from work for some time, and wanted to be sure that he was back in the swing of things before he reverted to his normal modus operandi. In these circumstances, he did what any sensible person would do. He rehearsed. He found a piece of scrap parchment, wrote out a trial paragraph or two, and then read through what he had written. If he was satisfied with the result (sometimes, perhaps, he was not satisfied and decided to repeat the experiment), he copied his draft into DB, possibly making some small improvements as he did so. We have no idea how often this happened. We only know that it happened at least twice while the scribe was working on DB-So; and we only know that because on these two occasions the piece of scrap parchment which came to hand was a blank page on the back of a C booklet.

For Finn, therefore, the first question that had to be answered seemed to be this: how could a government scribe get his hands on this Exeter manuscript? There were only two possibilities (Finn 1951, p. 564). Either Mohammed or the mountain must have moved. Possibly ‘an official’ had been sent to Exeter; in that case the next step would be to wonder why. Or possibly the manuscript had been sent to Winchester but then ‘returned’ to Exeter. In his later writings, Finn hints at the explanation which he preferred: he suspected that a team of government scribes had travelled around the country, visiting Exeter and some other places where similar manuscripts existed, compiling DB on the spot. This would square the circle: the production of DB could be regarded simultaneously as a centralized and as a decentralized operation. But he never developed this argument far, presumably because he knew that Galbraith would not accept it.37

The details are as follows. (i) At the moment when the DB scribe was about to start work on DB-So’s chapter 4, ‘Land of the bishop of Bayeux’, he had within reach the C booklet (or part of the C booklet) corresponding to chapter 5, ‘Land of the bishop of Coutances’. This booklet (fos. 121–53) comprises six quires. The last three pages of the last quire (152v–3v) were left blank by the C scribes; the blank page at the back (153v) is where the DB scribe wrote his draft of chapter 4. (ii) At the moment when he was about to start work on chapter 33, ‘Land of Robert son of Gerold’, he had within reach the C booklet corresponding to chapter 21, ‘Land of Roger de Corcelles’. This booklet (fos. 422– 36) comprises two quires. The last two pages of the second quire (436r–v) were left blank by the C scribes; the blank page at the back (436v) is where the DB scribe wrote his draft of chapter 33.

All of this is either wrong or beside the point. DB was not the work of a team of robots: it was the work of just one man. The Exeter manuscript did not originate in Exeter: it was written in the treasury at Winchester. Once these facts are known – and I do not think that facts is too strong a word – the question that puzzled Finn does not arise. Instead we need to look at the evidence afresh. Once we do that, the explanation is obvious. These passages written into C by the DB scribe are exactly what they look like. They are trials – draft versions of passages which the scribe was about to enter in DB. By and large, the DB scribe did not make himself a written draft. To the extent that it existed, the draft existed only in his head. Imagine him at the moment when he is about to write one more paragraph in DB. In his memory, there is entries in DB, whether he wrote any entries there at all – these questions, on Finn’s view of the case, are not decidable.

The textual evidence is presented in Tables 20–21. With regard to chapter 4, we are lucky enough to have the C text surviving as a proxy for the D text.38 Thus we can say that

35 The same goes for D-ExNkSk, which is ‘obviously not an exchequer production’ (Finn 1951, p. 561).

38

The paragraph in C uses a number of special-purpose abbreviations which the reader may not be familiar with. In extended form it would read: Samson capellanus habet (de episcopo baioccensi) unam mansionem quae uocatur Coma, quam tenuit Liuuinus Comes die qua rex Eduuardus fuit uiuus et mortuus et reddidit gildum pro viii hidis, has possunt arare

36

Why not Salisbury? asks Webber (1989), echoed in advance by Chaplais (1987, pp. 67–8). 37

But Chaplais (1987, pp. 70–1) was willing to consider it.

77

The survey of the whole of England scribe theta in C (467r5–14)

DB scribe in C (436v)

Samson capellanus h’t (de ep’o baioccensi) i mans’ qu˛e uocatur Coma, quam ten’ Liuuinus Comes die qua rex E f u 7 m 7 redd’ gildum pro viii hid’, has poss’ arare viii carr’. Inde h’t Samson in d’nio v hid’ 7 iii carr’ 7 uill’ iii hid’ 7 ii carr’. Ibi h’t S x uill’ 7 vi bord’ 7 (v)ii seruos 7 ii roncinos 7 7 vi animal’ 7 xx porc’ 7 c oues 7 xxv capras 7 lx agr’ nemusculi 7 xl agr’ prati 7 xl agr’ pascu˛e, 7 ual& (l’ redd’) per annum x lib’, 7 quando Samson recepit, ualebat tantundem. Huic addita est i mansio qu˛e uocatur Turnietta, quam ten’ Aluuardus (pariter) die qua rex E f u 7 m’ 7 redd’ gildum pro iii uirg’, has potest arare dim’ carr’, 7 ual& per annum xiiii sol’, 7 quando S recepit, ualebat tantundem.

ROBERTI FILII GEROLDI. IN SVMMERSETE. Robertus ten’ CERLETONE (7 Gozelinus de eo). Godman tenuit TRE 7 geldb’ pro v hid’. T’ra e’ xii car’. In d’nio s’t ii hid˛e 7 ibi iii car’ 7 vii serui 7 iiii uill’i 7 xv bord’ 7 iii coscez cum viii car’. Ibi molin’ redd’ v sol’ 7 l ac’ prati. Pastura iiii quar’ lg’ 7 iii quar’ lat’. Silua dimid’ leu’ lg’ 7 tntd’ lat’. Valuit x lib’. Modo vi lib’. Idem Robertus ten’ ...... Vitel tenuit TRE 7 geldb’ pro x hid’. T’ra e’ x car’. De ea s’t in d’nio iiii hid˛e 7 ibi iii car’ 7 viii serui 7 iiii coliberti 7 xi uill’i 7 xvii bord’ cum v car’. Ibi xxx ac’ prati 7 c ac’ pastur˛e. Silua iii quar’ lg’ 7 ii quar’ lat’. Valuit xviii lib’ quando R recep’. Modo redd’ x bacones 7 c caseos.

DB scribe in C (153v) DB scribe in DB-So (97ra)

Ep’s BAIOCENS’ ten’ COME 7 Sanson de eo. Leuuinus (com’) tenuit TRE 7 geldb’ pro viii hid’. T’ra e’ viii car’. De hac t’ra s’t in d’nio v hid˛e 7 ibi iii car’ 7 vii serui 7 x uill’i 7 vi bord’ cum ii car’. Ibi xl ac’ prati 7 xl ac’ pastur˛e 7 lx ac’ silu˛e minut˛e. Valuit 7 ual’ x lib’. Huic adiunct’ e’ TVRNIE. Aluuard tenuit pro Man’ TRE 7 geldb’ pro iii virg’ t’r˛e. T’ra e’ dim’ car’. Valuit 7 ual’ xiiii solid’.

TERRA ROBERTI FILII GEROLDI. ROBERTVS filius Girold ten’ de rege CERLETONE 7 Godzelinus de eo. Godman teneb’ TRE 7 geldb’ pro v hid’. T’ra e’ xii car’. In d’nio s’t iii car’ 7 vii serui 7 iiii uill’i 7 xv bord’ 7 iii coscez cum viii car’. Ibi molin’ redd’ v sol’ 7 l ac’ prati. Pastura iiii quar’ lg’ 7 iii quar’ lat’. Silua dimid’ leu’ lg’ 7 t’ntd’ lat’. Valuit x lib’. Modo vi lib’. Ipse Robertus ten’ ...... Vitel teneb’ TRE 7 geldb’ pro x hid’. T’ra e’ x car’. In d’nio s’t iii car’ 7 viii serui 7 iiii coliberti 7 xi uill’i 7 xvii bord’ cum v car’. Ibi xxx ac’ prati 7 c ac’ pastur˛e. Silua iii quar’ lg’ 7 ii quar’ lat’. Quando recep’: ualb’ xviii li’. Modo redd’ c caseos 7 (x cem) bacons.

DB scribe in DB-So (87vb) Ep’s BAIOCENSIS ten’ COME 7 Sanson de eo. Leuuinus (com’) tenuit TRE 7 geldb’ pro viii hid’. T’ra e’ viii car’. De ea s’t in d’nio v hid˛e 7 ibi iii car’ 7 vii serui 7 x uill’i 7 vi bord’ cum ii car’. Ibi xl ac’ prati 7 xl ac’ pastur˛e 7 lx ac’ silu˛e minut˛e. Valuit 7 ual’ x lib’. Huic M adiunct˛e s’t iii uirg’ t’r˛e in TORNIE. Aluuardus tenuit TRE pro uno M 7 pro tanto geldb’. T’ra e’ dim’ car’. Valuit 7 ual’ xiii sol’.

Table 21. Two versions of DB-So chapter 33. and I do not see that there is much to be gained from discussing them in detail. But one point at least is worth noting. In the draft version, the DB scribe is found using a formula – De hac t’ra s’t in d’nio .. hid˛e – which he had used for a time while working on DB-Wi before deciding to replace it with something simpler – De ea s’t in d’nio .. hid˛e (above, Fig. 5).40 In copying his draft into DB, he makes the same change here. With regard to chapter 33, neither C nor D is available, and the two versions that survive can only be compared with one another. Even so, there is one difference between them of which the sense seems clear. In the draft version, as in both versions of chapter 4, he uses the formula tenuit TRE et geldb’; in the final version he replaces this with teneb’ TRE et geldb’. This change reproduces a transition that occurred precisely while he was working on DB-So (above, Fig. 6).

Table 20. Three versions of DB-So chapter 4. there is nothing in the final version (DB-87vb) which could not derive from the draft version (153v), while the final version has one numerical error (xiii for xiiii) from which the draft is immune.39 The differences in wording are slight, viii carrucae. Inde habet Samson in dominio v hidas et iii carrucas et uillani iii hidas et ii carrucas. Ibi habet Samson x uillanos et vi bordarios et (v)ii seruos et ii roncinos et et vi animalia et xx porcos et c oues et xxv capras et lx agros nemusculi et xl agros prati et xl agros pascuae, et ualet (uel reddit) per annum x libras, et quando Samson recepit, ualebat tantundem. Huic addita est una mansio quae uocatur Turnietta, quam tenuit Aluuardus (pariter) die qua rex Eduuardus fuit uiuus et mortuus et reddidit gildum pro iii uirgatis, has potest arare dimidia carruca, et ualet per annum xiiii solidos, et quando Samson recepit, ualebat tantundem. The corrections are by the original scribe, with the exception of the fi rst one, made by another hand (which I cannot identify). These interpolated words are missing from Ellis’s (1816) edition, and from the excerpt printed by Finn (1951, p. 562). But they are, of course, crucially important if we want to know why this paragraph became a separate chapter in D-So, or why Sanson’s claim to an exemption from geld was disallowed (above, p. 64).

In the absence of any indication to the contrary, I think we can be confident that the obvious explanation is the right one. In both cases, what we have been calling the draft 40

Unless anyone thinks they can prove that D’s wording differed significantly from C’s, in ways which brought it closer to DB’s, there is a larger lesson to be learnt from Table 20. Apparently it was the DB scribe’s choice to use the formula De hac terra s’t in d’nio, his choice again to replace it with De ea s’t in d’nio. The wording that we find in DB, the changes of wording that we find in DB, were not dictated by the source text. They were the result of decisions made by the DB scribe himself.

39

This error was noted by Finn (1951, p. 563). It is, by the way, the only demonstrable instance of an uncorrected error for which the DB scribe is personally to blame.

78

Alien interventions version is indeed the earlier version; and what we have been calling the final version was indeed copied from the draft version. The scribe was at liberty to glance back at D-So, if that was what he wished to do, but there is nothing to prove that he did.

C-Wi C-Do C-Dn C-Co C-So D-So

How far to extrapolate from these facts is difficult to decide. Because in one case the C booklet corresponds to a chapter not yet written, in the other to a chapter written some time before, it may be fair to assume that the DB scribe had the whole collection of booklets available to him. On the other hand, because both cases relate to one county, it may be too much of a stretch to assume that he dealt with every county in the same way. Nevertheless, for Somerset at least, we know what was going on. While he was working on DBSo, the DB scribe had his source text, D-So, to hand; he also had some or all of the booklets containing the source text for his source text. All three versions of the survey text that were centrally produced – C, D, DB – were brought together for some interval of time, within the DB scribe’s reach.

DB-So DB-Dn DB-Co This is not an achievement to boast about. If it could be completed, this table would have 99 rows, with one entry in each row and 33 entries in each column; the table that we know enough to construct has nine rows, and the columns contain five, one and three entries respectively. Though it may be possible to make a little more progress than this,41 I cannot imagine that we shall ever come close to tabulating the sequence of events in full. The worst news is that we have no clue how much overlap existed between the compilation of one version and the next – how soon D was started with respect to C, how soon DB was started with respect to D.

Presumably this means that the DB scribe has some thought of checking with C if he came across any passage in D that he mistrusted; but how far he actually did this is hard to say. The fact that he made no corrections in C counts for nothing. He would not have had any means of detecting errors in C, other than errors which were obvious but trivial (grammatical slips, spelling mistakes, and so on); and he would not have had any motive for correcting these. The only proof that these booklets passed through his hands is the trial paragraphs written on two blank pages. That is the only usefulness which we know he found in these booklets. For the DB scribe, the C booklets were waste parchment, or on their way to becoming such. The C scribes would have thought this conclusion a trifle premature. Fortunately for us, they had it in mind to make some further use of the C booklets. If that had not been the case, none of this evidence would have had any chance of surviving.

But there is some good news too. The intervention by the bishop of Durham’s scribe can be placed precisely in this sequence of events. It must be later than C-So; 42 it must be earlier than D-So.43 There would have been no point in rewriting this stretch of the C text if D-So had already been written: the scribe assumed, and assumed correctly, that there was still time for the substitution to be made. Furthermore, we know – because the scribe tells us so – that this intervention resulted from a meeting between the king and the bishop of Winchester attended by the bishop of Durham. By a stroke of luck we can discover, from other evidence, exactly when that meeting occurred. The king’s movements, during the middle months of 1086, are sketched out for us by a well-known piece of narrative which survives as part of the ‘E’ text of the ‘Anglo-Saxon Chronicle’ (Swanton 1996, pp. 216–17). The king was at Winchester for Easter (5 April), at Westminster for Whitsun (24 May); and then ‘afterwards he travelled about so that he came to Salisbury for Lammas’ (Saturday 1 August),

4 From what we now know, we can try working out the sequence of events. Starting with Somerset, we can take it as given that C-So was earlier than D-So, and that D-So was earlier than DB-So:

and his council came to him there and all the men occupying land 41 As for the second column, we can be sure that D-Dn and D-Co were later than C-So, fairly sure that they were earlier than D-So. But we have no means of determining which of the two was the earlier: for all we know, the D scribes may have decided to dispose of Cornwall before dealing with Devon. If these booklets were included in the table, therefore, they would have to be marked with queries.

C-So D-So DB-So

42

Strictly speaking, we cannot be sure that C-So had been completed. We only know that this stretch of text is later than two of the C booklets: the booklet which it superseded, and the booklet into which it was written. We are permitted to believe, if we wish, that the C scribes were still at work on some of the other booklets. But I do not see why we should wish to.

We also know that Somerset was the last of the last five counties to be dealt with by the C scribes, and that the DB scribe had two more counties to deal with after he had finished with Somerset; so we can add four more entries in the first column, two more in the third:

43 Again we are free to believe, if we wish, that the D scribes had already got started on D-So. It is the chapter corresponding to DB’s chapter 2 which we know had not yet been written.

79

The survey of the whole of England and performed the task that he had been sent there to perform.

who were of any account over all England, whichever man’s men they were, and all submitted to him and were his men, and swore him loyal oaths that they would be loyal to him against all other men.

C-Wi C-Do C-Dn C-Co C-So

From Salisbury the king travelled south to the Isle of Wight, intending to cross into Normandy; he stayed there for some time; and then he made the crossing.

August 1086

This meeting in Salisbury was an extraordinary event.44 The chronicler knew that, and did his best to explain the significance of the occasion, as far as he understood it. For present purposes, the chronological implications are the only point which counts. It is obvious that the meeting must have been planned well in advance; probably the place and the date were decided at Westminster, in late May. The king then set out on a tour of the country, organizing his itinerary around the necessity of arriving in Salisbury towards the end of July; and all the other people required to attend – those who had been present at Westminster and those who were now being notified – were expected to plan their movements around the same necessity. They should arrive in Salisbury, without fail, before 1 August.

D-So DB-So DB-Dn DB-Co At this point in the sequence of events, C-So was already in existence. If we can be sure (as I think we can) that Somerset was the last county to be dealt with by the C scribes, that conclusion implies a much larger one: the entire C text had been compiled before the beginning of August. The D text, though possibly begun, had not yet been completed. As of early August, D-So did not yet exist. When it was brought into existence, some time later, it incorporated the revised text that had been added in C by the bishop of Durham’s scribe. It was a scribe involved in writing the latter part of D-So who amused himself by writing messages (nearly there, nearly there, nearly there) into his exemplar. Both DB-So itself and the trial passages written into C by the DB scribe are later than D-So; the DB-DnCo booklet is later still.

The bishops of Winchester and Durham would certainly have been summoned to this meeting. If they wanted to waylay the king, this was a good opportunity, and perhaps the only opportunity for which they could plan ahead.45 To put it differently, we know for a fact that the interview took place in Salisbury, and therefore – as was emphasized by Holt (1987, p. 44) – we can say, with virtual certainty, that it happened on or very close to 1 August. The king had other places that he wanted to visit, so he was visiting them during June and July; but he also wanted to get back to Normandy, so he headed south as soon as the Salisbury meeting was over. He stopped in the Isle of Wight for a time, perhaps only while he was waiting for favourable weather, and then he was gone. After that, anyone who needed to speak to the king would have had to follow him across the Channel. Plus or minus a few days, 1 August is the latest possible date, and to all appearances the only possible date, for the bishop of Winchester’s interview with the king.46 Immediately afterwards (presumably within a few days), the bishop of Durham’s scribe made the short journey to Winchester 44

As far as the C booklets are concerned, this is how the story continues. After the DB scribe had finished with them, they were used in some further operation which required them to be sorted into stacks, but which was interrupted and not completed (above, p. 53): the booklets which survive are those which did not relate to any of the counties (up to and including Wiltshire) on which this operation had been performed before the interruption occurred. After some further lapse of time, the abandoned stacks of booklets were put together and bound up to make a book; and that is the book which – unwisely rearranged when it was last rebound – survives to this day in Exeter.

((There is more to be said on this topic later (below, pp. 120–1).))

45

It is possible, I think, that the bishop of Durham, in anticipation of this meeting with the king, has ordered the treasury officials to send him the C booklet relating to the lands of the bishop of Winchester in Somerset. Then, once the king has given them carte blanche, the two bishops work over the text together, deciding what revisions are needed. When the bishop of Durham’s scribe sets out for Winchester, he takes the C booklet with him. His instructions are to write out a clean copy of the emended text, and then to destroy the original. That would explain – what does seem to need explaining – why he copied out the whole stretch of text afresh, rather than making corrections to the original. 46

It may seem obvious now that the meeting at Salisbury mentioned by this scribe is to be equated with the meeting at Salisbury reported in the English chronicle; but it was far from obvious before this section of C was discovered to be a last-minute addition. As far as I am aware, Holt (1987), citing Ker (1977), was the first person who had ever made the connection.

80

Chapter 8 The monks of Ely and the records of the survey – Part I

For more than sixty years, it has been understood that the written record of the ‘Survey of the whole of England’ evolved through several versions (Galbraith 1942). The earliest version, the one which I call B, differed from the subsequent versions in being cadastrally organized: the manors which formed the units of the survey were listed county by county, hundred by hundred, village by village. Unlike these subsequent versions – C, D and DB – no part of the B text survives in the original.1 The existence of such a version could be inferred from the surviving portion of the C text, the version which served as a vehicle for replacing the cadastral organization of B with the feodal organization intended for D (and for the epitome of D that I call DB); but little would be known of the character of B if this were the only evidence available to us.

reproduce as it stood. It is a private text, the property of the monks of Ely; and Ely scribes are unlikely to have felt inhibited from revising and annotating it, as seemed good to them. Three manuscripts are in question. I call them T, U and V.2 (Some readers may wish to know straight away that these are the same manuscripts which Hamilton (1876) called B, C and A respectively.) Each of them contains a copy of xEl; one of them also contains the copy of part of B-Ca. All three were written, in the mid or late twelfth century, in the priory attached to the cathedral church in Ely.3 Presumably they remained at Ely until the mid sixteenth century, when the monks were evicted and their library was dispersed. One of the manuscripts (my V, Hamilton’s A) reappears at the beginning of the next century. By that time it had come into the hands of Arthur Agarde (1540–1615), who gave it to Sir Robert Cotton (1571–1631).4 With the rest of Cotton’s library, though not until after surviving various adventures, it found a safe home in the British Museum in 1753. The other two manuscripts reappear towards the end of the seventeenth century, when they were in the possession of Thomas Gale (1635/6–1702). It does not seem to be known how or where he acquired them. Gale’s son, Roger Gale (1672–1744), donated a collection of manuscripts to Trinity College, Cambridge, in 1738; and these two were included in that collection.

In fact, by a stroke of good luck, a copy does exist of part of B. Towards the end of the twelfth century, a scribe working in the monastery at Ely made a copy – had both the opportunity and the motive for making a copy – of some portion of the B text. Our luck is not all good, however, because unhappily some part of this copy had been lost before the seventeenth century. The part that survives begins at the beginning of Cambridgeshire, and covers about three-quarters of this county before breaking off in mid sentence, at the foot of a verso page. How much has been lost is uncertain. It is doubtful too whether the scribe was copying directly from the original or from an earlier copy; and in any case the format is one imposed by this scribe, not one intended to resemble the original. Despite these drawbacks, the surviving text appears to be a straightforward transcript – not as accurate as it might be, but not deliberately edited – of a large part of B-Ca; and that makes it uniquely valuable.

The text that I call xEl was printed by Ellis (1816, pp. 497– 528),5 who knew of the existence of one of the manuscripts 2 Hamilton’s (1876) edition of xEl includes facsimiles of one page from each manuscript, lithographed by Frederick Netherclift. They are very pretty, but only accurate up to a point: the image had to be traced from the original by hand. Early attempts at photographic reproduction – such as the Ordnance Survey’s facsimile of ‘Domesday Book’ (1861–3) – were no great improvement.

Also from Ely, three copies survive of a compilation of extracts from some version or versions of the survey text. I call this text xEl. It covers all six counties in which the abbey of Ely owned land, but deals only with those manors – never more than a small proportion of the total number – in which Ely had an interest. Because of its complexity, this text will need to be discussed at greater length than the copy of B-Ca. But the reader should not let that obscure the fact that xEl is much less important than B-Ca, and that its reliability is much harder to judge. This is not an official text – the sort of text which a copyist might feel obliged to

3

The monastery at Ely was founded by Æthelwold, bishop of Winchester, in 970. Originally it was an abbey, wealthier than most but unexceptional: at the time of the survey it was ruled by abbot Simeon. In 1109 the church became the see of a new bishopric, and the monastery became a cathedral priory.

4

As is stated by a note at the foot of the first page: Ro: Cottoni Liber ex dono Arthuri Agarde (36r); the date 1609 is written below this, seemingly in different ink. As a Treasury official with easy access to ‘Domesday Book’, Agarde was better placed than anyone else to appreciate the significance of xEl. This manuscript is cited in an essay which he wrote about ‘Domesday’ matters, and which was published later by Roger Gale, the then owner of the other two manuscripts (McKisack 1971, pp. 86–7).

1 Of the C text roughly one-eighth exists in the original (Exeter Cathedral Library 3500, fos. 25–62, 83–494, 530–1); of the D text, roughly one-sixth (Public Record Office, E 31/1); of the DB text, roughly five-sixths (PRO, E 31/2, fos. 0–372).

5 ((It turns out that Ellis published xEl by mistake. The proof lies in a passage quoted by Prescott (2001, p. 183) from the Record Commission’s minute book. Ellis’s description of the text which he was proposing to

81

The survey of the whole of England

1

in Cambridge, but who chose (unwisely) to base his edition on V, the manuscript which, because it was in the British Museum, happened to be more readily available to him.6 The other text, the copy of B-Ca to be found in the same manuscript, remained largely unknown until it was eventually put into print by Hamilton (1876). By way of an appendix, Hamilton also produced a new edition of xEl: like Ellis (unwisely again, and without the excuse of ignorance), he based his text on V.7 Variant readings from the two Cambridge manuscripts were cited in the footnotes, with only moderate accuracy; and several shorter texts which follow xEl in the manuscripts were printed for the first time here.

T is a mid twelfth-century manuscript, catalogued by James (1902, pp. 145–6) and briefly described by Robinson (1988, p. 102). The pages were numbered in the seventeenth century, possibly by Gale; but the numbering is unsatisfactory. It does not cover the first eleven leaves, and it goes wrong about two-thirds of the way through.9 For that reason, and for the sake of consistency, I have taken the liberty of introducing my own foliation (Table 22).

It took almost twenty years before Hamilton’s book made any particular impact;8 but when it did, the impact was profound. For Round (1895), the publication of this copy of B-Ca opened up a world of new possibilities. In a long, badly organized article (two or three articles packed into one, interrupted by numerous digressions), Round let himself loose on an exploration of this previously unmapped landscape. By the time that he was finished, he had defined the terms for a radical reassessment of almost all the evidence relating to the survey. For Maitland (1897) too, this edition of B-Ca made a world of difference. Many things could be seen with clarity here – things which became frustratingly obscured in subsequent versions of the survey text, and most of all in DB.

The original contents, the work of a single mid twelfthcentury scribe, are listed and discussed in the following paragraph; I start by stripping away the accretions due to other hands. The book begins with a calendar (4r–9v), written by a contemporary or not much later scribe; it is too sparse to be of much interest (James 1902, pp. 145–6). Elsewhere, within the body of the book, numerous additions were made by late twelfth-century and later scribes, not just on pages previously blank (44r, 91v, 150r–1v), but also in two whole quires (fos. 44–51, 152–9) apparently inserted for the purpose. With two exceptions (see below), these additions need not concern us any further; but they prove that T was still at Ely in the fifteenth century, and was still occasionally consulted.

My objectives here are unambitious; I aim only to come to grips with the textual evidence. For information about manuscripts from Ely, and about other Ely matters, I have relied very heavily on Blake’s (1962) edition of the Historia Elyensis insule (see below). Though Blake was only incidentally interested in xEl and not interested at all in B-Ca, his treatment of the manuscript evidence will need to be cited frequently. Perhaps it should also be said that chapters 8–9 are my first and last word on the subject. The comments that follow are not intended as prolegomena for a new edition of xEl or B-Ca or both. There is need for a new edition, more intelligently designed and more accurately executed than Hamilton’s, but I have no thought of producing it myself. Here and there, however, I have offered some gratuitous words of advice which I hope may be helpful to anyone who undertakes the task.

The first scribe, whom I call T1, writes an elegant rounded hand. At its best, the script is very good, but it deteriorates considerably when the scribe starts ploughing his way through the tedious statistical data which make up most of xEl.10 At the moment when it left the hands of scribe T1, the manuscript consisted of 132 leaves and 16 quires (all of 8 except for the last one). The format is uniform throughout, with 23 lines to the page. There are three constituent booklets:

T = Cambridge, Trinity College O. 2. 41

booklet 1 (fos. 12–43): Libellus quorundam insignium operum beati Aedeluuoldi episcopi (12r–43v), a hagiographical tract concerning the foundation of the monastery at Ely by bishop Æthelwold;11 booklet 2 (fos. 52–91): Collectio priuilegiorum eliensis e˛ cclesi˛e (52r–84r), a transcription of royal, papal and other

print – ‘an original return or inquisition . . . respect[ing] a portion of the county of Cambridge’ – makes it certain that he was speaking of B-Ca, not of xEl. Apparently whoever made the copy copied the wrong section of the manuscript – and apparently Ellis never noticed.))

9

The person writing the numbers forgot what he was doing and switched from pagination to foliation; then he remembered and switched back again. (At the moment when he realized his mistake, it would have been easy for him to put the numbering right, but he did not bother to do so.) The upshot was that two of the verso pages were left without a number. A modern hand, perhaps James’s, has patched things up by adding two extra labels (187A, 188A).

6 Ellis’s edition does not include the appendix of shorter texts (65ra–9rb) which follows xEl in V. The opening stretch of the main text (36ra–7rb) was printed again by Ellis (1833, vol. 1, pp. 22–7); the variant readings quoted here come from T, not, as Round (1895, p. 124) supposed, from U.

10

The page lithographed for Hamilton by Netherclift (above, note 2) is fo. 95r. The page reproduced by Robinson (1988, pl. 62) is fo. 84r (where the first ten lines were written by scribe T1, the rest by scribe T2).

7

Hamilton’s edition of xEl was announced by Hardy (1865, p. 36) as being already ‘in the course of publication’. Hardy knew that there were two copies of xEl in Cambridge, not just one; he was also aware that V contained a second, unpublished text (but apparently did not understand what it was, nor that Hamilton was planning to publish it).

11

The full text was printed, from this manuscript, by Gale (1691); the prologue and the first few chapters were printed again by Blake (1962, pp. 395–7). In the prologue the author represents the work as his Latin translation of an English original: he was assigned this task, he says, by bishop Herveus (d. 1131).

8

To say that it ‘took the learned world by storm’ (Galbraith 1961, p. 123) is a characteristic lapse into exaggeration.

82

The monks of Ely – Part I collation

original signature

modern signature

17th-century pagination

proposed foliation

4 wants 1

1–3

8

4–11

8 8 8 8

1 2 3 4

1–16 17–32 33–48 49–64

12–19 20–7 28–35 36–43

8

5

65–80

44–51

8 8 8 8 8

I II III IIII V

6 7 8 9 10

81–96 97–112 113–28 129–44 145–60

52–9 60–7 68–75 76–83 84–91

8 8 8 8 8 8 12

VI VII VIII IX X XI

11 12 13 14 15 16 17

161–76 177–90 191–206 207–22 223–38 239–54 255–78

92–9 100–7 108–15 116–23 124–31 132–9 140–51

18

279–94

152–9

8

Table 22. Collation of Cambridge, Trinity College O. 2. 41. (The seventeenth-century pagination skips two pages in quire 12 (see note 9).) dates from 1144; and a third scribe, T3, started adding yet another batch, but broke off in the middle of a word, at the foot of the last recto page (91r). The latest document here dates from 1152 – but we have no idea what other documents the scribe might have added after this one, if he had not been interrupted. His script is appreciably different in aspect from that of the first two scribes, late rather than mid twelfth-century.

documents from the Ely archive, beginning with three charters of king Eadgar; booklet 3 (fos. 92–151): a copy of xEl, followed by several shorter texts relating to the lands possessed or claimed by Ely (92r–149v). Booklets 2 and 3 are linked together by a sequence of quire signatures (Table 22). Apparently the scribe began by writing booklet 2; then he continued with booklet 3, thinking of it as an appendix to booklet 2. Later, as something of an afterthought, he added booklet 1, intending it to form a preface for booklet 2.

These facts were all recognized by Blake (1962, pp. xxxix– xl), who pointed out that they enable the work of scribe T1 to be dated within a very narrow bracket. Of the documents copied by T1, the latest are a batch of papal letters which were issued in Rome at the end of April 1139, but would not have been available in Ely till several weeks after that: it was the arrival of these letters, so it seems, which prompted scribe T1 to think of compiling a cartulary. Of the documents transcribed by T2, the earliest is another papal letter, dated October 1140; and it seems a fair assumption that scribe T1 would have copied this letter too, if it had been available to him. It is certain, therefore, that scribe T1 was at work on booklet 2 after about June 1139, and likely that he had stopped work before about November 1140 (cf. Blake 1962, p. xxxiv). A similar or only marginally later date can be assumed for booklet 3, and (more loosely) for booklet 1.13

The scribe’s primary interest, it seems clear, was in booklet 2; but our interests diverge from his. I list the contents of the cartulary (Table 23) but have little more to say about this booklet. The original text, all written by scribe T1, fills the first four quires and overflows onto the first page of the fifth quire (84r). Only ten more lines were needed to complete the final document (a papal letter dating from 1139), but the scribe made a whole new quire, as if in the expectation that the cartulary would be continued, by himself or by others, as new documents entered the archive. At any rate, that is what happened. A second scribe, T2,12 added a further batch of documents (84r–9v), the latest of which

13

Robinson (1988, p. 102) regards the change of hand on fo. 84r as merely the start of a new stint, not implying a significant lapse of time; on this view T1 was probably at work after 1144, as T2 certainly was. That is

12

The change of hand on fo. 84r (above, note 10) was localized vaguely by James (1902, p. 146), determined precisely by Blake (1962, p. xl).

83

The survey of the whole of England Trin. Coll. O. 2. 41

BL Cott. Tib. A. vi

Blake 1962

52r–4v 54v–5v 55v–7v 57v–9r 59r–60v 60v–2v 62v–3v 63v 63v–4v 64v 64v–5r 65r–v 65v–6r 66r 66r–v 66v–7r 67r 67r–v 67v–8r 68r–v 68v–9r 69r–v 69v–70r 70v–1v 71v–2r 72r–v 72v 72v–3v 73v–4r 74r 74v 74v–5r 75r–v 75v–6v 77r–8r 78r–9v 79v–82v 82v–3r 83r–v 83v–4r -----84r–v 84v–8r 88r–9r 89r 89r 89r–v 89v -----90r 90v–1r 91r ------

— — — 99r–v 99v–100r 100r–1r 101r–2v 102v 102v–3r 103r 103v 103v–4r 104r 104r 104r–v 104v 104v–5r 105r 105r 105r–v 105v 105v–6r 106r 106v 106v–7v 107v–8r 108r 108r 108r–v 108v–9r 109r 109r 109r–v 109v 109v–10v 110v–11r 111r–12r 112r–13v 113v 114r 114r

ii.5 9 39 58 82 92 93 95 116 117 120 121 122 123 124 125 126 127 136 iii.2 3 4 5 6 7 10 11 12 13 15 16 18 19 26 55 54 56 65 66 67

114r–v 114v–16v 116v 116v–17r 117r 117r 117r–v

68 85 79 84 80 81 83

117v 117v–18r 118r -----118v 118v–19r 119r–v 119v–20r 120r ------

95 105 104

Eugenius III

91

archbishop Teodbald

90 106 134

bishop Nigel archbishop Teodbald bishop Nigel

Eadgar

Æthelred Cnut Eadward Victor II Eadward memorandum Willelm I

memorandum Paschalis II

Henric I

bishop Herveus Innocentius II bishop Nigel Innocentius II

Lucius II

Sawyer 779 780 781 907 958 1051 Jaffé–Loewenfeld 4350 Sawyer 1100 Bates 122 120 119 121 124 123 127 125 126 Jaffé–Loewenfeld 6212 6213 6211 6210 Johnson and Cronne 919 1048 931 1543 1656 1500 1499 1421 1542 1501 Holtzmann 17 Holtzmann 21 22 23 24 27 35 36 37 39 38 40 63 69 71 Saltman 98 99 Saltman 100

Table 23. Contents of the cartulary initiated by scribe T1. (Changes of hand are denoted by broken lines.)

84

The monks of Ely – Part I More is at issue here than just a question of dating. To the extent that this manuscript had an author, scribe T1 deserves the credit. As far as the evidence goes, he was the man who decided that the time had come to create a cartulary – who chose these documents (not all so obviously important that they would choose themselves), put them into this sequence, and supplied a title for the whole collection – and who decided to append a copy of xEl to it. Large and small, the decisions made by scribe T1, while he was working on this manuscript, exerted an influence on later generations. They helped to give a definite shape to the monks’ conception of their communal history.

close approximation) from the footnotes in his edition. Like T, U begins with a calendar (quires 1–2), a more interesting specimen than the one in T.16 On internal evidence, this calendar is certainly from Ely, and of about the same date as the rest of the manuscript; but the scribe who wrote it seems to occur only here, not in the following booklets. Apart from the calendar, U’s contents are briefly as follows: booklet 1A (fos. 1–106): HEI, Books I–II (1r–106v); three inserted leaves (fos. 107–9): a table of contents covering the whole of the rest of the manuscript, booklets 1B–C and 2–3;

U = Cambridge, Trinity College O. 2. 1 U is a later twelfth-century manuscript, also catalogued by James (1902, pp. 79–82) and briefly described by Robinson (1988, pp. 100–1).14 As well as numbering the quires, James had some thought of introducing a new foliation, but he did not follow through with this plan. Like everyone else, I have used the seventeenth-century foliation, presumably Gale’s, which replicates a late medieval foliation written in Roman numerals; it ignores the first 14 leaves but from then onwards is perfectly correct (Table 24).

booklet 1B (fos. 110–51): HEI, Book III, as far as the middle of chapter 92 (110r–51v); booklet 1C (fos. 152–78): the remainder of HEI, Book III (152r–76r), followed after two blank pages by one segment of the xEl text (177v–8v), the segment that is missing from the next booklet; booklet 2 (fos. 179–214): a copy of xEl (179r–207v), largely the same as in T except that one segment of the text is missing, followed by two documents that do not occur in T (207v–9v, 210v–13v);

This manuscript is important, most of all, because it contains the earliest copy – partly a working draft – of the Historia Elyensis insule, ‘History of Ely island’, a lengthy history of the church and monastery compiled by one of the monks and completed, in its original form, soon after 1170. I refer to this work as HEI from here onwards. In fact there are two full-scale versions of HEI, as well as several derivative versions. The original version, represented by U, was revised and expanded into a second version, represented by a manuscript known as the ‘Liber Eliensis’, the property of the dean and chapter of Ely, which I cite (on the few occasions when it will need to be cited) as W. Where these versions need to be distinguished, I call them HEI / U and HEI / W; but in general the reader should assume that I am speaking of the earlier version. Reasonably enough, Blake (1962) chose to print the longer version of the text, as it appears in W (which he called F).15 But the version represented in U (Blake’s E) can be reconstructed (to a fairly

booklet 3 (fos. 215–40): six hagiographical tracts (215r– 40v). Three scribes can be identified who wrote long stretches of text.17 Scribe U1 (Blake’s hand A) wrote the first and larger part of booklet 1A (1r–76r), as far as the middle of Book II, chapter 90. He thus established a format for the book, with 29 lines to the page, which subsequent scribes were obliged to conform to, more or less closely. Since this stretch of text includes the table of contents for Book II (43v–6r), it is clear that the whole of Book II was already in existence (presumably in the form of a working draft),18 and that the change of hand after 96r, coinciding as it does with the turn of a leaf, means nothing more than what it obviously means – that the task begun by this scribe was continued by someone else. The same scribe, U1, wrote the copy of xEl in booklet 2 (179r–209v),19 including one of

not impossible; but I do not see that Robinson has any solid reason for thinking that her interpretation is preferable to Blake’s. 14

((Another brief description is given by Love (2004, pp. lii–iii). Unlike me, she sees a change of hand at 233r, i.e. at the beginning of the final quire.))

16 This is the calendar collated as C by Wormald (1946, pp. 8–19). It is later than 1170 (because it includes the feast of Saint Thomas archbishop and martyr), but probably not much later (because it does not include obits for Henric II or bishop Galfrid, who both died in 1189). The page for October is reproduced by Robinson (1988, pl. 85).

15

HEI / W is a conflated text, largely derived from U (Blake 1962, p. xlii) but with numerous interpolations derived from other sources. For example, many passages from the Libellus (above, p. 82), appearing only in shortened form in U, appear word for word in W: the author of HEI / W took his cue from U but went back to U’s source for the unabbreviated text. (The source, I assume, was T. Blake (1962, p. xxxiv) thought differently, on the evidence of a single variant (in a passage missing from U) which is, at best, ambiguous. It seems to me that T has the authentic reading (Summonetur Wlnothus; adduxit secum . . . ), and that the phrase appearing only in W (ed. Blake 1962, p. 99) is to be read as an addition made there (a rather clumsy interpolation by the author of HEI / W), not as an omission from T.)

17 As far as booklet 1 is concerned, I am mostly following Blake’s (1962, pp. xxiii–iv) lead; what is said about booklets 2–3 is mine, and Blake is not answerable for it. 18 As Blake (1962, pp. xlii–iii) pointed out, the order of the chapters advertised here is not quite the same as in Book II itself, even in the portion written by U1: some rearrangement took place, after this list was drawn up, before the draft was copied into U. 19

85

Hamilton’s edition includes a facsimile of a page (179v) from booklet

The survey of the whole of England collation

modern signature

8 6

1 2

8 8 8 8 8 8 8 8 8 10 8 8 8

3 4 5 6 7 8 9 10 11 12 13 14 15

17th-century foliation

James’s foliation 1–8 9–14

1–8 9–16 17–24 25–32 33–40 41–8 49–56 57–64 65–72 73–82 83–90 91–8 99–106

15–22 23–30 31–8 39–46 47–54 55–62 63–70 71–8 79–86 87–96 97–104 105–12 113–20

107–9

121–3

three 10 8 8 8 8

16 17 18 19 20

110–19 120–7 128–35 136–43 144–51

124–33 134–41 142–9 150–7 158–65

8 8 10 one

21 22 23

152–9 160–7 168–77 178

166–73 174–81 182–91 192

10 10 8 8

24 25 26 27

179–88 189–98 199–206 207–14

193–202 203–12 213–20 221–8

10 8 8

28 29 30

215–24 225–32 233–40

231–40 241–8 249–56

Table 24. Collation of Cambridge, Trinity College O. 2. 1. (The leaves numbered 229–30 by James (a sheet from a fourteenth-century music book) are a late insertion between quires 27 and 28. Two similar leaves, which he did not number, form flyleaves at the front of the book.) anything) was written by other hands, is difficult to decide (Blake 1962, p. xxiii); I have not looked at it closely enough to form an opinion of my own. The presumption is, in any case, that only one brain was at work. Like Blake (1962, p. xlvi), I take it that booklet 1B would originally have continued as far as the end of Book III. As things stand now, however, this final section of HEI is a fair copy, booklet 1C, written by a scribe who had not contributed previously.

the documents not present in T. Scribe U2 (Blake’s hand B) wrote the rest of booklet 1A (76v–106v), as far as the end of Book II, and the first section of booklet 1B (110r–25v), as far as the end of Book III, chapter 43 (125v23).20 Up to this point, U seems to be a fair copy, replacing an earlier draft; from this point onwards, however, U itself begins to look like a draft, with chapters being added in piecemeal fashion. How much of this stretch (125v–51v) was written by scribe U2, how much (if

This man, scribe U3 (Blake’s hand C), was responsible for giving the book its present shape. He wrote the whole of booklet 1C (fos. 152–78), substituting three new quires for the latter part of booklet 1B.21 He also wrote the whole

2, showing approximately the same stretch of text as the facsimile of T (above, note 10). There are no other published reproductions of scribe U1’s work, as far as I am aware. 20

21

I do not know of any reproduction showing this scribe’s work.

86

A page written by scribe U3 (175v) is reproduced by Robinson 1988,

The monks of Ely – Part I of booklet 3 (fos. 215–40), the collection of hagiographical tracts. Between these two new booklets he inserted booklet 2, the copy of xEl written some time previously by scribe U1. It was scribe U3 who connected the beginning of this booklet with the end of booklet 1C, by adding the stretch of text (177v–8v) which in T forms part of xEl but which U1 did not include. And finally it was scribe U3 who added one more document (210v–13v) in the space available at the end of booklet 2.22

1170 for scribe U1, which may err on the late side, and a date of about 1180–1200 for scribe U3, which may err on the early side. Booklet 2 is linked with the first part of booklet 1A, to the extent that both the hand and the format are the same. That does not prove conclusively that this booklet was, at the moment when scribe U1 wrote it, intended to form an appendix for HEI; it could have been scribe U3 who decided that; but probably we are safe in assuming that this had been the intention all along. More specifically, we need to note that one segment of the xEl text will have to be considered separately from the rest. This is the segment giving lists of names, for fourteen hundreds in Cambridgeshire and three in Hertfordshire, of the men who had sworn to the truth of the facts reported from their hundred.24 Because the hand is different, it can be stated as a matter of fact that this segment has a different history from the rest of xEl. How far the difference goes is a question we shall have to come back to. But it is clear straight away that conclusions drawn from this stretch of text cannot be assumed to apply to xEl as a whole.

The final section of HEI’s Book III, recopied by this scribe, presumably from U2’s draft, carries the story forward as far as 1170. The author, speaking in the first person singular, explains that he was expecting to conclude his work with the death of bishop Nigel in 1169; but now he feels compelled to add a final chapter, giving a brief account of the life and death of the holy martyr, Thomas archbishop of Canterbury. From this it is clear that HEI was completed soon after 1170, possibly, as Blake (1962, p. xlviii) suggests, during the five-year interval when the see of Ely was vacant, 1169–74. We can thus be certain that scribe U3’s contribution is later than 1170; but we cannot say how much later. We may suspect that scribe U1’s contribution – the one which interests us – is earlier than 1174, but we cannot say how much earlier. The compilation of the chronicle is likely to have extended over several years, perhaps with some changes of plan along the way. Probably the author started work while bishop Nigel was still alive, not deciding till after the bishop was dead that his death would be a fit point at which to end (changing his mind again, however, after archbishop Thomas was murdered). But it does not seem possible to say how many years he spent on his task, or when he first began.23 For our purposes, these uncertainties do not seem to matter much. I assume a date of about

V = British Library, Cotton Tiberius A. vi, fos. 36–120 Tiberius A. vi is a composite volume, consisting of three very different and quite unrelated items put together by Cotton.25 In the fire which came close to destroying the Cottonian Library in 1731, this book suffered some damage, mostly along the top edge. When it was repaired, the leaves were cut and mounted one by one. From discontinuities affecting the seventeenth-century foliation, it appears that a few of the leaves – blank leaves, one would assume – may have been discarded at the same time.

pl. 86. 22

The table of contents (107r–9v) inserted at the front of Book III, written by a similar but (in my opinion) not the same hand, reflects the shape imposed on the end of the manuscript by scribe U3. The last few entries (109vb) are as follows: Passio sanctissimi Thome martyris cantuariensis archiepiscopi. Descriptio terrarum e˛ cclesi˛e sancte æd’eld’ in uolumen protensum. Vita beate Sexburge. Vita sancte Ermenilde. Vita sancte Ærchengote uirginis. Vita sancte Werburge uirginis. Vita beate Æd’elberge uirginis. Vita alme uirginis Withburge. The first of these entries refers to the final chapter of HEI; the second covers xEl; and the rest relate to the hagiographical tracts in booklet 3. Of the six saints in question here, three belonged to Ely, and the others were sisters or daughters of Ely’s saints. Saint Audrey, Ely’s patron saint, is the subject of HEI’s Book I, and accounts of her miracles are woven into Books II–III. ((Four of the tracts found in U are among those edited, from this and other manuscripts, by Love (2004, pp. 134–88, 12–22, —, 26–50, —, 54–66); the two that she omits consist just of excerpts from Beda.))

The item of interest here is a portion of a late twelfthcentury manuscript from Ely. When the leaves were cut, nobody thought to keep any record of the original collation; as it now stands, the manuscript consists of 85 single leaves, divisible into three booklets (Table 25). Excluding some additions made by later hands (69v–70v, 118v–20r), the contents are as follows: booklet 1 (fos. 36–70): a copy of xEl (36r–69r), largely the same as in T, followed by three blank pages; booklet 2 (fos. 71–98): a copy of part of the B version of the survey text (71r–98v), defective at the end; booklet 3 (fos. 99–120): a copy of the cartulary (99r–118r),

23

Some lapse of time is implied by the apologetic remarks in the preface to Book II (ed. Blake 1962, p. 63). Blake thought it possible that the compilation of HEI might have taken more than forty years. Because HEI / U is the work of one man, and because the preface to Book I already includes a prospectus for Books II–III, I do not find that credible; many authors have had to apologize for taking longer than expected to finish a book. But it might well have taken 5–10 years to research and write a book as big as HEI.

24 It is explained in the footnotes to Hamilton’s edition of xEl that in U this segment comes before the title; but nothing is said about the hand being different, nor about a new quire starting at this point. 25

The other contents are: the ‘B’ manuscript of the ‘Anglo-Saxon Chronicle’ (fos. 1–35), described by Ker (1957, pp. 249–50) and edited by Taylor (1983); and a fourteenth-century chronicle in French (fos. 121–99), entirely unoriginal (Taylor 1957, pp. 430–2).

87

The survey of the whole of England collation

17th-century foliation

modern foliation

thirty-five

38–72

36–70

twenty-eight

76–9, 90–113

71–98

twenty-two

118–39

99–120

T. There does not appear to be any evidence which tells against this view; there are some positive facts which tend to confirm it.29 If V consisted only of these two booklets, it would have no textual value whatever. Its value is confined to the text in booklet 2, of which this copy is the only one in existence. Unlike the scribes who worked on U, scribe V1 had no authorial or editorial ambitions; he was simply making a copy. To judge from the work he did in booklets 1 and 3, he was not the most accurate of copyists, but he did not take it upon himself to make improvements in the text. As long as his copy looked good on the page, he was satisfied. We may assume that the same applies to booklet 2, the unique copy of B-Ca. Because the rubrication is all in place, we may assume, furthermore, that the scribe completed the task (however he defined it) that he had set for himself.

Table 25. Collation of British Library, Cotton Tib. A. vi, fos. 36–120. (The seventeenth-century foliation jumps from 79 to 90, but there is no significance in that: the man numbering the leaves lost count, writing 90 insted of 80.) largely the same as in T but defective at the beginning, followed by five blank pages. The manuscript is later than 1152, on the evidence of booklet 3 – appreciably later than that, to judge from the style of the script. Hamilton (1876, p. xiii) suggested a date of about 1180. I would be inclined towards a later date, perhaps closer to 1200; but the only opinion to be trusted is the opinion of someone closely acquainted with the products of the Ely scriptorium,26 and that description does not apply to me. As far as I can see, an exact date is not going to be needed for present purposes.

Through no fault of his, what we see when we look at this booklet is not what the scribe would have wanted us to see. At the upper margin and outer corner, the leaves have been eroded by fire, and the parchment is discoloured and distorted; although Hamilton managed to decipher most of the damaged passages, here and there some parts of the text have been destroyed. Before that, before it came into Cotton’s possession, booklet 2 was already defective, not just at the end, where it breaks off in mid sentence at the foot of 98v, but also between fos. 96 and 97. It can be proved (see below) that four leaves have gone missing here; so we can feel fairly sure that booklet 2, at the moment when Cotton acquired it (and presumably already at the moment when Agarde acquired it), consisted of three quires of eight and a fourth quire from which the four inner leaves had dropped out. From the end of the fourth quire onwards,30 all the rest of this booklet had been lost.

For all three booklets the format is consistent – two columns and 31 lines to the page – and the original text is all the work of one man, scribe V1, writing a bold but rather uneven script.27 The headings and coloured initials have all been supplied. It seems clear enough that all three booklets were intended to belong together; but it is doubtful whether the existing order, fixed by Cotton’s binder, was the order that the scribe had in mind. There are no clues in V itself (quire signatures, table of contents, medieval foliation), but comparison with T suggests that booklet 3, the cartulary, was meant to come first, not last (and perhaps that V originally began as T does, with a copy of the Libellus). This is speculation; but it is not an unlikely idea that Agarde or Cotton might have changed the order of the booklets, so that V began with the beginning of something, rather than in the middle.

Up to a point, it is fairly easy to estimate the quantity of missing text, with the help of DB-Ca.31 If we compare DBCa with the surviving part of V, we find, roughly speaking, that one line in DB corresponds with two lines in V – partly because V has shorter lines, partly because the B text includes some categories of information which are absent from DB. If DB has a paragraph five lines long, the matching paragraph in V will be about ten lines long.32 Thus,

With regard to booklet 3, the cartulary, it is as certain as these things can be that V was copied from T, at a time when T already contained the additions made in that manuscript by scribes T2 and T3 (Table 23). The point was proved by Blake (1962, p. xl). To repeat the most obvious fact, in V the final document breaks off unfinished, in the middle of a word, just as it does in T; but here the break comes in the middle of a line (118rb11), not at the foot of a page.28 The presumption is that V’s booklet 1 was also copied from

29 For example, the word terre, turned up at the end of a line in T (148vb17), easy for a copyist to overlook, and in fact omitted from V. 30 As was first noted by Hardy (1865, p. 36), the text breaks off in the middle of this sentence: Et vii istorum homines re(gis) e(dwardi) fuerunt, viii h(idas) et i am u(irgam) habuerunt, et vicecom(iti) re(gis) v auras . . . (98vb). The missing words can be restored as et i inuuardum reddebant: the factual information comes from the corresponding paragraph in DB, the formula from the previous paragraph in B. 31 Here and elsewhere (especially in Appendix I), I shall be using the paragraph numbers introduced by Rumble (1981), not just for DB-Ca, but also for the corresponding entries in B-Ca and xEl-Ca.

26

When, for example, did the spelling c for t in words like inquisicio start to catch on in Ely? 27

The page lithographed by Netherclift for Hamilton (above, note 2) is fo. 71r, the beginning of booklet 2. I do not know of any other reproductions.

32

In counting the lines of V, I ignore two elements of the B text which are not represented in DB: the list of jurors for each hundred, and the overall assessment for a village divided into two or more manors. With those elements excluded, the multiplier comes out as a remarkably consistent

28

A second scribe, V2, began the additions that he wanted to make on the verso (118v), leaving this document unfinished.

88

The monks of Ely – Part I which had been submitted to the Treasury, in mid 1086, by the commissioners responsible for finalizing the report for Cambridgeshire. We cannot expect to prove this: we cannot rule out the idea that V was a second-hand copy (or an nth-hand copy, n > 1) of B – in which case it would follow that not all of its errors need be scribe V1’s fault. It is not obvious why anyone would have thought it worthwhile to make a copy of B; but the thought occurred to scribe V1, and so might perhaps have occurred at some previous date to someone else. The most we can say is that there is no reason not to think that V was copied directly from the original – and in the nature of the case a double negative is as close as we can get to a positive. When we have to deal with copies, rather than originals, ambiguities of this kind are inescapable.

for the missing portions of V, if we count up the number of lines in DB and multiply by two, we ought to get a reasonably accurate estimate of the number of lines in V. The question is of no great consequence, and I do not propose to discuss it in detail, only to report the conclusions that I have come to. (1) Between fos. 96 and 97, it is, I think, quite certain that four leaves have been lost (fos. 96A–D).33 (2) After fo. 98, I calculate that the missing portion of the survey of Cambridgeshire would occupy six whole leaves (fos. 98A– F) and some small part of a seventh (fo. 98G). But here we are extrapolating into a void, without any means of knowing how far we have gone adrift. After the end of the survey of Cambridgeshire, what, if anything, came next? In xEl (see below), the excerpts from the survey of this county are followed directly by excerpts from a description of the town of Cambridge. There is a parallel passage in DB-Ca (189r); but the details were mostly dropped from DB, and only a few snippets of interest to Ely found their way into xEl. Did V contain the full text of this description of Cambridge, appended to the survey of Cambridgeshire? Whatever the answer to that may be, a larger question comes next. Did V stop after dealing with this one county, or did it continue with another? These questions are worth asking, even if there seems to be little hope of answering them. They are, indeed, worth asking for that very reason. We need to remember how much uncertainty has resulted from the loss of part of this unique manuscript.

V’s copy of B-Ca is not our only resource. DB is another proxy for B, and one which has the advantage of being intact. The entries here have been reorganized, shortened and reworded; but the factual information should all be derived from B, and (errors aside) should all agree with V. From Ely, two other sources are available, though both consist only of excerpts of limited scope. First, all the paragraphs of B in which Ely is mentioned are represented by parallel entries in xEl, some of which, because they agree word for word with V, must have been copied word for word from B (see below). Second, with one exception, all the paragraphs of B which recite the names of the jurors for a given hundred recur in a segment of text which got itself connected with xEl. If V had never been written, or if it had been lost or destroyed, it would still be possible, from this other evidence, to arrive at a vague reconstruction of B-Ca; and that is what we are reduced to doing, for those stretches of text which do not survive in V. An outline reconstruction of the complete text, based on V, xEl and DB, is presented below (Appendix I).

The fact is that we have lost more than a quarter, perhaps much more than a quarter, of the text that scribe V1 was intending to preserve for posterity. Out of courtesy, we need to make some effort to visualize V as the scribe wrote it, as he meant us to see it. But sooner or later our attention has to shift. Beyond a certain point, we lose interest in knowing what was true of V; instead we want to know what was true of V’s exemplar. What was the manuscript which scribe V1 had in front of him, and how and when did it become available in Ely?

To appreciate how much difference it makes whether V survives or not, imagine what the situation would be if the stretch of text covering the first two hundreds had been lost (Table 26). We should still have the hundred headings and the names of the jurors; we should still have entries in xEl relating to three of these villages; but for the rest we should have to try to work out how DB’s entries fitted into this framework – a problem which is, in general, only approximately soluble. (And even if we could put the entries back in place, we could not hope to retrieve the original wording.) Because V has not been lost, we know all the answers without having to work them out. Much that would be uncertain if V did not exist becomes certain because it does.

2 Though the nature of V’s exemplar is a question which cannot be answered all at once, it has to be the simplest hypothesis that scribe V1, when he made his copy of B-Ca, was copying from the original – from the actual manuscript factor, approximately 1.94. In reconstructing the missing parts of V, we shall need to top up the estimate appropriately, after performing the multiplication, so as to allow for the presence in V of these elements absent from DB.

In its formal properties, V cannot be thought to have resembled its exemplar. The layout used in this booklet – short lines arranged in two columns, coloured initials and so on – is the same as in the other two booklets; presumably the scribe would have used this format whatever the exemplar looked like. There is nothing here which invites the thought that it might have been imitated from the earlier manuscript. Though it seems a safe guess that the lines in the exemplar

33

I pose the question thus: how many lines ought to occur in V between the beginning of para. 14/49 (96vb30) and the end of para. 14/56 (97ra10)? Summing the entries in DB (including these two) which ought to have counterparts here (Appendix I), I get a total of 248 lines. Multiplied by 1.94, that gives 481 lines; topped up (as explained in the previous note), that becomes 510 lines. This is the answer to the question as it was put. Subtracting the number of surviving lines (two at the bottom of 96vb, ten at the top of 97ra) and dividing by the number of lines per leaf (4 31 = 124), I get 4.016 as the estimated number of missing leaves.

89

The survey of the whole of England B-Ca / V

B-Ca / xEl / T

CAMBRIDGESHIRE Staploe hundred jurors’ names Kennett Badlingham Chippenham Snailwell Exning Burwell Soham Fordham Isleham Wicken Cheveley hundred jurors’ names Silverley Ashley Saxon Ditton Ditton Kirtling Cheveley

CAMBRIDGESHIRE Staploe hundred

B-Ca / ? / T Staploe hundred jurors’ names

Snailwell Soham

Cheveley hundred

Cheveley hundred jurors’ names

Ditton

Table 26. Reconstructing B-Ca with or without the evidence of V. were longer than the lines in V, I do not find any error in V which might have been caused by the scribe dropping a line. In V, quite frequently, words and short phrases are misplaced – some examples of this are cited by Round (1895, pp. 15–16) – as if notes added in the margins of the original were being imported into the text by an unintelligent copyist; but I cannot see any pattern here. With the help of xEl and DB, we can make some progress in detecting and correcting V’s errors; whether we think that these errors are all to be blamed on scribe V1 depends on whether we think that he was copying directly from the original.

If V can be trusted on this point, B had no title. Scribe V1 left three lines blank, as if with the thought that somebody might like to invent a suitable heading, but nobody ever did. There is no preamble; the text begins, as it means to continue, in a brisk and businesslike fashion: In Cambridgeshire. In Staploe hundred. These are the men who swore . . . (a list of eight names). In this hundred Kennett defended itself for 3.5 hides in the time of king Edward . . .

and then it launches into its description of the manor coinciding with this village. When it comes to a village divided into two or more manors, the text takes on this shape:

For monitoring V1’s performance as a transcriber, the best evidence comes from the copy which he made of xEl, if we can assume (as surely we can) that he was copying from T. As far as I have checked, the most obvious trait is a tendency to lengthen shortened words: he often writes dominio where T has d’nio, homines where T has ho’es. He is more inclined than T to write verbs in full, numbers as words instead of numerals. Sometimes this tendency results in errors – potest for pot(uerunt), manet for man(erium), hund’ for h(idis) – which do not redound to his credit. On the whole, however, considering the length and the tediousness of the text that he was copying, his copy is a respectable piece of work. It is also worth noting that he seems to alter the spelling of place-names as a matter of course: if he recognizes the name, he spells it as he would normally spell it, not as he finds it spelt in his exemplar. Probably he did the same in making his copy of B-Ca; but of course it cannot be taken for granted that he treated every exemplar with the same degree of respect or disrespect.

In this hundred Burwell defended itself for 15 hides. Of these 15 hides the abbot of Ramsey holds 10.25 hides . . . Of these 15 hides Alan holds 2.5 hides from count Alan . . . Of these 15 hides Gaufrid holds 1.25 hides from count Alan . . . Of these 15 hides the nuns of Chatteris hold 0.5 hides . . . Of these 15 hides Hardwin de Escalers holds 0.5 hides . . .

Once Burwell’s 15 hides have all been accounted for, the bulldozer moves on to the next village – and so on, and so on, manor by manor, village by village, hundred by hundred, until finally there is not (or would not be, if the text were complete) a single acre of land in Cambridgeshire of which we do not know who holds it, and whether he holds it directly from the king or from one of the king’s barons. From reading B, much more than from reading DB, one gets a sense of the inexorable force that was driving the sur-

90

The monks of Ely – Part I V.37 Possibly a note of the omission was intended to appear each time; in the text as it survives, however, there is only one more note of this kind.38 Otherwise the only clue that we get is the fact that the assessments reported for the component manors fail to add up to the assessment reported for the village. With only three words to go on, in breui suo, we can hardly hope to understand exactly what was meant. But my guess would be that these entries were cancelled in the B text (occasionally with a note explaining why) some time after the compilation of the C text,39 and that the scribe of V, or some previous scribe (whichever scribe it was who copied directly from B), omitted them because they were cancelled, without thinking that perhaps it might be better to include them nevertheless (if they were still decipherable).

vey forwards. Straight questions were put; straight answers were expected. There was to be no tolerance for delay or indecision. In B, as in DB, the basic unit of inquiry is the manor. Most of the questions being asked are demands for information relating to a given manor; and the same questions are repeated for every manor in turn. For example, the abbot of Ramsey’s manor in Burwell is described in these terms: And of these fifteen hides the abbot of Ramsey holds ten hides and one virgate from the king. There is land there for sixteen ploughs; there are four on the domain and twelve for the villains. There are three hides in domain and forty acres. Forty-two villains and a half, eight slaves, meadow for ten ploughs, pasture for the village’s livestock, two mills (paying) six shillings and eight pence. Livestock on the domain: two spare oxen, a hundred sheep, twenty-f ive pigs, four horses. All sources of profit included, it is worth sixteen pounds; when (the abbot) got possession, (it was worth) sixteen pounds; in the time of king Edward, twenty pounds. This manor has always belonged and (still) belongs to Saint Benedict’s church.34

3 From the answers that were recorded, it ought to be possible to reconstruct the questions that were being asked. We can do that; we can do much more than that. From the B text for any single county, it ought to be possible to arrive at a reconstruction of the program which governed the entire survey (Table 27). The questionnaire at the heart of the program is (if we wish to use this expression) the commissioners’ terms of reference; but we need to tread carefully here. These are the terms of reference issued to the commissioners responsible for the survey of Cambridgeshire, as they were given effect by those commissioners. How far the instructions varied from county to county, how much room there was for differences of interpretation, or for outright misunderstanding – these are questions for which as yet we have no adequate answers.

The wording of the text, though at first it is slightly unstable,35 soon settles down into a shape which repeats itself in paragraph after paragraph. Occasionally some fact turns up for which the usual formulas will not suffice; but as soon as this fact has been disposed of, the pattern reasserts itself. In one important respect the pattern fails. As was recognized by Round, the king’s own manors – more precisely those manors which belonged to king Willelm because they had formerly belonged to king Edward – are not recorded in the B text as we have it. The next village after Burwell is Soham, assessed at 11 hides, and one of the manors here happened to belong to the king. We are given no description of it. Instead we find this note:

For each hundred as such, the only information recorded is the jurors’ names; for each village as such, the only information recorded is the total number of hides for which it has to pay geld. In subsequent versions of the survey text, this information all drops out. Once the text had been reorganized on feodal lines, there was no place in it for any data above the level of the individual manor. The information which had to be dropped from the main text could have been collected into appendices; but that did not happen. In C, in D, in DB, it is only from incidental remarks that we learn of the jurors’ existence; we never learn their names. Again, we are never told what the total assessment is for

Of these 11 hides the king has 9.45 hides in his brief,36

after which the other four holdings in Soham are described in the normal way. Seven more manors belonging to the king are similarly omitted from the B text, as it appears in

34

Et de his xv hidis tenet abbas de Ramesio x hidas et i uirgam de rege. Terra est ibi xvi carrucis, iiii or in dominio et xii uillanis, et iii hide in dominio et xl acre, xlii uillani et dimidius, viii serui, pratum x carrucis, pastura ad pecuniam uille, duo molendina de vi solidis et viii denariis, pecunia in dominio ii o animalia ociosa, c oues et xxv porci, iiii or runcini. In totis ualentiis ualet xvi libras, et quando recepit xvi libras, tempore regis Edwardi xx libras. Hoc manerium semper iacuit et iacet in ecclesia Sancti Benedicti. Based on V-72ra, but checked for the formulas against other entries in B / V and for the facts against DB (para. 7/9).

37 In DB-Ca the stretch of text corresponding to the ‘king’s brief’ is paras. 1/1–9. Except for the last one, the places in question all fall within the scope of the surviving portion of B-Ca. 38

Et de his viii h(idis) et xl ac(ris) habet rex i h(idam) et iii uir(gas) in breui suo (V-95vb, para. 1/8).

35

This is a point, noticed by Galbraith (1961, p. 126), which might be worth investigating more closely. The difficulty will be in deciding how much of the fluctuation originated in B, as the commissioners’ scribes adjusted themselves to their task, and how much of it in V, as the Ely scribe adjusted himself to his.

39

There is no indication in the surviving C booklets that the king’s manors had been given special treatment in the B text. On the contrary, the C booklets covering the king’s manors seem to have been compiled at the same time as all the rest, from the same source text, by the same scribes. So the absence of such entries from B-Ca / V should probably be taken to imply, not that they were never present, but rather that they were removed from the text at some later stage.

36

De his xi h(idis) habet rex ix h(idas) et dimi(diam) vi ac(ras) minus in breui suo (V-72va, para. 1/1).

91

The survey of the whole of England for each county for each hundred in this county who are the men who swore? for each village in this hundred how many hides did it answer for TRE? how many now? for each manor who holds it? if not from the king, from whom? how many hides? (Report the TRE number, regardless of whether the current number is the same.) how many ploughs does the land suffice for? how many ploughs on the domain? how many ploughs for the villains? how many hides of domain? (Do not answer this question unless some deduction of danegeld has been claimed.) if there are fewer ploughs than the land suffices for, how many more might be made on the domain? how many more for the villains? how many villains, bordars, slaves? anybody else worth mentioning? how many mills? and what are they worth? any other assets worth mentioning? how much meadow? how much pasture? how much livestock -- cows, sheep, pigs, horses -on the domain? altogether how much is the manor worth? how much was it worth when the man who holds it got possession? how much was it worth TRE? who held it TRE? on what terms? did the king get anything from him? if there is any current dispute, get the jurors to give you a statement of the facts. (Do not try to arbitrate; your remit is just to report.) next manor next village next hundred next county

Table 27. A reconstruction of the program governing the survey. any village. In the case of a village (like Kennett) which comprises only one manor, of course the assessment of the manor is the same as the assessment of the village; but DB does not tell us explicitly that this is the only manor that needs to be counted. In the case of a village (like Burwell) divided into several manors, DB does not give us the total at all. We can recover it (errors aside) by tracking down the corresponding entries in the relevant chapters of DB and adding up the assessments recorded there; but we have to do that for ourselves, without any guidance from DB.

with matters affecting the geld – but I think that he was missing the point. The assessment information is still to be found in DB – in an arrangement which, though certainly less convenient for some purposes, would presumably not have been adopted unless it had been thought preferable for some other purposes.40 It is the omission of the jurors’ names which carries greater significance, because it was irreparable. These names had not been recorded for no reason: they had been recorded because it might become necessary to know who was legally answerable for the truth of the statements vouched for by some given hundred. If it turned out that the facts had been misrepresented, the jurors

It is an established fact, therefore, that some of the information assembled and set down in writing by the commissioners conducting the survey was systematically omitted from C, and hence from D and DB. This is important: it means that the B text was not altogether superseded by these subsequent versions. The omission of the assessment figures was given great emphasis by Galbraith (1942) – it seemed to him to prove that the survey was not especially concerned

40 Galbraith, towards the end of his life, appears to have thought that his exposure of the ‘geld fallacy’ was the chief contribution he had made to ‘Domesday’ scholarship. That is rather sad, because the ‘geld fallacy’ is not a fallacy at all. What Galbraith insisted on finding inexplicable had in fact been explained long before, by Ballard (1906, pp. 249–50) – admittedly not the most credit-worthy source, but one which Galbraith was prepared to trust implicitly when it came to Eyton’s conjecture.

92

The monks of Ely – Part I were the men on whom retribution would fall first. Therefore it cannot be true (as has sometimes been thought) that the commissioners’ report was thrown away as soon as the D text had been brought into existence. Even after that, the B text still had some value of its own: it would retain that value, perhaps not for ever, but at least for as long as there was any likelihood of litigation resulting from the survey. For that reason it had to be kept; and that means, by the way, that the B text would still have been available, for some length of time, should anyone have wanted to consult the assessment information in its original, unfeodalized arrangement.

93

Chapter 9 The monks of Ely and the records of the survey – Part II

In dealing with xEl we have two dangers to avoid. On one side, we risk being distracted by questions which, however important they may be for the history of Ely, or for the history of Cambridgeshire, are not especially relevant to the survey. On the other side, we risk being trapped by questions which can be discussed interminably, without our making any progress towards an answer. If we want to stay clear of these dangers, we shall have to choose a course carefully – and having made our choice we shall have to travel fast.

planatory title (92r) – a famous passage, frequently quoted, not infrequently out of context.2 There are two odd things about this title, considered as a title for xEl: it seems to be speaking of a single county, whereas xEl speaks of six, and it seems to be speaking of an entire county, whereas xEl speaks only of those manors in which Ely had an interest. This title does not mention Ely, or Saint Audrey, or the abbot and monks.3 It would in fact be much more suitable as a title for B-Ca; but we have no evidence for making that link, or at least not for making it directly. At all events, we have reason to doubt whether this title is properly connected with xEl.

I start by discussing two technical matters. First, I consider the anatomy of the text, as it appears in the earliest surviving copy. Second, I consider the relationships between this and the other two copies. After that, I propose to discuss some aspects of the evidence, selected because they are significant, and because it is possible to draw conclusions about them with a fair degree of confidence.

(ii) The lists of jurors. In T and V, the title is followed by a stretch of text reciting the names of the jurors (eight for each ordinary hundred, twice as many for a double hundred) for fourteen Cambridgeshire hundreds and three Hertfordshire hundreds (92r–4v). In U this stretch was originally omitted; then it was supplied, by a different scribe, at the end of the preceding quire. What we take this to mean will depend on how we construe the relationship between T and U (see below), but there must have been some room for doubt, in the mind of at least one Ely monk, whether this segment ought to be included or not. Nevertheless, it must certainly derive, as xEl-Ca does, from B-Ca, that being the only version of the survey text in which the jurors’ names were to be found. At some stage (not before the 1080s, not after the 1130s), the Ely monks must have got their hands on a portion of the B text which covered two of the counties of interest to them, but (so it seems) not the other four. Scanning through this, somebody extracted the lists of jurors’ names, stringing them together to make this stretch of text.4

1 The contents of the text, as it appears in T, divide into 17 segments (Table 28). Additional segments appear in U and V. It is not impossible that some of these are authentic parts of xEl accidentally missing from T; but there is (unless someone can prove otherwise) no justification for regarding them as such. I attach no significance to them, except as indications that xEl was thought important – important enough for additions like these to become attracted to it. Some portions of the text as it appears in T are fairly sure to be accretions of the same kind, dating from between the 1080s and the 1130s.1 The most obviously incongruous segment is the list of parish churches in Norfolk (122r), which has no ostensible connection with the survey: presumably there was some space here in the original, and somebody decided that the space might as well be made use of for inserting a copy of this list. Six other segments, it seems to me, are not, or cannot safely be assumed to be, parts of the original text, and I discuss these briefly, one by one, before dealing with the core components.

(iii) A statistical digest covering the manors held by the abbey in domain, reporting five numbers for each: domain 2

The passage begins: HIC SUBSCRIBITUR INQUISITIO TERRARUM, quomodo barones regis inquirunt . . . (T-92r). At the very least, one should insist on having this passage quoted from T, not from V. 3

By contrast, the table of contents inserted into U (above, p. 87) refers to xEl as a Descriptio terrarum e˛ cclesi˛e sancte æd’eld’ in uolumen protensum. (I do not understand what was meant by the last three words, ‘into a stretched-out volume’.)

(i) The title. In all three copies, xEl begins with a long ex-

4

From his treatment of Hertfordshire, it appears that this scribe’s intention was to copy the lists only for those hundreds which were of interest to Ely. In Cambridgeshire, if that was his plan, he had one hundred to omit – and he did indeed omit one hundred, but not the right one. Instead of omitting Whittlesford, he omitted the hundred before it. By accident or on purpose, he also changed the order from that existing in B-Ca, making two transpositions.

1

Three segments, Ca, Ht and Nk-1, end with a note reporting the total value: De toto quod habemus in tota scira . . . (110v, 114r, 122r). (For Ca and Ht the total is the same as in segment xEl-s; for Nk it is a slightly smaller amount.) The first person plural – ‘we’ meaning the monks – proves these notes to be additions made in Ely; but this is the only context where it occurs.

94

The monks of Ely – Part II xEl / T 92r 92r–4v 94v–110v 110v–13r 113r–14r 114r–15r 115r–17r 117r–22r 122r 122r–5r 125r–33r 133r–42r 142r–3r 143v–5v 146r–v 146v–9r 149r–v

segment title jurors (CaHt) xEl-Ca xEl-s (CaHtExNkSkHu) xEl-Ht xEl-Ex-1 xEl-Ex-2 xEl-Nk-1 parish churches (Nk) xEl-Nk-2 xEl-Sk-2 xEl-Sk-1 xEl-Hu domain manors villains’ ploughs tenancies Hamo de Sancto Claro

Hamilton 1876

(v) A schedule of the abbey’s non-domain lands, categorized as ‘thegnland’ and ‘soke’, organized tenant by tenant and county by county (146v–9r).

97 97–101 101–21 121–4 124–5 125–7 127–30 130–6 136–7 137–41 141–53 153–66 166–7 168–73 174–5 175–82 182–3

(vi) A schedule of the Ely lands held by Hamo de Sancto Claro (149r–v). In T there is a change of format at the start of this section, and perhaps a change of appearance in the script. Apart from the heading, it consists entirely of excerpts from xEl-Sk-2 relating to the lands which in 1086 were held by the bishop of Bayeux. But the heading proves that it was not compiled till much later, in or after the 1120s.9 The end of this schedule is the end of the text, as it appears in T; four blank pages follow. In each of the other two copies, U and V, there is more to follow. In U two fairly long documents occur at this point.10 Scribe U1, seemingly without hesitation, continues with a document listing the barons who are still in possession of some of Ely’s lands – lands which they ought to hold from the church or else not hold at all (207v–9v, ed. Hamilton 1876, pp. 184–9). Scribe U3, leaving one page blank, adds a document listing the lands to which the monks (we are told) succeeded in proving their claim, at a trial held by bishop Goisfrid and others, but of which they have still not been able to get possession (210v–13v, ed. Hamilton 1876, pp. 192–5, ed. Bates 1998, pp. 413–17).11 Two shorter documents occur in V: a list of the monks’ fisheries (69ra–b, ed. Hamilton 1876, pp. 190–1), loosely related to a list that was added in T (156v), and a memorandum explaining that a cartload of lead from the Peak amounts to only four-fifths of a London cartload (69rb, p. 191) – a fact worth knowing, no doubt, if one had a church to roof, but not relevant for us.

Table 28. Contents of xEl as represented in T.

ploughs, men’s ploughs, villains, bordars, slaves (143v– 5v). The factual information is nearly all derived from xEl, but – as the title explains (this title does mention Saint Audrey) – the entries have been rearranged to bring together all the manors administered by the same reeve (which means that they have been rearranged very drastically).5 Subtotals are entered at the end of each reeveship, if it comprised more than one manor, and that supplies a check on the numerical data. This and the next three segments – not thought worth printing by Ellis (1816) – are all derived from xEl, but the presumption is that they are independent from T and the other copies of the complete text; so they may have some textual value, each within its own narrow compass.6 It also has to be remembered, however, that they may have been edited, more or less extensively, by scribes drawing from other sources available in Ely.7

2 Up to a point, it is easy to work out the relationships between these three surviving copies of xEl. T is the earliest copy, and – at least in a superficial sense – very obviously the best one. By failing to base their editions on T, Ellis and Hamilton went wrong before they started; the next editor of xEl, one hopes, will not make the same mistake. For the other two copies, the basic pattern is this: U very frequently differs from T and V; V very frequently differs from T and U; but it never happens that U and V agree in differing significantly from T. In this passage, for example:

(iv) Another statistical digest, reporting for each manor a single figure, the number of villains’ ploughs (146r–v).8 5

The title is perfectly explicit about this: et h˛ec distinguuntur sicut prepositi tenent quisque preposituram suam (T-143v). I do not know why Finn (1960, p. 397) made a mystery of it. 6

There is a good example of this in the paragraph relating to Streetly, where xEl reads ii c’ & dm’ h’ in d’nio (96r). From B / V, confirmed by DB (para. 5/15), it is clear that this ought to read ii c’ & dm’ & dm’ h’ in d’nio, ‘two ploughs and a half and a half hide in domain’. The digest of xEl has the right number of ploughs, ii c’ et dm’ (143v). This tends to prove that the omission of & dm’ was an error originating in T, and that the other two copies, because they share the error, both derive from T.

T: T’c & post: ual’ xx lib’, 7 modo xxx lib’ (115r) 9 Hamo de Sancto Claro was given custody of the lands which reverted to the king after the death of Eudo Dapifer in 1119–20 (Farrer 1925, p. 168). Presumably this schedule was drawn up, then or soon afterwards, because there was some hope of persuading the king to give back the lands which had been stolen from Saint Audrey.

7

Presumably that is where we look for an explanation of the disconcerting fact that this digest reports the presence of slaves on the manors in Huntingdonshire. Neither DB nor xEl mentions the existence of slaves; but somebody knew that they were there, and how many of them there were.

10

They were discussed – how helpfully the reader must decide – by Finn (1960, pp. 398–407).

8

11 The trial took place before 1075; this text, which does not pretend to be a contemporary record, is earlier than 1086.

In T it carries the title UILLANORUM. The scribe of V, not seeing the sense of this, changed it to Nomina uillarum.

95

The survey of the whole of England U: t’c ual’ xx lib’ 7 modo xxx ta (190r) V: Tunc & post: ual’ uiginti li (48vb)

– a phrase which occurs quite frequently – with preter written in such a way (p’t’) that a copyist might mistake it for pot’ (to be read, one assumes, as potest).14 That makes nonsense; and the U scribe, refusing to copy nonsense, substituted the word sine, which carries the right meaning, but is actually not the right word.

there are two omissions in U, one different omission in V. For reasons already explained (above, p. 88), we can take it as certain that V was copied from T: it shares T’s errors (as far as these can be detected), and has numerous errors of its own.12 V’s readings are of no textual value, and of interest only as a means of assessing this scribe’s reliability. The sole question remaining to be answered, then, is the relationship between T and U.

It is not difficult to find evidence which suggests, seems almost to prove, that U was not copied from T. In my experience, however, this evidence always turns out to be inconclusive, when it is looked at more closely. I cite just one example. In the paragraph relating to Wisbech, as it appears in T, the value clause reads: Inter totum ual’ c s, quand’ rec’: c s, t r e: lib’ (108v). The numeral which ought to come before lib’ is missing.15 In T this omission coincides with the end of a line; and that suggests that the error originated here, through a momentary lapse of concentration on the part of scribe T1. In that case we would expect to find the same error in any copy derived from T, but not in any other copy. If we look at V, what we find is a gap: t r eaduuardi
 lib’ (45ra).16 Looking at U, at first sight we seem to find proof that this copy does not derive from T: t r e vi lib’ (186v). But there is something odd about the spacing here. On closer inspection it appears that scribe U1 (like scribe V1) initially left a gap here, and then (unlike V1) inserted the numeral afterwards, not using up all the space that he had left for it. Instead of contradicting it, therefore, this evidence tends rather to confirm the view that U was copied from T, with the qualification that some of the errors inherited from T were then successfully corrected.17

This is one of the places where we risk being sucked into some interminable discussion. Round claimed to have proved that U is independent from T, and subsequent commentators have assumed this to be true; but the proof dissolves on inspection. Despite the emphatic language – ‘A careful analyis . . . has satisfied me beyond question . . . ’ (Round 1895, p. 124) – it has to be remembered that Round was relying on Hamilton’s apparatus, which is far from perfectly accurate. Only one passage was cited specifically by Round (1895, p. 131) as proof that U could not have been copied from T, because T has ‘blunders’ from which U is immune. Here is that passage, as it appears in the manuscripts: T: IN lolesuuorda: fuit quidam sochem’, sub abb’e eli i h’ & dm’: tenuit t r ead’, potuit dare pot’ licent’ eius sine soch’m’; & modo pi uicecomes, tenet eam sub abb’e eli. Valet x s (103v) U: IN losewrd’a fuit quidam soch’ sub abb’e ely i h’ 7 dim’, ten’ t r e, pot’ dare sine lic’ eius sine soch’, 7 modo pic’ uicec’ ten’ eam sub abb’e ely, ual’ x s (184r)

The crucial point, I think, is this. Unlike scribe V1, the scribes who worked on manuscript U were not just making copies. They were participating – they were conscious of participating – in the creation of a work of literature. Of course they would not copy nonsense: if the source text needed to be knocked into shape, they were willing to treat it roughly. The question which we have been discussing, the relationship between xEl / T and xEl / U, should thus be subsumed into a larger question, the relationship between T and HEI / U; and that is an Ely question, not one for us. By reframing the question, I release myself from the obligation to answer it; but it seems to me that the author of HEI was working directly from T – from booklet 1 for the Libellus, from booklet 2 for the documents that he quotes,

There are indeed two errors in T, but it is silly to call them ‘blunders’. They are mere slips – small mistakes, obvious to any reader, and easily put right by any copyist who has his wits about him (if he thinks himself at liberty to alter the text).13 The readings found in U do indeed make better sense, but that does not prove anything at once. They could be original readings, preserved by U but garbled by T, as Round supposed them to be; alternatively they could be corrections made by an editorially minded scribe, copying, but not copying thoughtlessly, from T. This is not just to say that the evidence is ambiguous; on the contrary, as far as the first of T’s ‘blunders’ is concerned, it is fairly certain that Round’s interpretation is the wrong one. At this point, without much doubt, the original reading would have been preter licentiam eius, ‘without his (the abbot’s) permission’

14 The error would have originated either in T’s exemplar or in T itself. No manuscript has the reading quoted by Round, ‘sine licentiam [sic]’. Scribe U1 wrote sine lic’; to all appearances he knew just as well as Round that sine takes the ablative. 15 By consulting DB, we can find out what this number should be: TRE: vi lib’ (DB-Ca-192ra).

12

One variant cited by Hamilton (1876, p. 137, note 23) would, if it were correctly reported, tend to disprove this conclusion. But in fact the words modo xl sol’ are omitted only by U, not by both T and U. The V scribe, however, did occasionally allow himself to correct an obvious error. For instance, he corrected T’s Inter ual’ x lib’ (109r) to Inter totum ualet x li (45rb), by analogy with other entries.

16

The gap means: ‘Do not blame me for this. The error is my exemplar’s fault, not mine. Feel free to supply the numeral if you can.’ 17

If scribe U1 had to go looking for the missing numeral, where might he have found it? There are two or three possible answers. He might have found it in T’s exemplar, if that manuscript still existed. He might have found it in the manuscript which would later serve as the exemplar for V’s booklet 2. Or he might have found it – just as we did – in DB-Ca’s chapter 5, if Ely had procured a copy of that.

13 Round also mentions a ‘blunder’ in V. This is the omission of tenuit, again the sort of error which could easily be corrected by a copyist of moderate intelligence.

96

The monks of Ely – Part II xEl / T

from booklet 3 for xEl. As far as xEl is concerned, I find nothing which could not be accounted for on the theory that U is a fair copy of an edited copy of T. In fact, the editing continued in U itself, where there are numerous blanks or erasures and marginal additions, and this manuscript is going to pose some serious problems for xEl’s next editor.18

94v–110v 113r–14r 114r–15r 115r–17r 117r–22r 122r–5r 133r–42r 125r–33r 142r–3r 110v–13r

3 For the time being, it seems safe to proceed on the assumption that T is the only complete copy of xEl which has any textual value.19 The excerpts listed above were presumably derived from T’s exemplar (which, in the absence of evidence to the contrary, is presumably also the original), not from T itself; so these can be used, for what they are worth, as independent evidence. More remotely, T can be compared with the official versions of the survey text: against B for most of Cambridgeshire; against D for Essex, Norfolk and Suffolk; against DB for the whole of Cambridgeshire, and for Hertfordshire and Huntingdonshire. That gives us a lot of territory to cover, and I shall be driving through it as fast as I think safe.

segment xEl-Ca xEl-Ht xEl-Ex-1 xEl-Ex-2 xEl-Nk-1 xEl-Nk-2 xEl-Sk-1 xEl-Sk-2 xEl-Hu xEl-s (CaHtExNkSkHu)

Hamilton 1876 101–21 124–5 125–7 127–30 130–6 137–41 153–66 141–53 166–7 121–4

Table 29. Core segments of xEl as represented in T. there is only one segment, and all the manors in which Ely had any interest are listed here, in strictly cadastral order, regardless of their current status. In some sense, therefore, perhaps in a significant sense, xEl-Ca is exceptional. Many paragraphs of xEl-Ca are reproduced, in virtually the same order,20 word for word or nearly so, from B-Ca. One example will be proof enough. I quote B’s description of Willingham (para. 5/39), as it appears in V, and xEl’s description, as it appears in T:

To map out the route ahead, I list the core segments of xEl again, in a slightly different order (Table 29). Where there are two segments relating to the same county, one segment covers the manors of which Ely is safely in possession, and the other segment covers the manors of which Ely has lost – but still hopes to regain – possession. It seems clear that the pair of segments for Suffolk ought to be in the order shown here, like the matching pairs of segments for Essex and Norfolk, though in T the order is reversed. As for segment s, there is no reason for it to come between Ca and Ht, as it does in T: the logical place for it would be at the end (or else at the beginning). Wherever we put it, it differs in character from the other segments; it is also, in my judgment, the most valuable segment, and I keep it for last for that reason (below, pp. 103–6). The fact that the Sk segments have become transposed in T is a hint that xEl may originally have consisted of a batch of booklets, the order of which was liable to be perturbed. Nevertheless, the order of the counties, as they appear in T, is not haphazard: in segment s, the paragraphs which summarize the abbot’s holdings run through the counties in precisely this order.

B / V-97vb: IN hoc hundr’ Wiuelingeham pro vii h’ et dimid’ se de t r e et modo: pro v. Et de his vii h’ et dim’ tenet abb’ de ely vii h’, vii carrucis e’ ibi t’, ii e c’ et iiii h’ in dominio, v carr’ uillanis, xii uillani, viii cotarii, i seruus. Pratum vii carr’. Pastura ad pecuniam uille. De maris vi sol’, iiii xx ou’ ii xx por i us runc’. In totis ualentiis ual’ c sol’ et qn’ recep’ totid’ t r e viii li. Hoc manerium iacet et iacuit semper in eccl’ia S’ Ædel de ely in dominio. Et de his vii h’ et dimid’ tenet i sochemannus de comite alano i uirgam . . . Et de his vii h’ et dimid’ Rogerus tenet de picoto uicecomite i uirgam . . . xEl / T-103r: Wiuelincgaham pro vii h’ & dm’: se defendit tpr’ R ead: & modo pro v h’; In hac uilla ten7 abb’ eli: vii h’; vii c’ ibi e’ t’ra; ii c’ & iiii h’ in dominio; v c’ hom’; xii (uill’i); viii (cot’); i (s); pratum ad vii c’; pastura ad pecc’ uille; de marisca: vi s, quat’ xx o; xxii p & i runc’. In totis ualentiis ual7 c s, quand’ rec’ c s, tpr’ R ead viii lib’; Hoc man’ iacet & iacuit in e˛ ccl’a S’ Æld’ in d’nio. Not only does xEl agree verbatim with B: it even begins by giving the total assessment for this village, an item of information which was only to be found in the B text. But then it simplifies things, replacing ‘And of these seven hides and a half’ with ‘In this village’, and focuses on the abbot’s seven hides, ignoring the other two virgates. What could be clearer? The paragraph in xEl is an edited excerpt from BCa, put into this form by someone preoccupied by matters affecting Ely.21

Because Ca is the only county where we can compare xEl with B, we are bound to start with this segment; but a word of caution is needed before we can begin. By analogy with Ex, Nk and Sk (in Ht and Hu the question did not arise), we would expect to find two segments for Ca as well. In fact 18

As I understand it, what U represents is an attempt to construct an improved version of the xEl text, suitable for inclusion as an appendix to HEI. The attempt did not quite succeed. Scribe U3, we may think, when he recopied the end of HEI / U, ought also to have recopied xEl, but he did not bother to do so. The scribe who recopied HEI in its definitive form, HEI / W, did not include xEl at all.

20

The only difference in the order is that one whole hundred has moved: Radfield comes after Chilford, not, as in B, after Staine.

19

Any statement about xEl which begins ‘In all three manuscripts . . . ’ is immediately to be distrusted – and so is the author who made it.

21

97

Round (1895, p. 9) quoted a similar pair of paragraphs, the pair relating

The survey of the whole of England Though the order of the paragraphs is consistently the same, the degree of verbal resemblance varies greatly. We have only to look at the previous paragraph (para. 26/48) to find a much looser relationship between xEl and B:

– which incorporate additional information and are much more narrowly concentrated on Ely’s own affairs. This second layer of text, I suppose, is the product of some further investigation, a partial, private survey commissioned by abbot Simeon; but this is an Ely question, and I do not propose to pursue it. From our point of view, this revision of the text has greatly reduced its value. With caution, we can use xEl to reconstruct the missing portions of B-Ca, and more vaguely to form an idea of the B text for other counties.23 But for every paragraph the question has to be asked: is this something close to a word-for-word excerpt from the official survey text, or is it the product of some subsequent revision in Ely? To the extent that the B text survives, xEl is of no interest except for correcting errors introduced by V. For the rest, xEl is equivocal.

B / V-97rb-va: IN hoc hundr’ Oura pro xv h’ se de t r e . . . Et de his xv h’ tenet Rad’ de hardeuuino ii h’ et i uirgam . . . Hanc t’ram tenuerunt x sochemanni t r e. Et i us istorum: homo Abb’is de Ely fuit. Dimidiam h’ habuit, non potuit dare neque uen. Et ii istorum homines predicti abb’is iii uirgas habuerunt. Vendere potuerunt, soca remansit abb’i. Et vii alii homines abb’is de Rames’ fuerunt, i hidam habuerunt, uendere uel dare potuerunt: sine soca. xEl / T-103r: IN ouro fuit quidam soch’m’ nomine Stanhardus qui dm’ h’ habuit sub abb’e eli, non potuit ire ab eo, nec separare ab e˛ ccl’a; & valet xx s & modo h’t harduinus. Et alii ii o soch’m’: iii v’ habuer’ potuer’ dare l’ uendere sine soc’ cui uoluer’, & modo ten7 harduinus, 7 valet xv s.

With the other five counties, we are on much weaker ground. In the absence of a straightforward copy of B, we have only two terms of comparison, xEl and either D or DB. The textual evidence, in these circumstances, is irredeemably ambiguous, just as it is for that portion of Ca where V is not available. It was Round’s suggestion, with regard to Ex, Nk and Sk, that xEl was copied from the surviving D booklets for these counties: on this view, any facts reported in xEl which are not to be found in D should be explained away as interpolations made by Ely scribes. Whether we think this likely or not, I do not know that we have any sure way of disproving it. One piece of evidence, frequently cited as proof that xEl-Nk-1 could not have been copied from D-Nk, does not prove any such thing. It is instead a good illustration of the ambiguity intrinsic to the textual evidence, and I discuss it, briefly, for that reason.

There is information in xEl which is not to be found in B, and even to the extent that the substance is the same the wording is divergent. It is possible, of course, that an Ely monk, reading the original, might occasionally have added a note in the margin reporting some relevant fact which he happened to be aware of; conversely it is possible that some phrases might have been omitted accidentally by scribe V1, while he was making his copy of B-Ca. But neither explanation is adequate for what has happened here. We are not dealing with casual additions in xEl or casual omissions in V. The whole entry has been reworded. What in V is a regular part of the B text is replaced in xEl by a paraphrase – a new version of the entry which agrees with B only loosely, and only up to a point.22

As was first pointed out by Johnson (1906, p. 4), there is a paragraph in xEl-Nk-1 – the final paragraph, relating to the manor of Bergh (T-121v–2r) – which has no counterpart in D-Nk (apart from one portion of it, which in D forms a separate entry). Except that it comes at the end of its segment, this paragraph has no odd features: it seems to be an integral part of the text. At first sight, therefore, this seems to prove that xEl was derived from a version of the survey text earlier than D – a version from which this paragraph had not yet gone missing. The conclusion is probably right (see below), but this evidence does not prove it. If one looks at D, at the place where this entry would be expected to occur (DNk-214v19), one discovers two odd things. First, there is a hundred heading – Heinesteda hund’ dim’ – which would be right for Bergh, but is wrong for the paragraph (relating to Pulham in Earsham half-hundred) which actually follows this heading. Second, there is a caret mark at precisely this point – between the hundred heading and the place-name Pullaham. It seems clear that somebody noticed the omission; and if he was capable of detecting it, presumably he would also have been capable of correcting it. (The miss-

We are, I think, obliged to conclude that xEl-Ca as we have it is a palimpsest. The underlying text consists of the paragraphs – like the one for Willingham – which, in their wording as well as in their order, are manifest excerpts from B. But many of the entries have been overwritten with new paragraphs – like the one for the abbey’s sokemen in Over

to Melbourn (para. 5/34). After noting some of the differences in wording, he added a careless remark which has done no little harm: ‘These prove that verbal accuracy was not aimed at by the transcribers’ (1895, p. 10). They do not prove that. The statement is true for at least one transcriber; it need not be true for both. These variants do not tell against the view that V is what it seems to be, a straightforward copy of B, and that xEl is also what it seems to be, a concatenation of edited extracts from B. There is no symmetry here: one text is vastly more reliable than the other. Another inept remark of Round’s – ‘the Domesday scribes appear to have revelled in the use of synonym and paraphrase’ (1895, p. 26) – has also been frequently quoted. Why is it misbegotten epigrams like these that historians tend to repeat? 22 Round (1895, p. 18) has a good example – an entry in B / V-84va which is almost identical with a pair of entries in DB (191rb, 198rb) but does not make sense as it stands. This is replaced in xEl / T-40ra by a longer sentence which explains the facts in more detail (and which makes it possible to understand how the error in the B text came about). The compiler of xEl, Round thought, appeared to have ‘corrected the original return from his own knowledge of the facts’ (1895, p. 19); but where did this knowledge come from? Not out of the compiler’s head, we may be sure.

23 That some revision has occurred in the other counties seems clear from examples like this: a sentence which in D reads Rog’ bigot tenet de abb’e, sed prius tenuit de rege (D-Nk-214v) has expanded in xEl into Hos ten’ R. bigot de rege, set abb’ diratiocinauit eos coram ep’o constantiensi, modo tenet eos predictus R. bigot de abbate (xEl / T-121r).

98

The monks of Ely – Part II a1 c1 c2 c3 e1 f1 l1 n1 p1 r1 s1 s2 t1 w1 w2

among the manors of which she had possession, it does not need to be proved that the source text was organized cadastrally; that is obvious at once. The question is whether it was organized in the same way as B; and the simplest way of framing that question is to ask whether the hundreds were arranged in the same order.25 By the time that they appear in D and DB, the hundreds have been chopped up into pieces, and the pieces have been distributed among the chapters where they belong. But in Cambridgeshire, as in many other counties, the original order is partially and approximately preserved within each separate chapter. It has long been understood, therefore, that by recombining the information available in different chapters of D or DB one can hope to reconstruct the order of the hundreds, as they appeared in B. This, by and large, is more easily said than done: the evidence is often inadequate and inconsistent.26 Because of the complexity of the compilation process, perfect orderliness is hardly to be expected; at any stage, for any number of possible reasons, blocks of text may have got themselves rearranged. With luck, however, the message that we want to hear may be audible through the noise.

Armingford Chesterton Cheveley Chilford Ely Flendish Longstow Northstow Papworth Radfield Staine Staploe Thriplow Wetherley Whittlesford

Table 30. Cambridgeshire hundreds. ing text – not just the omitted paragraph but also the omitted hundred heading – could have been added at any stage, as long as B-Nk or C-Nk was still available.) This hypothetical corrector did not write the missing passage in the margin; if he had done that, it would still be there. But he might perhaps have written it on an inserted slip; and the slip might have dropped out again, at some later date. If this evidence stood alone, it would be perfectly possible to argue, on Round’s behalf, that xEl was copied from D during the span of time when this paragraph was to be found there – after the defect had been patched, before the patch fell away.24

As it happens, Cambridgeshire is not one of the easiest counties to deal with. There are fifteen hundreds here (Table 30), and the information available from DB is not good enough for the sequence to be fully reconstructed.27 The conclusions that I come to are these. (i) There are four hundreds (c2, r1, s1, s2) which seem to belong together but cannot be ordered satisfactorily with respect to one another. Usually they come at the front of the chapter, but in one case (chapter 14) they come at the back. (ii) With these four excluded, the remaining hundreds are reasonably wellbehaved, and the order into which they fall is this:

4 Rather than wasting our time with evidence which is bound to be inconclusive, we need to find some alternative line of attack. For Cambridgeshire, because of the existence of V, it can be proved that xEl-Ca and DB-Ca both derive from BCa. Suppose that V did not exist. Would it still be possible to prove that xEl’s source text for this county was the same cadastrally organized version of the survey text from which DB also derives? Can we find some argument which works successfully in Cambridgeshire – i.e. which produces the answer which we know to be the right one for this county – and then apply the same argument elsewhere? In this section I shall be going over the same ground that was covered by Sawyer (1955, pp. 186-90); but I have worked out all the evidence again for myself. (I should also like to think that I have explained things more clearly than Sawyer did.)

DB-Ca:

f1 c3 w2 t1 a1 w1 l1 p1 n1 c1 e1

With the same four hundreds excluded, the order that exists in xEl is: xEl-Ca:

f1 c3 t1 a1 w1 l1 p1 n1 c1 e1

Apart from the omission of the single hundred (w2) in which Ely had no interest, the two sequences are identical; and that is enough to prove that the source text for xEl-Ca 25 In some other county it might happen that we could prove the weak conclusion (that the source text was organized cadastrally) without being able to prove the strong one (that it was organized in the same way as B). In fact this does not happen; so I shall be arguing directly for the strong conclusion.

Because xEl-Ca consists of a single segment, with the manors on which Saint Audrey had some claim interspersed

26 I do not have any algorithm for retrieving the latent order from a collection of partial sequences which are only roughly consistent with one another. An interactive program would be needed – one which identifies the most troublesome chapter, or the most troublesome hundred, asks for permission to delete it, and then tries again.

24 There is more to be said on the subject than this, but the rest is Ely business. The manor of Bergh, held by Godric Dapifer ‘under the abbot’, is only one half of the story: the other half concerns the manor of Apton. In xEl-Nk-2, we are not told who currently has possession of Apton; instead we are told that ‘Saint Audrey is supposed to be getting this land in exchange for Bergh’ (T-125r). In D-Nk, Apton is listed among Godric’s manors, without any mention of Saint Audrey (203r–v). Ambiguity strikes again: which is the original and which is the edited version?

27 I do not know whether anyone else has attempted to reconstruct the hundred order for Cambridgeshire from the evidence of DB alone. Sawyer did not try this: he took the order as given, on the evidence of B (patched up at the end from xEl), and then showed that this order is approximately preserved within chapters of DB (Sawyer 1955, p. 180). Hart (1974) did the same.

99

The survey of the whole of England Witham Dunmow Uttlesford Chelmsford Rochford

must have been B-Ca. (As we happen to know from V, the actual ordering in B was this: s2 c2 s1 r1 f1 c3 w2 t1 a1 w1 l1 p1 n1 ....

w4 d1 u1 c2 r1

and the partial sequence recovered from DB lines up exactly with that.)







We thus have a method, demonstrably successful in Cambridgeshire, which is capable of proving that xEl derives from B. But this method will only work under certain conditions. It will not work on a segment of xEl which corresponds approximately with a single chapter of D or DB. In such a segment, the order of the entries, and therefore the order of the hundreds, will be roughly the same regardless of whether the source text was cadastrally or feodally organized. Whether xEl’s compiler had to work through B for himself, finding the paragraphs that he wanted, or whether he was able to profit from work already done by government scribes, there will be no difference in the outcome that we can recognize. Therefore we can only hope for the method to work on a segment of xEl which corresponds with multiple chapters of D or DB; and immediately that means that we are restricted to three segments, Ex-2, Nk-2 and Sk-2. These are (by chance) the same three counties for which D survives and DB does not. I discuss these segments in turn, in this order, which happens to be the most convenient one.





















Ex-2 











Figure 9. Order of the entries in xEl-Ex-2 mapped onto the order of hundreds in B-Ex. is the result. The ordering of the entries, neither cadastral nor feodal, was presumably imposed by the man who made these excerpts: apparently he had some preconceived idea of the order into which they should be put.28 However that may be, the method which we are using fails to prove that he was copying from B-Ex. Segment Nk-2. There were 33 hundreds in Norfolk, and the order as it appeared in B-Nk can be reconstructed rather easily from D-Nk. Johnson (1906, p. 4) worked it out with almost complete success, making only one small misjudgment.29 There were fourteen hundreds in which Ely had some interest, but only ten of these are represented by entries in xEl-Nk-2. The order of these entries correlates perfectly with the hundred order that existed in B-Nk (Fig. 10), except for two small discrepancies. One of the entries for Clackclose hundred (c1) has gone astray, tucking itself in at the end of the subsequent hundred; and the single entry for Henstead hundred (h2) has dropped back to the end of the hundred which ought to follow it. A few hiccups of this kind are only to be expected; they do not prevent us from seeing the overall picture. For Nk-2, it is clear, our method works very well: we can take it as a proven fact that this segment was copied from B-Nk.

Segment Ex-2. The hundred order as it appeared in B-Ex is approximately reconstructable from D-Ex, and the sequence worked out by Round (1903, pp. 409–10) is mostly clear enough. There are some puzzling features, and several possible explanations for them; but these are Essex questions and need not detain us here. Only five hundreds are represented in Ex-2, and for these the order is reasonably certain, except that Uttlesford (u1) and Chelmsford (c2) should possibly be transposed. Either way, the result we get is negative (Fig. 9). There is no correlation between the order of the entries, as they appear in xEl, and the order of the hundreds, as they appeared in B. This is not to say that the order has been feodalized: that is not true either. Two entries relate to manors owned by Eudo Dapifer, but they have not been put together; two entries relate to manors owned by Goisfrid de Manneville, but they have not been put together either.

The same correlation, somewhat disturbed towards the end,30 exists in segment Nk-1. There it proves nothing; but I see no reason why we should not assume that what is demonstrably true for Nk-2 was true for Nk-1 as well. Both segments, I take it, were compiled by an Ely scribe who worked his way systematically through B-Nk, excerpting all the entries of interest to him and Saint Audrey. By the time that he was finished, he had made two booklets:

This segment of xEl is exceptional in two respects. In all the other segments, not excluding Ex-1, hundred headings are regularly present, barring some sporadic errors and omissions. In this segment the headings are consistently omitted, though in T the scribe did vaguely attempt to leave space for them. (The V scribe did the same thing, imitating his exemplar; possibly the T scribe too was imitating his exemplar.) More strikingly, this segment is unique in having its own title: Has terras calumpniantur [sic] abbas de Eli secundum breues regis (T-115r). It looks as if Ex-2 may have been compiled separately, at an earlier stage than the rest, and then subsumed into the larger project of which xEl

28 For example, perhaps he had been given a list of names of the places of interest to Ely and instructed to fi nd the relevant entries in the survey text. 29 Johnson’s ordering transposes Greenhoe South (g3) with Grimshoe (g4). The evidence is inconsistent, but chapters 21 and 31 outweigh chapter 22. There is a misprint in the sequence reported by Sawyer (1955, p. 187): it ought to read either 1, 6, 1, 5, . . . (if he was following Johnson) or 1, 5, 1, 6, . . . (if he was making the same adjustment as me). 30

The last entry, out of order, is the entry relating to Bergh (above, page 98).

100

The monks of Ely – Part II

c1 f4 g4 g3 s1 g5 l1 m1 b2 h2 e1 d2 l2 d1

Clackclose Freebridge Grimshoe Greenhoe South Shropham Guiltcross Launditch Mitford Brothercross Henstead Earsham Diss Loddon Depwade

seeming sometimes to come near the middle of the order, sometimes right at the end. Ely was interested in 19 of the Suffolk hundreds, and we can ignore the other six, which luckily include the two most troublesome cases. As for h1, the Ely chapter in D-Sk (chapter 21) puts it in the middle of the sequence, between p2 and r1, and no harm will be done if we follow that lead.33 Now, when we come to compare the order of the entries in segment Sk-2 with the hundred order reconstructed for BSk, we find a surprising pattern (Fig. 11). Piece by piece, we can recognize the same sort of correlation that we found in Nk-2, but here it is overwritten with some seemingly wild oscillations. A very similar pattern, with fewer irregularities, exists in segment Sk-1.34 The patterns resemble one another, not just broadly, but also in some details – most notably in the fact that both segments have an entry for Ipswich hundred (i1) intercalated into Bosmere hundred (b5). In the source text, it seems, part of b5 was followed by i1, which was followed by the rest of b5. Reconstructing B from D, we cannot hope to recover small features like this – that is one of the reasons why we cannot expect to find perfect correlation – and xEl is telling us something here which otherwise we would not know. Whatever these oscillations mean, it seems certain already that the two xEl segments for Suffolk were compiled at the same time, by the same method, from the same source text.







Nk-1 
























































Nk-2 

























Figure 10. Order of the entries in xEl-Nk-1 and xEl-Nk-2 mapped onto the order of hundreds in B-Nk.

Sawyer was at least on the verge of discovering these oscillations, but seems hardly to have been able to believe that they existed, let alone that they made any sense.35 Finn (1960) seems also to have had some inkling of their existence.36 The phenomenon is certainly real, and not beyond the reach of explanation. It results from a division of labour between two scribes, who take half of the B text each, and who jointly compile a pair of booklets like the pair for Norfolk. Scribe I takes the first portion of B, covering the hundreds from t1 to h1; scribe II takes the second portion, covering the hundreds from r1 to l2. Both start scanning through B; whenever one of them comes across an interesting paragraph, he copies it into the appropriate booklet; and the outcome is that, in both booklets, paragraphs copied by scribe I from his portion of B alternate with paragraphs copied by scribe II from his.37 If the origi-

one for the abbot’s manors, one for the manors belonging to other barons on which Ely had some claim. He might have scanned through B-Nk twice, making one booklet at a time; but the more sensible plan would be to scan through B-Nk just once, compiling the two booklets in parallel. It can be proved (see below) that this is how the booklets for Suffolk were compiled; so probably it happened here too. Each time the scribe came to a relevant entry, he read it through and made a decision. If Saint Audrey was securely in possession of this manor, he copied the entry into booklet 1; if Saint Audrey had a claim on the manor which she had not yet been able to make good, he copied the entry into booklet 2. And then he resumed his scan of B-Nk, looking for the next entry that he would need to copy. In the end, therefore, his booklet 1 was precisely analogous with one of the C booklets which had been or was about to be compiled by the treasury scribes, a booklet which would carry the title ‘Land of the abbot of Ely in Norfolk’ (corresponding with chapter 15 in D-Nk). But booklet 2 was made to the scribe’s own formula.

33 The reader who doubts that should redraw Figure 11, moving h1 to the end, and then wait to see what difference this makes to the following discussion. (The answer will turn out to be: no significant difference.) 34 The data tabulated by Sawyer (1955, p. 189) can be mapped onto Fig. 11 by relabelling the hundreds A, B, C, . . . , I, K, . . . , T.

Segment Sk-2. There were 25 hundreds in Suffolk, and the ordering as it stood in B-Sk can mostly be reconstructed without much effort.31 Two hundreds, Blackbourn (b3) and Bradmere (b6), stand out as being very badly behaved, and I do not see how they can be fitted into the sequence; 32 a third hundred, Hartismere (h1), is also hard to handle,

35

‘It is tempting to see . . . two sequences intermingled, but this would be difficult to prove and hard to imagine in practice’ (Sawyer 1955, p. 189). In fact, the result seen here is like the result that one gets by performing a riffle-shuffle on a pack of cards; but that is a simile, not an explanation.

36 ‘In Suffolk the [xEl] entries for each of five adjacent Hundreds [b5, c2, c1, w2, l2] are frequently not together, Hundred by Hundred’ (Finn 1960, p. 389). The word ‘adjacent’ goes to show that he was seeking inspiration in the wrong place. Instead of looking at the text, he was looking at a map.

31

Sawyer (1955, p. 189) seems to have thought that he could reconstruct the order entirely, but does not say how he managed it. 32

37

This is a just-so story, and there are other ways of telling it. We might suppose instead that one man did all the scanning and the other man did

They are anomalous in other ways too (Davis 1954, pp. xxvi–ix).

101

t1 t2 l1 b1 s2 i1 b5 c2 p1 p2 h1 r1 c4 b2 b4 c3 c1 w2 l2

Thedwestry Thingoe Lackford Babergh Stow Ipswich Bosmere Claydon Parham Plomesgate Hartismere Risbridge Cosford Bishop’s Blything Colneis Carlford Wilford Loes

The survey of the whole of England











Sk-1















Sk-2









Figure 11. Order of the entries in xEl-Sk-1 and xEl-Sk-2 mapped onto the order of hundreds in B-Sk. (paras. 1–11); scribe II writes the second block (paras. 40– 6); then scribe I takes over again. If we read this table row by row, we may think that the sequence is chaotic; if we read it column by column, we discover that the order of the entries is approximately the same as in D-Sk, and that the order of the hundreds is approximately the same as the order reconstructed for B-Sk. The reader is welcome to draw up a similar table covering segment Sk-2.

nal booklets survived, it would be obvious at once from the changes of hand that the oscillations are a by-product of this division of labour; even without the originals, thanks to the existence of D we can still work out what was happening. To illustrate the point, I work things out more fully for booklet 1, i.e. the original for segment xEl-Sk-1 (Table 31). The paragraphs are identified with the corresponding paragraphs in D-Sk’s chapter 21, as they were numbered by Rumble (1986).38 Scribe I writes the first block of text

Why the scribes chose to collaborate in this manner – why they chose to make one pair of booklets together, rather than each of them making a pair by himself – is another question. On the face of it, the plan is not a sensible one, because the scribes will get in one another’s way. From time to time, one of them will come across a paragraph which he needs to copy, only to find that the booklet into which he needs to copy it is already being used by his colleague. This would

all the copying. At any moment (except in the initial phase, when there is nothing yet for Writer to do), Finder is scanning through one portion of the B text, marking those entries that Writer will need to copy, and Writer is making his way through the other portion, copying those entries that Finder has already marked. Every so often, they change places (or exchange their portions of the B text, which comes to the same thing), and oscillation results. This is not a different conjecture; it is the same conjecture expressed in a different way. 38 I have simplified things slightly by ignoring some small transpositions. In hundred c3, for example, the actual sequence is paras. 48, 50, 51, 49, 52, but I write this as paras. 48–52. It will be noticed that there are two

entries here, near the bottom of the first column, which D-Sk puts into a different chapter.

102

The monks of Ely – Part II scribe I

hundreds

21/1–3 4 5–9 10–11

t1 t2 l1 b1

12–14 16–21 15

s2 b5 i1

22

b5

23–5 26–30

b5 c2

31–5

c2

21/36, 67/5–6 21/38

p1 p2

39

h1

scribe II

r1 c4 b2

21/40–1 42–4 45–6

b4

47

c3

48–52

c1

53–5

w2 c1 w2 l2

71–92 56–70 93–4 95–100

l2

101–4

And that will do the trick.40 In the event, it seems, the fair copy did not get made – or at least it did not get made till fifty years later, when scribe T1 did the job. By that time, not only were the booklets out of order: the fact that they needed special treatment had been forgotten, and scribe T1 just copied them as he found them. For present purposes, these oscillations are important only because they make it certain that booklet 1 was compiled simultaneously with booklet 2. Without this evidence, though we might be willing to assume that, we would not be able to prove it. As far as booklet 2 is concerned, the oscillations are ultimately irrelevant. It is the piecemeal correlation between the order observable in xEl and the order reconstructable for B which proves the point which we were hoping to decide: that segment Sk-2 must have been derived from the B text – or, if we think it necessary to state the conclusion more diffidently, from a B-like manuscript which, because not demonstrably different from B, was presumably identical with it.41 Thus it can be proved, directly for segments Nk-2 and Sk2, indirectly for segment Sk-1, that xEl is derived from B. The same is self-evidently true for segment Ca. Given these facts, it seems a fair conclusion – pending proof to the contrary – that the same applies to every segment of xEl. The basis for this conclusion is not as strong as it might be, because there is clearly some degree of heterogeneity among the segments, and accordingly some room for doubt as to whether what is true for one is true for all. That said, I cannot see any reason why we should be reluctant to think that the descriptive segments of xEl were all derived, as some of them certainly were, from the B version of the survey text. Conversely the evidence which we have developed from xEl can be taken to prove the existence of the B text, not only for Cambridgeshire, but also for Norfolk and Suffolk. Since the existence of the B text is also implied, by the surviving C booklets, for five counties on the other side of England, there is another obvious conclusion which it would be perverse not to draw: the B text existed originally for every county.

Table 31. Oscillations in segment xEl-Sk-1. probably not be a serious problem,39 but it could have been avoided altogether if each scribe had made his own pair of booklets. So we are left wondering why the less efficient plan was preferred. (The answer is, I would guess, that scribe I wanted to keep a close eye on the work that scribe II was doing.) Whatever the reason for its adoption, the plan is a recipe for entropy: it shuffles hundreds together in an arbitrary way. But perhaps there was more to the plan than meets the eye. I suspect – but cannot hope to prove – that it may have been intended to make a fair copy later, and to put the paragraphs back into their proper order at that stage. Perhaps scribe I was intending to do that himself; but let us suppose (to make things harder for ourselves) that the task was going to be assigned to somebody else, scribe III. As long as scribe III understands what he has to do, as long as he can distinguish scribe I’s writing from scribe II’s (he should pay attention to the hundred headings too), he will have no difficulty in achieving the desired result. For each booklet in turn, he makes two passes through his exemplar: on the first pass he copies all paragraphs written by scribe I, on the second pass all paragraphs written by scribe II.

5 One segment remains to be discussed. I have not yet explained why segment s, despite its different character, should be accepted as part of the original xEl text, and why 40 It is a nice question whether xEl’s next editor would be justified in rearranging the Suffolk segments along the same lines as scribe III. If this were known for certain to have been the original intention, an editor would have no choice but to respect it. 41 That is the conclusion for which Sawyer seems to be heading; but then at the very last moment he veers away. The source text, he decides, was not B itself: it was a copy of B in which the entries had already been reorganized to some extent. He had discovered, he thought, ‘a stage of the enquiry’ – of the compilation process, it might be better to say – ‘which has not yet been noticed’ (Sawyer 1955, p. 188). This conclusion is a non sequitur; but since it has been generally ignored, there seems to be no point in explaining why.

39

In these circumstances, the scribe does not have to stop working. He can insert a bookmark at the place where he has found an entry and then continue scanning. When the booklet that he has been waiting for becomes available, he goes back and copies the entry that he found in the first place, plus any other entries that he has found and marked in the interim. And then he resumes his scan from the point at which he broke off.

103

The survey of the whole of England it is especially significant. That is what I now propose to do.

that is given to Ely’s affairs, it has often been casually assumed that segment s originated in Ely; but that, as Finn (1960, p. 390) pointed out, is manifestly not the case. The monks of Ely had no interest in knowing the aggregate statistics for the manors in Cambridgeshire that were held by the knights of Hardwin de Escalers (para. 12). They were interested in knowing, for instance, that one of Hardwin’s men, Radulf by name, owned a manor of 2.25 hides in the village of Over, because this manor included 0.5 hides of Ely thegnland and 0.75 hides of Ely soke (para. 26/48); 43 but they were not concerned with Radulf’s other hide, the soke of which belonged to the abbot of Ramsey. It was of no significance to them that Hardwin’s men held a total of 34.625 hides in Cambridgeshire – and of even less significance that Hardwin’s men held a total of 18.5 hides in Hertfordshire (para. 13), on none of which (as far as the evidence goes) did Saint Audrey think she had a claim. An Ely scribe might decide (as scribe T1 and others did later) to copy these paragraphs if they already existed; that is likely enough. But no Ely scribe would have had either the motive or the opportunity for bringing them into existence in the first place.

Segment s is itself segmented. It consists of sixteen paragraphs, the contents of which can be summarized as follows: (1) The abbot of Ely in Cambridgeshire (2) Picot, Hardwin, Wido and others in the same: Ely thegnlands (3) The same in the same: Ely soke (4) The abbot of Ely in Hertfordshire (5) The same in Essex (6) The same in Norfolk (7) The same in Suffolk (8) The same in Huntingdonshire (9) Picot the sheriff in Cambridgeshire (10) The same in the same: Ely thegnlands (11) The same in the same: Ely soke (12) Hardwin de Escalers in Cambridgeshire (13) The same in Hertfordshire (14) The same in Cambridgeshire: Ely thegnlands (15) The same in the same: Ely soke (16) Wido de Rainbuedcurt in Cambridgeshire: Ely thegnlands, Ely soke

It has often been remarked that these summaries are similar to some that occur in a manuscript in Exeter – the same manuscript which largely consists of the surviving portion of the C text (Exeter Cathedral Library 3500).44 In fact, to say that they are similar is to misrepresent the case: it would be more accurate to say that they are identical.

Each paragraph begins by giving the name of the baron and the name of the county in question; then, for each of a series of designated categories of information (see below), it reports the total for this baron in this county. For some categories, but not for all, it distinguishes between the manors held by the baron in domain and the manors held by his knights. Trying to anticipate the sort of questions which might be put to him, the compiler worked out the answers in advance. Somebody, for example, might want to know the total assessment for all the manors held in domain by sheriff Picot in Cambridgeshire. Paragraph (9) supplies the answer (58.125 hides). Somebody might want to know the aggregate number of domain ploughs on all the manors held by Picot’s knights in the same county. The same paragraph supplies the answer to this question too (17.5 ploughs). There are some slight variations (in the wording, in the order of the last two items), and in some cases the scheme is simplified because the distinction between domain and non-domain manors does not apply; but on the whole these paragraphs all conform closely to the same underlying pattern.

The summaries in question are contained in a two-leaf booklet (fos. 527–8) which got itself connected with the surviving C booklets. They were written out very neatly (527v–8r) by the scribe whom I call mu (above, pp. 41–2); but since this seems sure to be a fair copy, they may originally have been compiled by one of his colleagues, rather than by mu himself. What they give is a condensed description of the abbot of Glastonbury’s lands in four counties: Wiltshire, Dorset, Devon and Somerset.45 The program which generated these summaries is represented in Table 32.46 Except that no distinction was made between et appreciata in breue abb’is de eli (T-112r). Similar notes follow paras. 14, 15 and 16. (There is no mystery about the expression ‘the abbot of Ely’s brief’: it refers back to paras. 1–3.) An itemized list of the holdings in question, derived by an Ely scribe from xEl-Ca, can be found in one of the segments appended to xEl (T-146v–7v); the ‘others’ mentioned in paras. 2–3 can be identified from this. Erchenger the baker, for instance, is listed as the tenant of 1 hide of thegnland and 0.55 hides of soke.

Some special treatment was required for the lands of a monastery like Ely, because of the complexity of the tenurial arrangements existing here. There are separate paragraphs for the thegnlands (para. 2) and the soke (para. 3); the totals reported here include the subtotals reported later for three individual tenants, Picot the sheriff (paras. 10–11), Hardwin de Escalers (paras. 14–15), and Wido de Rainbuedcurt (para. 16).42 Because of the prominent place

43 This is the paragraph quoted already (above, p. 98), as it appears in B / V and xEl / T. 44 ((Galbraith, for example, got as far as saying that they are ‘precisely similar’ (1961, p. 116); but he only said it in a footnote, and then said nothing more. As far as I am aware, the first person to realize that they are practically identical, and that this fact cries out for an explanation, was Roffe (2000, pp. 181–2).)) 45

A poor copy of the fourth paragraph was inserted by a later (perhaps twelfth-century) hand in a blank space elsewhere (173r).

42

46

A note following para. 11 explains that the data being counted here (paras. 10–11) have already been counted in the paragraphs giving the grand totals (paras. 2–3): H˛ec t’ra predicta de Thainl’ et soca est scripta

The summary covering the lands of Saint Petroc in Cornwall (528v) is by a different hand, but follows the same program, minimally adapted to suit the case.

104

The monks of Ely – Part II for each baron for each county summing over all manors held in domain how many manors? how many hides? how many ploughs on the domain? how many villains, bordars, slaves, others? and how many ploughs do they have? how much is this land worth? summing over all manors held by his knights how many manors? how many hides? how many ploughs on the domain? how many villains, bordars, slaves, others? and how many ploughs do they have? how much is this land worth? summing over all manors held by his thegns how many manors? how many hides? how many ploughs on the domain? how many villains, bordars, slaves, others? and how many ploughs do they have? how much is this land worth? summing over all manors how many ploughs is this land sufficient for? how much has it gained in value since it came into his hands? next county next baron

Table 32. Program for generating the summaries in Exeter Cathedral 3500, fos. 527v–8r. knights and thegns, the summaries included with xEl were generated by exactly the same program.

ing is determined by the content: there is only one way of saying ‘ten hides’, ‘twenty ploughs’, ‘fifty villains’. As far as the xEl summaries are concerned, it is to be assumed that they had already been copied at least once before being copied into T, and we cannot be sure how accurately the surviving copy represents the original. Nevertheless, there are two specific agreements to be noted.

If they have not had cause to think about such questions before, readers may think that one summary is bound to look very much like another. The answer to that (for those readers willing to admit that the thought did cross their minds) can also be found in the Exeter manuscript. A different scribe (not one whom I recognize) copied out another batch of summaries (530v–1r),47 and the program for generating these is represented in Table 33 Though the loops are the same, the questionnaire is quite differently structured, as well as being much shorter.48 Only three of the questions being asked here (the ones marked with an asterisk) are exactly equivalent to questions being answered in the summaries written by scribe mu.

(1) In every county where the survey was carried out, one of the questions which had to be answered was this: how many ploughs would there be on this manor if the land were being fully exploited? It is not clear exactly how the question was worded, nor whether the wording was the same with respect to every county; the answers are variably phrased. In B-Ca, the formula used is x car’ ibi e’ t’ra or some permutation of that (x car’ e’ ibi t’ra, t’ra e’ ibi x car’). The DB scribe, in the first booklet he wrote, DB-YoLi, experimented with several formulas – quas poss’ arare x car’, et x car’ poss’ ibi esse, ubi x car’ poss’ esse – before deciding that T’ra e’ ad x car’ was clear enough; but then he started wondering whether to omit the e’ or the ad or both. The Exeter summaries use a distinctive formula:

In the wording too, we find some striking agreements between the summaries surviving in Exeter and the summaries surviving in this Ely text. Up to a point, of course, the word47

The second page (531r) is reproduced by Darlington (1955, opp. p. 217).

48

The man who compiled these summaries was aiming for cumulative totals. In the case of Rotbert son of Girold, for example, he starts by giving the numbers already arrived at for Wi + Do; then he gives the numbers for So; and then he sums each pair of figures to get the numbers for Wi + Do + So (530v). Carrying on like this, he would eventually have arrived at numbers for the whole of England. To judge from the surviving copy, however, these were just desultory experiments; it is hard to believe that the compiler had enough momentum behind him to get very far with this plan.

H˛ec terra sufficit x car’ (four instances), which (unless I have missed something) occurs nowhere else except in the xEl summaries, some of which (but not all of which) use the identical formula:

105

The survey of the whole of England for each baron for each county summing over how many how many how many how much summing over how many how much next county next baron

all manors manors? hides? ploughs is this land sufficient for? * is it worth? all manors held in domain hides? * is this land worth? *

Table 33. Program for generating the summaries in Exeter Cathedral 3500, fo. 531r. H˛ec terra sufficit x car’ (seven instances).

probably by the same man. And that means that they must have been compiled, from some version of the B text, in the king’s treasury at Winchester.

(The other paragraphs replace this with T’ra x car’ or T’ra ad x car’.) In both cases, moreover, the total reported is the total for all manors, regardless of whether they are held in domain or not.

How far this plan was followed through with, we are not in a position to decide. The summaries written by scribe mu look as if they may have been recopied for some special reason, perhaps in response to a demand for information on this particular topic. The summaries included with xEl are just the ones which caught the attention of a visiting Ely scribe. Even so, these accidental survivors span eleven counties (including Saint Petroc’s lands in Cornwall), and it would be a strange fluke if these were the only counties for which summaries had been compiled. On the other hand, it would hardly seem sensible to draw up briefs of this kind except for baronies of a certain size. It is not impossible – I do not think it should be put more strongly than this – that the Treasury officials drew up a summary for every important baron, for every county where he was a holder of land. That it could have been done is clear: how far it was done we are never going to know.

(2) The Exeter summaries do not report the total value. We can work this out for ourselves, if we wish, by adding together the values reported for each category of manors (domain, knights, thegns); but the answer is not written down for us. Instead the compiler recorded a different number: not the value, but the change in value; not (as perhaps we might expect) since the time of king Edward, but since the time when the current owner got possession. If the value has increased, he writes that ‘this land has improved by ten pounds in the hands of abbot Turstin’: H˛ec t’ra emendata e’ in manu turstini abbatis x lib’. If the value has decreased, he writes that the land ‘has worsened by ten shillings’,

But that, in a sense, does not matter. Regardless of how far the program was carried out, the program itself is significant. It tells us what the Treasury thought about the survey. These, in the Treasury’s view, were the most important facts, and this was the most convenient way for those facts to be arranged. For anyone who hopes to understand what purposes the survey was intended to achieve, this is crucial evidence. Thanks to Ellis, it has all been available in print for nearly two hundred years: perhaps someone should finally make use of it.49

e’ peiorata de x sol’. If the value is unchanged, he writes nothing. Presumably what he did was add up all the Modo values, add up all the Quando recep’ values, and then subtract the second sum from the first one. But only the result of this subtraction got recorded. The idea that one ought to be tracking the change in value seems to me a very sophisticated notion, and I do not know that we find any trace of it elsewhere – except in the summaries from Ely. At precisely the same point in the program, precisely the same calculation has been performed, and precisely the same formula is used for reporting the result. This land ‘has improved by ten pounds in the hands of abbot Simeon’:

6 By way of conclusion, I give a short account of the history of xEl, as far as I think I can make sense of it. Much of what follows is guesswork, and I do not pretend otherwise; but it seems to me that the account given here is more coherent, and more consistent with the facts, than any alternative

emendata de x lib’ in manu Symeonis abb’is. The holdings covered by the xEl summaries had all gone up in value, and therefore we find no instance of the peiorata formula. Apart from that, I cannot imagine any better proof that these summaries – the ones in Exeter, the ones in xEl – were all compiled at the same time, in the same place, very

49 ((For the reader’s convenience, I have reprinted both stretches of text in Appendix II.))

106

The monks of Ely – Part II do not insist on that.51 What we are definitely guessing that he does is this: he copies the lists of jurors’ names for every hundred in which Ely has an interest. That may look like an odd thing to do, fifty years after the event, when hardly any of the jurors can still have been alive; but it may have seemed less odd at the time, when the new bishop was trying to assert his authority, both against the monks and against the church’s tenants. Whatever the motive for making them may have been, these extracts from B attach themselves to the top of the stack of xEl booklets.

account that I can think of. I have simplified it in some respects; no doubt I have made it too simple. The story begins in Winchester, with the arrival at the Treasury of a scribe from Ely (perhaps one of the monks, perhaps an employee of the abbot’s). It is not clear when we want to think this happened; but probably it happened in abbot Simeon’s time, i.e. no later than 1093. The man from Ely brings with him a letter (perhaps from the king, perhaps from someone else) ordering the Treasury officials to give him access to the commissioners’ reports (the B text) for those counties in which Ely has an interest, and to let him make whatever excerpts he likes.50 The Treasury officials comply. Working quickly but not in a rush, the Ely scribe combs through the commissioners’ reports, finding and copying every paragraph which makes any mention of Ely. (For some of the time at least, he has a companion working alongside him.) While he is in Winchester, he comes across a collection of summaries compiled by a Treasury scribe, and copies some of those as well. He then returns home with a batch of booklets containing all the excerpts that he has made, perhaps intending to copy them out again more neatly, when he has the time. But that intention, if it existed, comes to nothing.

Eventually, in 1139–40, scribe T1 makes a copy of the whole stack of booklets, turning the entire contents into one consecutive text. Some time later, the booklets cease to exist, but this copy survives; and two other copies made from this one also survive. That is the end of the story, so far as xEl is concerned. Meanwhile the bound volume of B (or the collection of ‘hundred rolls’, whatever it may have been) is not returned to the Treasury: it remains in Ely, and finally scribe V1 decides to make a copy of it. Perhaps he copies the whole text, covering several counties; perhaps he copies only part of it; we do not know and never will. The original then disappears, and so does some part of the copy that has been made of it. All that survives is a fragment of this copy, broken off at a point about three-quarters of the way through the first county. And that is the end of the story, so far as the B text is concerned.

For the next fifty years, these booklets sit on a shelf in Ely. They are in the monks’ custody by now; they are not forgotten. From time to time, for one reason or another, somebody looks through them, extracting some information which seems of interest to him; and a number of these derivative texts attach themselves to the bottom of the stack of booklets. At some stage, a portion of the B text becomes available in Ely. If anyone wishes to visualize it as a collection of rolls, I have no objection at all. Simply to fix ideas, let us think of it rather as a bound volume, B-Ca.. Ht.. , of which we can stipulate three properties: it begins with Ca; it contains one further county of interest to Ely, namely Ht; it does not contain any of the other four, Ex, Nk, Sk or Hu. We can also be sure that it arrived in Ely before 1140; so the likelihood is that it arrived at the same time as bishop Nigel, who had earned his promotion to the episcopate by serving as the king’s treasurer. (Why only one B volume arrived is probably not worth asking: there are too many possible answers. Perhaps the other volumes no longer existed; perhaps the Treasury would not release them; perhaps Nigel only wanted to borrow this volume, the one which covered Ely itself.) Within the next few years, somebody copies some extracts from this book. Perhaps he copies the title, HIC SUBSCRIBITUR INQUISITIO TERRARUM . . . ; we

50 Probably the letter came from someone else, with authority to issue such orders: if the letter had come from the king, we would expect it to have been preserved. One of the writs which does survive (Bates 1998, no. 127, to be punctuated in the manner suggested by Round (1895, p. 133)) is important in several respects – as proof, for instance, that abbot Simeon was displeased with the treatment accorded to him by the commissioners – but cannot be connected specifically with the compilation of xEl.

51

Not knowing where this passage came from, we cannot think of relying on it. Does that matter? If we want to know what questions were being asked, we can work things out for ourselves, from the evidence of B-Ca; it would be just laziness to quote this passage instead.

107

Chapter 10 The conduct of the survey: the fieldwork phase

Even in 1086, the survey was not quite the only thing on people’s minds. That very year, bishop Rotbert of Hereford was working on a tract which he hoped would persuade his readers that it was not the year 1086 after all – that in fact it was already the one thousand one hundred and eighth year since the Incarnation. This theory did not originate with Rotbert: as he was careful to explain, it originated with a monk of Mainz, Irish by birth, Marianus by name. (As Marianus himself had pointed out, the difficulties with the conventional chronology had been recognized already by Beda.) Marianus had died in 1082; bishop Rotbert took up the cause. His contemporaries were not convinced, but this tract of Rotbert’s did not disappear altogether. Some people read it; during the twelfth century at least two scribes were told to make a copy of it; and those two copies survive. The tract would be, for modern readers, of very little interest, were it not for the fact that Rotbert allowed himself, at the end of one of his chapters, to indulge in an entirely irrelevant digression. This present year, he says, the year mistakenly supposed to be the year 1086, is the twentieth year of king Willelm, the same year in which, by order of the king, there has been made a survey of the whole of England.

this account, but also to know for certain when (to within a few months) it was written, and by whom it was written. But it is not very long or very detailed. Because he is wandering from his theme, Rotbert rations himself to less than a hundred words (91 words, to be precise); and he assumes that his readers will know nothing in advance. The things which he chooses to tell us, therefore, are the most elementary things: for somebody who might wish to learn about the survey, these are the first things to know. Because of the surviving documentation, we are, as it happens, not nearly as ignorant as Rotbert expected us to be. For the most part he tells us nothing that we cannot work out for ourselves. It also has to be said that we do not know, with any exactitude, how closely bishop Rotbert had been involved in the workings of the survey. There are no personal touches (not that one would expect there to be) in this short account. It is certain that Rotbert would have been expected to attend the meeting of the king’s court in Gloucester at which the survey was discussed (see below), as well as the synod which followed; it is not quite certain that he did actually attend. It is not known that he served as one of the investigators; it is not known that he did not. But there were four counties where his church owned land – Herefordshire, of course, plus Gloucestershire, Worcestershire and Shropshire – and for these four at least he had no choice but to participate. As far as we know, bishop Rotbert was given no reason to complain that he and his church had been unjustly treated.

Hic est annus uigesimus Willelmi regis Anglorum, quo iubente hoc anno totius Angliae facta est descriptio. After the twelfth century, this passage seems to have been entirely lost sight of until it was discovered and put into print by Stevenson (1907).1

For us, the chief significance of Rotbert’s account lies in the next-to-last sentence. This is where we find the explicit statement – immediately recognized by Stevenson (1907, p. 75) as ‘the greatest addition to our knowledge of the Survey’ resulting from his discovery – that the operation proceeded in two stages:

Thus, from Rotbert’s own words, we know that this is a strictly contemporary account of the survey, written before the end of 1086. His last sentence is a rather cryptic remark about the disturbances which occurred in the aftermath of the survey – Et uexata est terra multis cladibus ex congregatione regalis pecuniae procedentibus – and that cannot have been written, it seems to me, till July or so at the earliest.2 We can count ourselves fortunate, not only to possess

Alii inquisitores post alios et ignoti ad ignotas mittebantur prouincias, ut alii aliorum descriptionem reprehenderent et regi eos reos constituerent.

1 As far as I am aware, no further copies have been found. With only two copies, neither of which is uniformly better than the other, Stevenson had no option but to pick and choose between them. (Among the variants cited by Stevenson, there are, by my count, four places where A is obviously better than B, eight places where B is obviously better than A.) In one place I am inclined to think that Stevenson made the wrong choice. It seems to me that B’s tuguria tantum habentibus, ‘those just owning huts’, should be preferred to A’s tuguria tantum habitantibus, ‘those just occupying huts’: it makes for a sharper contrast with domos et agros possidentibus, ‘those possessing houses and arable land’.

One team of investigators produced a written report (the word for which is descriptio); a second team of investigators was then sent to check this report and to notify the king of any misconduct that might come to light on the part of the into the chronicle compiled soon afterwards by the monks of Worcester (Stevenson 1907, pp. 76-8, ed. McGurk 1998, p. 44). Hence it was known to historians long before Stevenson’s discovery of the source text. Torn from its context, however, it did not make much sense. I discussed this sentence previously (above, p. 66) and have nothing further to say.

2

This remark found its way, in shortened form, into the additions made at the end of a copy of Marianus’s chronicle; and from there it found its way

108

The fieldwork phase full length.7 Though the author is apparently still the same man,8 it seems clear that there is some discontinuity in the text at the end of the annal for 1086; and that seems to imply that his account of the survey is a strictly contemporary account, written in 1086–7. It is thus to be treated with the same respect as bishop Rotbert’s account. But the tone of it could hardly be more different. Instead of the stiff, unemotional remarks which we get from Rotbert, here we are given the impressions of somebody still recovering from the shock of seeing the surveyors in action.

first team. Rotbert’s language is emphatic – he makes the point twice (alii post alios, . . . alii aliorum), in case anyone missed it the first time – and there cannot be any doubt as to his meaning. He tells us, moreover, that the second team consisted of men who had no stake in the counties which they were investigating. In a neatly turned phrase (which works well in Latin but is not easy to imitate in English), Rotbert tells us that they were sent to counties which they did not know and where they were themselves not known (ignoti ad ignotas mittebantur prouincias).3 That is explicit. That the opposite was true for stage 1 – that the original survey had been carried out by local men – is not stated in so many words; but it is certainly implied. Rotbert assumed that his readers would be capable of seeing the intended contrast for themselves. As far as his modern readers are concerned, that expectation seems first to have been met by Barlow (1963, p. 285).4

Towards the end of his annal for 1085, this author tells us that at Christmas the king held court at Gloucester for a period of five days (perhaps 23–27 December 1085).9 One early item of business was the appointment of three new bishops; then the king and his council turned to the question of the survey. About this, we are told (in words which have been very frequently quoted), there was ‘much deliberation and very deep discussion’. After this meeting, presumably without delay, the survey got started; and this is how the chronicler describes it:

An anonymous chronicler, writing in English, left us another account of the survey which is arguably also contemporary. The only surviving copy of the English text is a manuscript written at Peterborough in the 1120s (Oxford, Bodleian Library, Laud 636),5 but it is clear, from allusions in various Latin chronicles, that copies of an English chronicle not very different from this one were quite widely available. At least one of those copies – the copy (possibly borrowed from Canterbury) which became the exemplar for the Peterborough copy – would have been, if it had survived, of greater value than this one.

Then he sent men of his over all England into every shire and had them find out how many hundred hides there were in the shire, and what the king himself owned in the way of land and livestock on the land, and what customs he ought to have from the shire in any twelve-month period. He also had it recorded how much land his archbishops owned, and his bishops and his abbots and his earls, and (though I am making too long a tale of this) what or how much each man who was a tenant of land in England owned, in the way of land and livestock, and how much money it was worth. So very closely did he have it investigated that there was not one single hide, not one yard of land – not even (it is shameful to tell it, but he did not think it shameful to do it) one ox or one cow or one pig was left out – that was not set down in his record. And all the records were brought to him afterwards.10

Because this copy was written in a single stint, as far as 1121, and because there is, after 1079, no other English manuscript with which it can be compared, the evolution of the text has to be traced by looking for internal evidence. There is only one point which seems at all significant here. In describing the events of 1086, the author twice makes a point of telling us that he does not know – only God does – what the sequel will be.6 Moreover, the annal for the next year begins, uniquely, by writing out the date in English, at

7 ‘One thousand and eighty-seven years after the birth-time of our Lord Jesus Christ, in the twenty-first year that William ruled and governed England, as God granted him, . . . ´(Swanton 1996, p. 217). This annal runs on into the author’s obituary of Willelm I, an attempt to draw up a balanced account of the good things and bad things which ought to be remembered about him. (He writes, he says, as someone who has looked upon the king, and lived in his court for a time; but the looking, it seems clear, was only done from a distance.) Here he reverts very briefly to the survey: Willelm was king of England, ‘and by his astuteness it was so surveyed that there was not one hide of land in England [of which] he did not know who had it or what it was worth, and [which was not] afterwards set down in his record’ (p. 220). The wording echoes that of the previous passage (thet næs an hid landes, . . . on his gewrit gesett), but the indignation has faded: the word geapscip, though I gather that it often carries a negative charge (‘crookedness, craftiness’), is here intended as a compliment (‘shrewdness’), not a reproach. Anyone who needs to sum up the survey in a single sentence might do worse than think of quoting this one.

3 Rotbert’s language is classical Latin: he uses the word prouincia (twice) in preference to comitatus. Stevenson seems to hesitate briefly (at first he translates prouincias as ‘districts’); but then he cites a passage from Hemming’s cartulary (below, p. 116) where it is clear that comitatus and prouincia are synonymous. 4

It is worth noting, perhaps, that the passage was misconstrued by Stenton (1943, p. 609 = 1971, p. 618). 5

Published in facsimile by Whitelock (1954). Plummer’s (1892–9) edition, though it needs to be checked against the facsimile, is still valuable. The concluding stretch of text, from 1070 onwards, was printed again by Clark (1958, 1970); there is now a new edition of the entire chronicle (Irvine 2004). Several translations into modern English are available: except where I indicate otherwise, the passages quoted below are taken from Swanton’s (1996) version.

8

In speaking of the famine of 1087, he refers back (‘as we already told’) to his remarks about the bad weather of 1086.

9

‘And afterwards the archbishop and ordained men had a synod for three days.’ This has to be read as a parenthesis: the business of which we hear next was conducted at the king’s court, not at the archbishop’s synod.

6 The aetheling Eadgar, thinking himself unfairly treated by the king, has gone off on some adventure of his own: ‘May the Almighty God give him honour in the future’. The weather has been uncommonly bad this year: ‘May God Almighty remedy it when it be his will’ (Swanton 1996, p. 217).

10 I translate this passage rather loosely; there are tighter translations to be found, such as Swanton’s (1996, p. 216), if that is what the reader would prefer. Anyway the sense is clear; and it would be unwise to argue anything from nuances in the wording. (In particular, the last sentence has

109

The survey of the whole of England Though there is nothing here that we cannot work out for ourselves from other evidence, an eye-witness narrative, written down within a year or two of the event, has to have some special value. Though we might have preferred a little less indignation, a little more information, the account seems perfectly reliable, as far as it goes. But it was not written by somebody who had seen the workings of the survey from the inside: it is the report of a horrified observer.11

them a copy of the schedule listing the questions that they were required to ask. They would (even in the eleventh century) probably have needed some written authorization – a letter addressed to all the barons of B shire, French and English, ordering them to obey the commissioners’ instructions as if they came directly from the king; a letter to the sheriff of B shire, ordering him to put himself and his agents at the commissioners’ disposal. Some scribes would certainly have accompanied the commissioners, but nothing is known about them.

1

From what bishop Rotbert says (from what he says explicitly about the second teams, and from what he thus implies about the first teams), we can be certain that the commissioners were local men – men who knew and were known by the barons of the target county. Beyond that we can be sure of almost nothing. In Wiltshire and presumably elsewhere, the commissioners were expected to collect belated payments of geld (above, pp. 67–8): for that reason they make an occasional appearance in the Wiltshire geld account as ‘Walter and his companions’. That proves (if we need it to be proved) that the survey of each county was conducted by a team, not by one man; but apparently one man was understood to be in charge. It seems that this Walter must be Walter Gifard,14 whom we shall meet again in Worcestershire (below, p. 116). Walter held only one manor in Wiltshire – but did hold it in domain (DB-Wi-71va) – and perhaps we may suspect that he would not have been put in charge of the survey here unless he was also in charge of the survey of some other nearby county where he owned a larger block of property.15 The commissioner named S to whom Lanfranc wrote a letter (below, p. 114) was certainly concerned with more than one county (if we can be sure, as I assume we can, that Lanfranc was being exact). These are just straws in the wind, but they suggest that we ought to allow for the possibility that one team of commissioners may sometimes have been responsible for two or more (presumably adjoining) counties. Eventually, with much effort and some luck, we may be able to work this out a little further, but as yet we do not even know what sort of evidence is going to be admissible.

Even if the chronicler had not heard of it before, the decision to conduct a survey had probably been made well in advance of the meeting at Gloucester.12 By December 1085, the king was not asking his councillors whether they agreed that the survey was a good idea. That had already been settled; it was too late now for anyone to ask the question put by Joab to king David: Sed quid sibi dominus meus rex uult in re huiuscemodi? (2 Samuel 24:3).13 The business of the meeting was to discuss the implementation of the policy; and essentially that means two things. For the country as a whole, the king and his council had to agree on a final list of the questions to be asked. For each county, they had to appoint a panel of commissioners to carry out the hard work. The men appointed would not have been chosen unless the king trusted them; but they were being given power which they might be tempted to abuse, and no doubt it had already been agreed – and made known to everyone concerned – that the survey of each county would be verified eventually by a second team of commissioners. Some deadlines would also have had to be settled – most urgently a deadline for the submission of the B text, in its unverified form. Those decisions having been made, the survey could begin at once; and presumably it did. It was the middle of winter, to be sure, but the king did not expect anyone to delay doing their duty on account of some inclement weather. As they departed from Gloucester, the commissioners entrusted with the survey of B shire would have carried with

It will, I fear, always be a struggle to find anything sensible to say about stage 1 of the survey – the stage which ended with the completion of the B text in its original (unverified) form. To make any progress, we shall have to learn how to filter out those changes in the written record which were made during stage 2 of the fieldwork phase, and then during all three successive stages of the compilation phase. (And we shall have to begin by unlearning

no deep meaning. It is just the author’s way of wrapping up the paragraph: he began with the king, and so now he ends with the king. With dismal regularity, one finds it being assumed that ‘afterwards’ means ‘at Salisbury on 1 August’. The chronicler does not say this; nor does he imply it.) 11

By appending this account to his account of the meeting at which the decision was made, the author runs on into the early months of 1086, following the story through to its conclusion. The next annal begins by telling us where the king was at Easter (5 April 1086); but quite possibly the author is fetching back in time as he starts a new line of narrative.

12

14

Except for the coincidence in date, there is nothing to suggest that the survey was a consequence of the invasion scare of 1085. But the scare is what brought the king back to England; if he had stayed longer in Normandy, the survey would presumably not have happened till later – in which case it might not have happened at all.

The only other Walter occurring in Wiltshire is Walter (here and usually called Walscin) de Dowai. He held two manors in this county, but neither of them in domain (72ra), and I take that to be enough for us to rule him out.

15

The head of Walter’s barony was Long Crendon in Buckinghamshire, and this was the county where he held the largest number of manors (DBBu-147ra–8ra). One cannot get from Buckinghamshire to Wiltshire without passing through at least one other county. I merely mention these facts; I draw no conclusions from them.

13

Joab was advising against the proposal to count the people of Israel. The king overruled him. Joab and his colleagues went off to conduct the census: nine months and twenty days later they returned to Jerusalem and Joab reported the final figure to the king. Then the pestilence began . . . .

110

The fieldwork phase many wrong ideas – above all the idea that properties of B can be instantly inferred from properties of DB.) For one county, Cambridgeshire, we know what the B text looked like – not with regard to its external appearance, but with regard to its internal structure. In the form in which it survives (above, pp. 89–91), B-Ca is some distance away from its original form: it includes passages added by the second team of commissioners; it excludes all paragraphs referring to the manors held by the king as the successor of king Edward. (On top of this, the copy is not a very accurate one; on top of that, part of the copy has been lost.) But those are not the problems which concern us here. We are asking how the B text was put together in the first place. How did the facts pass from the people who knew them to the people who wanted to know them? That is what we should like to know.

and the monks were dependent on the benevolence of the king’s agents. We are told, in a vague but heartfelt remark, that they ‘suffered many hardships’.18 One document is chiefly of interest here, the one which takes up the largest amount of space (fos. 2va–5rc).19 It consists of a manor-by-manor description of the lands in Kent which belonged to the archbishopric of Canterbury, in the largest sense of the word. There are three segments: the first covers the lands held by the archbishop himself (or in some cases by his knights); the second the lands held by the monks of Christ Church; the third the lands held by the bishop of Rochester.20 Here I propose to concentrate on the first segment (fos. 2va–3va), leaving the others to be dealt with (I hope) in another context. The order in which the manors are listed here is largely the same as in another document which had got itself included in the same collection, a schedule of the farms and other payments due to the archbishop from his domain manors (fo. 5va–c).21 This order was one which made sense to the archbishop’s officials, not to anyone else (Table 34). The first 18 paragraphs cover the manors in Kent, the last eight the manors elsewhere (Surrey, Middlesex, Sussex, in that order). Five manors listed in the longer text (their names are printed bold) are missing from the schedule of farms, for the simple reason that none of them was held in domain;22 otherwise the ordering is (except for Sundridge)

The commissioners responsible for the survey of Kent,16 at some early stage in their proceedings, sent out a questionnaire. We do not know how widely it was distributed; we only know that a copy of the questionnaire was addressed to archbishop Lanfranc; and we know this because a copy survived of the document drawn up in response. It was one of a batch of documents which seem all to have originated in the archbishop’s chancery: except for one comment which cannot have been added till after Lanfranc was dead, they can all be assumed to date from before May 1089. After that, they came into the hands of the monks of Christ Church (or of Holy Trinity, to use the name which Lanfranc would have preferred). The originals were, one assumes, an unprepossessing batch of business records, of various sizes and shapes. It was the monks who decided to promote them to the status of holy writ by having them copied out (seemingly by a professional scribe) on eight huge sheets of parchment (Canterbury, Cathedral Library Lit. E 28).17 They did this (I suppose) because it soon became alarmingly obvious that the king – Willelm II by now – had no intention of allowing a new archbishop to be appointed within the foreseeable future. Almost four years elapsed before Lanfranc’s successor was chosen, another nine months before he showed his face in Canterbury. During all that time, the archbishopric was in the king’s hands,

18

Though the text known to historians as the ‘Acta Lanfranci’ includes some passages of pseudo-history aimed at Saint Augustine’s (in this version it dates from about 1120), it is mostly based on a reliable text written after but perhaps not long after the arrival of Lanfranc’s successor. The concluding paragraph, covering the period from May 1089 till December 1093, has a contemporary ring to it. Post obitum Lanfranci caruit aecclesia Christi pastore iiii or annis, mensibus ix et diebus ix, in quibus multa aduersa perpessa est. Anno uero dominicae incarnationis millesimo xciii datus est pontificatus Cantuarberiae Anselmo beccensi abbati ii nonas martii, uiro probo, bono, apprime erudito, et sui temporis nominatissimo. Venit autem Cantuariam vii kalendas octobris, multis eum causis rationabilibus detinentibus quod prius uenire nolebat, et sacratus est ii nonas decembris (Cambridge, Corpus Christi College 173, fo. 32v, published in facsimile by Flower and Smith 1941). 19

In this document and the one before it (fo. 2rb–c), rubrics were inserted later by a different scribe. They have no particular claim to be authentic: the man who wrote them could have invented them extempore, and his place-name spellings, which usually diverge from those in the original text, suggest that this is precisely what he did.

16 We have no clue who they were, unless it is to be inferred, from the fact that Adam son of Hubert was a member of the second team which visited Worcestershire (below, p. 116), that he (like Walter Gifard) had also been a member of one of the first teams. If so, Kent is the county in which he would be most likely to have served.

20 It was pointed out by Urry (1967, p. 26) that a copy of another version of this text, lacking the third segment, occurs in an early thirteenth-century Christ Church register (Canterbury, Dean and Chapter, Reg. K, fos. 70rb– 2va). This version has been very heavily reworked, losing most of its value in the process; but there are some indications, as Urry observed, that it was not derived from Lit. E 28. As for the third segment, a version of this was copied into the cartulary compiled at Rochester in the 1120s (Strood, Medway Archives and Local Studies Centre, DRc/R1, fos. 209r–10r). All this evidence I hope to discuss in detail at some future date.

17 This manuscript has come to be known, unaptly, as the ‘Domesday Monachorum’, an eighteenth-century joke which we should not feel compelled to repeat at every opportunity – or ever. It was published in facsimile by Douglas (1944) and is described and discussed in a valuable paper by Cheney (1983). Though Douglas (as far as I can see) omitted to mention this fact, the facsimile is only two-thirds the size of the original – two-thirds linear, so less than half by area. (The dimensions of the facsimile have sometimes been quoted as the dimensions of the original; let the reader beware.) The leaves are ruled for three columns and 54 lines. The coloured initials (red, blue, green, purple) are original; the scribe left spaces for document and paragraph headings but never got round to supplying them.

21

It was noted by Lennard (1959, pp. 119–20) that the coloured initials emphasize the division of this schedule into 26 paragraphs; and he was inclined to infer that it represents a rota of fortnightly payments (p. 131). But I do not see how that can be reconciled with the fact that the list is organized cadastrally, county by county. 22 In three cases this is clear from DB, where the corresponding entries (Brasted, Eynsford, Ulcombe) have dropped down into a separate sub-

111

The survey of the whole of England very nearly the same.23 Thus it seems certain that the longer text was drawn up by one of the archbishop’s officials, and that it was based, so far as the order of the paragraphs was concerned, on a list or schedule not very different from the one which survives independently.24 We may guess that the monks’ manors were dealt with in a similar way, but there is no contemporary list of the farms that were payable from them.25

1 2 3 4 5 6 7 8 9

By ordering the entries as it does, this text seems to be saying that it originated inside the archbishop’s administration. In other respects it seems to be saying that it was drawn up for a purpose decided outside that context – not by the archbishop, but by the commissioners responsible for the survey of Kent. To begin with the obvious point, the lands of the archbishopric extended over nine counties (below, p. 114), but this text deals only with one. (We cannot say whether similar texts were produced for all or any of the other counties; if they were, they failed to survive.) The facts reported are all facts which we know were of interest to the commissioners: of course they were of even greater interest to the archbishop himself, but we do not know that he had any immediate motive for assembling this particular range of facts at this particular time.

10 11 12 13 14 15 16 17 18 19 20

In any case it is clear that some textual relationship exists between this Canterbury text and the B text of the survey. The best proof of this comes from the second segment, in the paragraph relating to the town of Sandwich, because, fortunately for us, the monks of Saint Augustine’s had enough of an interest in this place to include a version of the relevant entry in their excerpts from B-Ke (below, p. 123).26 I print the two paragraphs side by side (Table 35),

21 22 23 24 25 26

schedule of farms

archbishop’s manors

Westgate Petham Bishopsbourne Wingham Aldington Lyminge Sundridge Reculver Herne Boughton under Blean Teynham Charing Gillingham Maidstone Northfleet Bexley Otford Wrotham East Malling Darenth

Westgate Wingham Bishopsbourne Petham Aldington Lyminge Reculver Herne Boughton under Blean Teynham Charing Pluckley Gillingham Maidstone Northfleet Bexley Crayford Brasted Otford Sundridge Wrotham East Malling Darenth Eynsford Ulcombe

Croydon Mortlake Hayes Harrow South Malling Tarring Pagham Lavant Tangmere

Table 34. Order of the entries in the schedule of farms (Lit. E 28, fos. 5va–c) compared with that in the description of the archbishop’s manors (fos. 2va–3va).

chapter (DB-Ke-4rb–va) covering the lands of the archbishop’s knights. (The indications are that this subchapter was created by the D scribes.) In the other two cases the same was true, but DB has lost sight of the fact: in the light of later evidence, we can be sure that Pluckley and Crayford were also both out on lease, to Willelm Folet and Hugo de Port respectively (cf. Du Boulay 1966, pp. 364, 338). (There is a mention of Crayford in the schedule of farms, but it forms a sort of footnote to the Bexley paragraph.) Another blunder in DB (probably not the DB scribe’s fault) is the absence of an entry for Teynham: to judge from the ordering of the archbishop’s knights’ manors (DB-Ke-4rb–va), it ought to follow the entry for Pluckley.

so that readers can judge for themselves, but the verdict is not in doubt: one text is derived from (some version of) the other.27 Either the Canterbury text (or something like it) was one of the sources used in compiling the B text; or else it was extracted from the B text (or something like it), the order of the entries being changed (so we should have to suppose) to bring them into line with a Canterbury text resembling the schedule of farms. From every point of view, it seems to me, the first theory is to be preferred.

23

But very different from the ordering in DB. Very loosely one might say that the order is back-to-front; but such slight correlation as exists is accidental. DB tends to run from west to east, and this Canterbury text, like the schedule on which it was based, tends to work outwards from Canterbury. 24

The order is reproduced exactly (except for the omission of Brasted) by another document in the same collection (fos. 2rb–c); but this – despite some suggestions to the contrary – is obviously just an epitome of the longer text, of no independent value.

paragraph, reports only the first figure, but tries to make it clear that the information is already out of date (DB-Ke-3ra). His attempt to explain this has been misunderstood by some historians as proof that DB is one year earlier than the Canterbury text. The mistake originated with Ballard (1920, pp. xix–xx); it was repeated by Douglas (1944), and has been repeated occasionally since.

25 A later schedule does exist which claims to be describing the arrangements put in place by Lanfranc (Urry 1967, pp. 26–7). But it has no claim to be contemporary – certainly not in this version, which dates from about 1200, probably not in any version remotely resembling this one.

27 In this sort of situation, there are only two possibilities to be considered. Historians (when they wish to give the impression that they are thinking logically) often say that there are three: A from B, B from A, or A and B both from X. But when it is not a serious possibility that A as it survives was copied from B as it survives, or vice versa, the existence of X can be taken for granted straight away. The question is whether X was A-like or B-like.

26

The Canterbury text is careful to state that the farm is in the process of being increased: in the last complete financial year Sandwich paid 50 pounds, but in the current year it is due to pay 70 pounds. Though it has dropped out of the excerpt made for Saint Augustine’s (which gives only the current figure), a similar statement must have been included in B-Ke, and eventually also in D-Ke. The DB scribe, in his version of this

112

The fieldwork phase Canterbury, D & C, Lit. E 28, fo. 3vb–c

PRO, E 164/27, fo. 21r–v

Sandwic est manerium sc’˛e trinitatis, et de uestitu monachorum, et est læth et hundretus in se ipso, et reddit regi seruitium in mare sicut douera, et homines illius uill˛e antequam rex eis dedisset suas consuetudines, reddebant xv lib’. Quando archiep’s recuperauit, reddebat xl lib’ et xl milia de allecibus. Et in pr˛eterito anno reddidit l lib’ et allecia sicut prius. Et in isto anno debet reddere lx et x lib’ et allecia sicut prius. In tempore E regis erant ibi ccc et vii mansur˛e. Nunc autem lx et xvi plus.

Sandwich burgum Sc’e Trinitatis est de uestura monachorum, et est hundred in se ipso, et reddit regi seruicium in mare sicut illi de Doura, et homines illius uille, antequam eis rex dedisset suas consuetudines, reddebant xv li’. Et quando archiep’c recuperauit, reddebant xl li’ et xl mil’ de allecibus, modo uero debent reddere lxx li’ et alleces sicut prius. Tempore regis Edwardi erant ibi ccc et vii mansure, modo sunt lxxvi plus. In isto burgo habet Sc’s August’ unum agrum, et ibi sunt xxx mansure que reddunt monachis iiii mil’ de allecibus uel x s’, et regi faciunt seruicium in mare si[cut] alii. In isto agro habet eciam Sc’s August’ unam eccl’iam.

Table 35. Two descriptions of the town of Sandwich. What this text represents, therefore, is an outline description of the manors in Kent belonging to the archbishopric, compiled on Lanfranc’s behalf by one of his administrative assistants, in response to a series of questions asked by the commissioners conducting the survey of this county. At least one copy of this text was in Lanfranc’s possession at the time of his death: perhaps a duplicate had been kept on file; perhaps the original had been returned to him, after the commissioners were finished with it.28 What survives is a copy of that copy.

If it could be assumed (as it evidently could be, in the case of archbishop Lanfranc) that the owner of every manor would have this much information instantly to hand, a questionnaire of this kind would have been an effective method for assembling an early version of the B text, into which the additional data which the commissioners were planning to collect could be inserted as it became available.29 To what extent this method was actually used – that is another question. We have already constructed a list of the questions which shaped the survey of Cambridgeshire, as it is reflected in B-Ca (Table 27). If we subtract from this the questions which Lanfranc has already answered for us (and two questions at the end which belong to stage 2), what we are left with is this:

It is possible – not impossible – that the same questionnaire was circulated among all the barons in Kent who were known or assumed to be holding directly from the king. If so, the archbishop’s response would probably have been atypical, because the manors belonging to him had mostly been in his church’s uninterrupted possession since the time of king Edward. But there were some exceptions. Taking note of the way in which these exceptions are dealt with here, we can reconstruct the questionnaire in a form which could have been sent out to any baron, not just in Kent but (with one or two words changed) in any other county. It would have looked something like this:

for each manor how many ploughs does the land suffice for? how many ploughs on the domain? how many ploughs for the villains? how many hides of domain? if there are fewer ploughs than the land suffices for how many more might be made on the domain? how many more for the villains? how many villains, bordars, slaves? anybody else worth mentioning? how many mills? and what are they worth? any other assets worth mentioning? how much meadow? how much pasture? how much livestock -- cows, sheep, pigs, horses -- on the domain? how much was the manor worth when the man who holds it got possession? how much was it worth TRE? next manor

for each manor what is its name? who held it in the time of king Edward? from whom? who holds it now? from whom? how many sulungs did it defend itself for in the time of king Edward? how many now? how much it is worth? if any of your men possess parts of this manor for each man how much does he hold? how much is it worth? next man next manor

These were the facts which remained to be discovered, by whatever means might be employed, before the survey of

PS - remember to say plainly which hundred each manor belongs to

29

It is a point to note that – except for special cases like Sandwich – the only value reported is the current value. Apparently that was the only figure being asked for at this stage.

28

There is some slight evidence (above, note 20) that more than one copy of this text found its way into the Christ Church archive.

113

The survey of the whole of England cludes with some more polite remarks.

any given county could be brought to completion. And these are the proceedings which were witnessed, somewhere in the country, by the English chronicler; but he was more interested in telling us what a disgraceful business it was than in telling us how it was managed.

It is difficult to draw any definite conclusions from this. Altogether there were nine counties in which the archbishop held land (Du Boulay 1966, pp. 43–6), and (since Lanfranc speaks in the plural) two or more of these must have fallen within S ’s ambit. From the schedule of farms (above, p. 111), we know that the archbishop’s domain manors were distributed over four counties – Kent, Surrey, Middlesex, Sussex – and none of those can be in question; Hertfordshire is also excluded, because none of the manors here belonged to the monks.32 That leaves us with four counties where the monks owned land but the archbishop and his knights did not: Buckinghamshire, Oxfordshire, Essex, Suffolk. Two or more of these were the counties where S ’s interests intersected with Lanfranc’s. Taking a gamble, perhaps we might think of narrowing the choice to one or the other contiguous pair of counties; and in that case the better bet, it seems to me, would be Buckinghamshire and Oxfordshire.33 There were three Canterbury manors here, and all three appear in the list of properties which Lanfranc had recovered for his church;34 so possibly there might still have been some doubt in people’s minds as to what Lanfranc had done with them. In Essex and Suffolk, by contrast, the monks’ manors had been continuously in their possession (apart from some small encroachments) since the time of king Edward, and it seems unlikely that anyone would have been uncertain about their status. This does not get us any closer to deciding who S might have been, but we cannot expect an answer to every question. Nor can we be certain beyond all doubt that S is the right initial: one capital letter, out of context, may easily be miscopied.35

Away from Kent, we have evidence of the archbishop’s involvement with the survey at a slightly later stage in its proceedings. While the B text was still in a plastic state – still open to last-minute corrections – a letter was delivered to inviting him to comLanfranc from someone named S ment on a point which, in the B text as it stood, was not as clearly stated as might perhaps be wished. Again we have no idea whether this letter was unique or nearly so, or whether the commissioners were sending out letters like this in large numbers. The only sure fact is that none survive, not even S ’s letter. All that we have is Lanfranc’s reply, which (somehow or other) came to be included in a rather meagre collection of his correspondence put together by one of his admirers shortly after his death (ed. Clover and Gibson 1979). Whoever he was, the compiler was so short of material that he could not afford to pick and choose: he included even letters like this one, so brief and so lacking in context that they do not make much sense. The significance of this particular letter (available in print since the seventeenth century) seems to have gone unremarked until Barlow (1963) pointed out that it contains an explicit reference to the survey of 1086. Something of the gist of S ’s letter can be got from Lanfranc’s reply. (A messenger is passing to and fro, but his role in the business is obscure.) S is responsible for the survey of some number of counties – presumably he would have said which counties they were, but Lanfranc does not bother to repeat the names – including some in which Lanfranc’s church owns land. The question he asks is: which of these manors does Lanfranc hold in domain? (Though no doubt it was delicately worded, the point of the question would be that the domain manors are the ones which will fall into the king’s hands, when Lanfranc goes the way of all flesh.) As it stands now, says S , the record does not state which manors belong to the archbishop himself and which belong to his monks. But S offers to have the text amended, to make this distinction clear, if Lanfranc would like him to do so.

From the evidence of the surviving geld accounts, we can be sure that the second instalment of the current geld fell due at a time intermediate between stage 1 and stage 2 of text is printed by Barlow (1963, p. 289) and, with the rest of the collection, by Clover and Gibson (1979, p. 170). 32 The land in Hertfordshire was (except for two acres) all held from the archbishop by Anskitil (de Ros); some of it was claimed by Westminster, some of it by Saint Alban’s (DB-Ht-133rb). 33

But in this case it would have to be assumed that S failed to keep his promise: if DB is an accurate proxy for B, the record was not amended. (That is why Barlow (1963) was deceived into thinking that only Essex and Suffolk would fit the bill. He could not do what we can: consult Du Boulay (1966).) To my mind, this is not an obstacle: reasons can easily be imagined why the opportunity might in the end have been missed. But the only S to be met with in these two counties is Suain the sheriff, who held one manor in Oxfordshire, not in domain (DB-Ox-160ra).

Lanfranc’s reply is short and to the point – as short as it could be without being discourteous.30 He thanks S for his concern. He answers the question that was put to him: in the counties which S is responsible for investigating, none of the Canterbury lands are held in domain, all are assigned to the maintenance of the monks.31 And he con30

34 The manors recovered by Lanfranc, with the king’s support, are conveniently listed, county by county, in the Christ Church obit of Willelm I (Le Patourel 1948, pp. 24–6). Haddenham (DB-Bu-143vb) is in a category by itself. This was Lanfranc’s personal property, subsequently given by him to the monks of Rochester (Flight 1997, pp. 12–13).

Barlow detected some fearfulness in Lanfranc’s letter; I do not.

35 If we ventured to think that ‘S’ might be wrong, perhaps a mistake for ‘G’, we would face an embarrassment of possibilities. Since the name Walter was often spelt Gualterus, Lanfranc’s correspondent might even be Walter Gifard.

31

Scias autem in illis comitatibus, quorum exquirendorum tibi cura commissa est, me nichil in dominio habere, sed omnes in illis partibus nostrae aecclesiae terras ad uictum monachorum per omnia pertinere. The full

114

The fieldwork phase the survey. As stage 1 drew to an end, the B text for each county was delivered to the treasury; there it was checked against the geld account; and the result of that checking was a list of queries which the treasury expected to be answered in stage 2. For four counties, the version of the geld account which registers this checking survives in the original (above, pp. 61–6): they are largely the work of scribe alpha (the treasurer’s clerk, as I suppose him to have been).

missioners had some good excuse, perhaps the king would not be unforgiving. But the commissioners were, we may be sure, expected to make every effort to deliver their reports on time, especially for those counties which were first in line to be dealt with in stage 2. If these were delayed, the whole programme might be derailed. The king was at Winchester for Easter: we know this because the English chronicler tells us so. He says nothing whatever that implies a connection between this meeting and the progress of the survey. All the same it seems likely that the Easter court provided an opportunity for the king and his council to review the results from stage 1 before finalizing the arrangements for stage 2.

It is not known when exactly the geld instalments were due, or whether the dates were the same for every county.36 The best evidence comes from Dorset, where scribe alpha records a number of late payments: though the treasury is now in possession of most of this money, he assumes that the commissioners will impose some penalty for the fact that it was not paid on time. There were, it is clear, two ‘appointed terms’, and in some cases the treasury’s complaint is merely that money which ought to have been paid at the first term was not paid until the second. It was presumably a more serious matter that some money had gone unpaid at the second term. The men of Roger de Bello monte have paid 1260 pence for 17.5 hides, but they did not pay till ‘after the feast of Saint Mary’.37 One step worse than that was for the payment not to be made till ‘after Easter’; and worst of all was the case of Rotbert de Oilleio, who ‘withheld’ a sum of money till ‘after Easter’ and has still not paid it even now.38 This is good evidence, as far as it goes; but it is very thin, and in some respects not of much help. (We are given no clue as to the date of the first term, but that need not concern us.) Even so, I think it safe to say that in Dorset, and possibly everywhere, the second term ended on Lady Day (25 March), and that a short period followed after that (ending this year on 5 April) during which one could pay the money (as Roger de Belmont’s men did) and hope not to incur the full penalty for late payment.

As soon as he had the B text and the geld account, scribe alpha set to work compiling his list of queries for each county in turn. For him too, the urgency would be greatest for the counties which were going to be first in line; once stage 2 was safely under way, he could perhaps afford to deal with the remaining counties in a more leisurely fashion. His list of queries for Dorset was clearly not compiled till Easter was well in the past; and corrections and additions continued being made to it, over some period of time. The final corrections may have been made at the very last minute, perhaps as late as June or July, just before a copy of this text was sent off to the commissioners responsible for this county.

2 Stage 2 of the survey is a little less obscure than stage 1. As we know from bishop Rotbert’s account, the basic intention was for each county to be visited by a second team of commissioners, disinterested parties who would monitor the first team’s work and report whatever derelictions they might find to the king. In one county – Worcestershire – we know that the team consisted of a bishop and three other men; all four names are recorded.

The same deadline, so I am willing to guess, applied to stage 1 of the survey. The commissioners appointed at Christmas would have been told told, before they left Gloucester, that they had exactly three months to get the job done: the finished report for every county was to be handed in to the treasury no later than 25 March. In the nature of the case it would not be surprising if a few reports missed the deadline; provided the delay was slight, provided the com-

The story goes something like this. Upon their arrival in Worcester, the commissioners discovered that bishop Wulstan (who had possibly offered them accommodation and would certainly have invited them to dinner) was intending to recruit them into a scheme of his own.39 One matter much on his mind was an unresolved dispute with the abbot of Evesham (the details of which we may ignore). There had already been a trial, some years earlier, and Wulstan could produce two writs connected with that: a writ ordering the trial to be held, a subsequent writ approving of what had been decided.40 But he had still not been able to make the abbot comply. Once Wulstan knew who the commissioners for Worcestershire were going to be, he sent a

36

As far as Berkshire is concerned, there is an explicit statement in DB that the geld was payable in two equal instalments of 3.5 pence, at Christmas and at Whitsun (DB-Be-56va). But the very fact that this information was recorded suggests that it was anomalous; furthermore it is in the past tense. It seems that the men of Berkshire were making a protest: in view of his promise to maintain the laws of king Edward, the king should not demand more than 7 pence a hide, and should let it be paid at the traditional dates. The chances that the king listened to this complaint are small. 37

Et pro x et vii hidis et dimidia reddiderunt homines Rogerii de bello monte c et v solidos post festum sanctae Mariae (ch-Do-20v). The place in question was assessed at 30 hides, of which 12.5 hides were deductible as domain (DB-Do-80rb).

39

Bishop Rotbert of Hereford owned some land in Worcestershire (DBWo-174rb) and is said to have been a close friend of Wulstan’s; but his name does not appear here.

38

Sed Rotbertus de oilleio retinuit inde xv solidos usque post pascha, with the interlined note quos nondum habet rex (ch-Do-19v). As is clear from the geld account (but not from DB), Rotbert was farming some of the land which had fallen into the king’s hands.

40

The two writs are printed by Bates (1998, nos. 347–8), and by Thorn and Thorn (1982, app. V, ‘Worcester H’, nos. 1, 3).

115

The survey of the whole of England messenger in search of the bishop of Coutances, who had presided over that trial, asking him to write a letter setting out the facts; and the bishop (who was probably somewhere in England at the time) obliged.41 Armed with these documents, Wulstan made the commissioners his accomplices in ambushing the abbot of Evesham. When the abbot arrived, he found himself caught in a situation where he could not avoid coming to terms with the bishop. In the presence of the monks of Worcester, of some of the monks of Evesham (presumably just those who were accompanying the abbot), and of the four commissioners for Worcestershire (regis principibus qui uenerant ad inquirendas terras comitatus), a new document was drawn up to celebrate the abbot’s discomfiture.42 The list of witnesses is headed by the abbot of Gloucester, another visitor coopted into Wulstan’s scheme;43 it also includes a monk of Saint-Rémi, presumably acting as his monastery’s proctor in connection with the survey.44

Thus we have three sources of information. Two are strictly contemporary: a letter from the bishop of Coutances to the commissioners for Worcestershire, and the agreement drawn up in the commissioners’ presence between the bishop of Worcester and the abbot of Evesham. The third is only slightly later: Hemming’s commentary on the Oswaldslow text. It would have been helpful if one of these documents had given us a date or a hint of a date; unfortunately none of them does that. But at least we can be perfectly sure who the commissioners were: Remigius bishop of Lincoln,47 Henric de Ferieres, Walter Gifard,48 Adam. None of them owned land in this county: they would not have been chosen if they had. Henric and Walter were important barons, just the sort of men whom we would expect to find entrusted with important tasks like this. The last name is more of a surprise. Hemming tells us, no doubt rightly, that this Adam is Eudo Dapifer’s brother – i.e. the same man who occurs in DB as Adam son of Hubert – and he was only a second-tier baron, a tenant of the bishop of Bayeux.49 Still, with the bishop in prison, Adam was a firsttier baron for the time being; and through his brother he had a personal connection with the king’s court.

This dispute with the abbot of Evesham overlapped with a larger question, the extent of the liberties enjoyed by the church of Worcester in the triple hundred of Oswaldslow. At one of the formal sessions convened by the commissioners, this question was brought up and a form of words was agreed: the bishop was satisfied with it, the county was ready to swear to it, and the commissioners were willing to include it in their report to the king.45 They also allowed the bishop to make a copy for himself – and a copy of that, some years later (after but not long after the death of Willelm I), was included by the church’s archivist, a monk named Hemming, in a compilation of documents concerning the possessions of the church of Worcester (BL, Cotton Tib. A. xiii, fos. 119–34). This particular document comes right at the end (fo. 133r–v), followed only by an explanatory paragraph contributed by Hemming himself.46

From what we know about Worcestershire, it does not seem rash to infer that in every county the second team consisted of four men – one bishop and three barons. If we can trust a vague remark by Hemming, the team which dealt with Worcestershire had also dealt or was about to deal with ‘a number of other’ counties (in hac prouincia et in pluribus aliis). There are indications elsewhere that it was indeed the rule that the second team (unlike the first one) should be headed by a bishop – that is, by someone who, as well as being a baron, was also literate in Latin, and could answer for the accuracy of the written record. In Wiltshire the second team appears in the geld accounts as ‘bishop Willelm and his colleagues’ (above, p. 67): this Willelm is, almost certainly, the bishop of Durham, who appears to have been connected somehow with the survey of Somerset too (above, pp. 71–3). More doubtfully, there is one item of evidence (below, p. 121) suggesting that the bishop of Winchester and the bishop of Coutances may each have dealt with one or more of the counties where the monks of Ely owned land, perhaps even (if we argue from silence) that between

41

This letter is printed by Bates (1998, no. 350), and by Thorn and Thorn (1982, app. V, ‘Worcester H’, no. 4). There is evidence which seems to suggest that bishop Goisfrid led the team which visited one or more of the counties where the abbot of Ely owned land (below, p. 121). 42

This agreement between bishop Wulstan and abbot Walter is printed by Thorn and Thorn (1982, app. V, ‘Worcester H’, no. 5).

43

The abbey of Gloucester owned very little property in Worcestershire – just half a hide in Droitwich (DB-Wo-174rb) – but apparently that was enough reason for abbot Serlo to come to Worcester. 44

His name was Alfwin. Saint-Rémi owned nothing in Worcestershire, but it did possess lands in Staffordshire (DB-Nn-222vb, DB-St-247va) and (one hide only) in Shropshire (DB-Sh-252rb).

45

pp. 57–8) is Hemming. Ker seems to be working towards that conclusion himself, but eventually veers away from it (p. 72), for reasons which do not seem good to me.)

46

47 The bishop’s retinue included two monks (named Ulf and Rannulf) and one clerk (named Nigel), all of whom witnessed the agreement mentioned above (note 42).

Printed by Thorn and Thorn (1982, app. V, ‘Worcester F’). The version appearing in DB-Wo (172va) is somewhat shorter, but much of it is word for word the same. Hoc testimonium totus uicecomitatus uuireceastre dato sacramento iurisiurandi firmauit . . . tempore regis Willelmi senioris, coram principibus eiusdem regis . . . qui ad inquirendas et describendas possessiones et consuetudines tam regis quam principum suorum in hac prouincia et in pluribus aliis ab ipso rege destinati sunt, eo tempore quo totam Angliam idem rex describi fecit. As I understand him, Hemming is saying that this text was written down in an annex added to B-Wo (the autentica regis cartula, as he calls it, using the same phrase twice) and that the annex is (as he knows or supposes) still kept in the king’s treasury with the records of the survey (quae in thesauro regali cum totius Angliae descriptionibus conseruatur). (I am assuming that the scribe called ‘hand 1’ by Ker (1948,

48

Henric is named before Walter in the agreement. The bishop of Coutances reverses the order, and Hemming does the same. Hemming also calls Walter ‘earl’: presumably this was true at the time when he was writing, but it was not true at the time of the survey. 49

His holding was mostly in Kent, but he also occurs in Surrey and Oxfordshire (DB-Sy-31vb, DB-Ox-156rb). (The Adam who occurs in Hertfordshire, thought by Farrer (1925, pp. 293–4) to be the same man, was not: the Ely records call him Adam son of Willelm.)

116

The fieldwork phase them they dealt with all six of these counties.

hundred juries were being shunted in and shunted out with the least possible delay (below, p. 119). A small county, such as Hertfordshire, could quite possibly have been dealt with in one day. An unusually large county, such as Yorkshire, might take more than three days. On the average, however, three days per county might be enough. But this makes no allowance for travel time. If a single team were employed, it would lose a large amount of time transporting itself from one county to the next, on top of the time that it spent conducting business. And that would take too long.

If all of this is right, we have evidence for the involvement of at least four bishops (three English, one Norman) in stage 2 of the survey. Multiplying by three, we may guess that at least twelve barons would have been involved. These are the smallest numbers that seem at all likely; if we double them, the numbers that we get – eight bishops, 24 barons – are probably about the largest numbers that we would be willing to consider.50 With 32 commissioners and 32 counties, each commissioner (if the workload were evenly distributed) would have to deal with four counties: if we decrease the number of commissioners, the number of counties with which each will have to deal increases. If there were only 16 commissioners (the number that we started with), each would have to deal with eight counties, more or less.51

For other reasons too, it is clear that no single team of four commissioners could possibly do the job. Whoever they were, sooner or later they would come to a county where at least of them had to recuse himself, because he was an interested party. Men who were important enough to serve as commissioners – men like the ones whom we have met in Worcestershire – were likely to own land in several counties. Suppose that we have chosen a team to conduct stage 2 of the survey of Worcestershire: the bishop of Lincoln, Henric de Ferieres, Walter Gifard, Adam son of Hubert. There are several counties – though only one adjoining county (Shropshire) – into which we could send the same team.53 But most counties are precluded.54 We could send the bishop of Lincoln into Gloucestershire; but if we did, we should have to find a replacement for Henric de Ferieres. We could send the bishop of Lincoln into Wiltshire; but if we did, we should also have to find a replacement for Walter Gifard.

That some division of labour must have been involved seems clear enough from scheduling considerations. Stage 2 did not begin till after Easter; it was completed before the end of July (by which time the C text for every county was already in existence). If a single team were employed, that would imply an average allowance of no more than three days per county. From one angle, that number does not seem unreasonable to me. I do not doubt but that the commissioners transacted their business at a speed which would seem astonishing if one thought of comparing it (as Eyton did) with the proceedings of the itinerant justices of Henric II – still more so if one thought of comparing it (as Maitland did) with the proceedings of the itinerant justices of Henric III. Those analogies seem inappropriate to me. In the late eleventh century, a three-day meeting was a very long meeting.52 We can take it for granted, I think, that the

At the same time, the bishop of Lincoln has duties towards his church which are not superseded by his duties towards the king. On his own account – as bishop, not as commissioner – he is sure to want to attend the meeting in Lincolnshire, and probably the meetings in all the other nine counties where his church owns land. It is not clear that his attendance is demanded; but we cannot reasonably deny him the opportunity to be present at any meeting which affects his church’s interests. Similarly, we cannot think of making it impossible for Henric de Ferieres to attend the meeting in Staffordshire, or for Walter Gifard to attend the meeting in Buckinghamshire, or for Adam son of Hubert to attend the meeting in Kent. They also have interests to protect, for themselves and their heirs. Such concerns may not be as exalted as the duty owed by a bishop to his church, but we cannot expect a baron to disregard them. If this were the thirteenth century, such men would have lawyers to represent them. In the eleventh century they expect and are ex-

50 A conjecture of Eyton’s (1877, pp. 106–9) implied that 36 commissioners were employed; Ballard (1906, pp. 12–13) thought that 28 might be enough. As far as I am aware, no one has found these numbers inconceivably too large or too small. The reader should understand that I have nothing to say about Eyton’s conjecture, beyond what is said between the lines of the present section. (I am not even sure whether it should be taken to refer to stage 1 or stage 2.) In fairness, however, I ought to add that Eyton is not to blame for the harm which his conjecture has done. It was Galbraith (1942) who took a casual suggestion and – seemingly without any serious thought – made an axiom out of it. It was other historians who allowed this axiom to take hold, to the point that any account of the survey had to begin with an intricately detailed mapping of Eyton’s conjecture, as it was modified, on a mere whim, by Ballard. There is, I gather, an unpublished essay on this subject among Eyton’s papers: anyone who thinks that the conjecture is worth taking seriously might do him and the rest of us the courtesy of starting with that. (Historians who cite an article of Stephenson’s (1947) as proof that Eyton was right are admitting one of two things, either that they have not read the article for themselves, or that they have failed to understand it.)

Patourel 1948, p. 22). I note, by the way, that the only reliable account of what happened on Penenden Heath is the document printed (but misinterpreted) by Douglas (1933, pp. 51–2).

51

These are only order-of-magnitude estimates, not to be taken literally. It is unlikely that the load was evenly distributed. A more realistic guess might be that there were (bishops included) 20–24 men who visited five or six counties each and 4–8 men who visited one or two counties each, where and when they were needed to make up the number.

53

These four would be eligible to serve in Yorkshire, Middlesex, Hertfordshire, Sussex, Hampshire, Worcestershire, Shropshire, Cheshire, Dorset, Devon, Cornwall. Choose any four barons – and the chances are that the allowable counties will be patchily distributed like this.

52 The author who wrote a fictionalized account of the meeting on Penenden Heath in 1072 expected us to be impressed with the fact that the meeting had to last for more than one day – in fact for as much as three days: Et quoniam multa placita . . . ibi surrexerunt . . . quae prima die expediri non potuerunt, ea causa totus comitatus per tres dies fuit ibi detentus (Le

54 It is to Eyton’s credit that he recognized this difficulty (Eyton 1877, p. 108), less to his credit that he promptly wriggled out of it. Like everyone else at the time, he was ignorant of bishop Rotbert’s account of the survey, and for him that is an allowable excuse.

117

The survey of the whole of England the DB scribe was following (nearly) the order set by the D scribes; they had been following (nearly) the order set by the C scribes; and they had been following (nearly) the order set by the arrival of the B texts. The order in which the DB booklets were written was approximately the same – not exactly but nearly the same – as the order in which the counties had been dealt with in stage 2. How far the orders differed in detail is more than we can say; but I think we can be sure that the juggernaut began its journey in the far north and ended it in the far south-west.

pected to speak for themselves. But they are also the only men who can be trusted to conduct the survey. If this were the thirteenth century, we could send full-time judges or career officials to carry out a job of this kind. In the eleventh century, we have to rely on the barons; and that puts us into a dilemma. It means that we have to find some way of reconciling the duty which a baron owes to the king with the duty which he owes to himself. The only acceptable solution, as far as I can see, is for the meetings to be scheduled to occur in rapid succession, without actually overlapping to any significant extent. Things have to be arranged in such a way that we find ourselves, on any given day, in this sort of situation. One team of commissioners is wrapping up its business in county A: the formal sessions are over, but some work still needs to be done on the final report. A second team has just started its proceedings in county B, and will be busy for a few days yet. A third team is already on the spot in county C, talking matters over with the sheriff, making sure that all necessary preparations have been made. And the individuals who will make up a fourth team are already, separately or together, on their way to county D. In a few days’ time, the juggernaut will have moved on, and the commissioners who were at work in county A will be mobile again. One or more may have to hurry off to another county where the king requires his services; one or more may have to hurry off to a meeting which he has reasons of his own for attending; and anyone who does not have urgent business can relax, at least for a while.

What this map brings to mind, I think, is a map of a military campaign – perhaps an advance through enemy territory, where the castles need to be cut off and captured one by one. The man who planned this campaign was a man experienced in warfare, a man accustomed to issuing orders – accustomed also to having his orders obeyed, instantly and without question. There was no ‘man behind the survey’, no man in the shadows acting as the king’s alter ego. The king himself was in command.

The commissioners are not the sort of men who travel alone. They need servants with them, to look after themselves and their horses; they need an armed escort. Upon their arrival, they and their retinue will require accommodation, food and drink, fodder. All this, I suppose, was arranged by the sheriff, at the king’s expense. Perhaps the sheriff had to account for his expenditure; but no such accounts survive. If the commissioners had found some cause for complaint, with respect to these or other advance arrangements, we might learn what ought to have been done by hearing what had failed to be done; but no such complaints are on record. There is no hint that any sheriff showed himself uncooperative or incompetent.

It needs to be stressed that the logistical constraints which came into play in stage 2 were very different from what they had been in stage 1. There was no room for improvisation: the schedule had to be decided in advance and stuck to. Coordinating the movements of the commissioners – a small number of men who can be trusted to do as they are told – is not the difficult part. It is child’s play compared with the job of mobilizing the local juries. About that I say something below; I mention it here only to make the point that the date of each meeting had to be fixed several weeks in advance. If Willelm I’s chancery had kept copies of outgoing letters, we should probably be able to map out the advance of the juggernaut day by day. There would be letters addressed to the men of B shire, letters addressed to the sheriff of B shire, letters addressed to the individual commissioners; and some of these letters would be sure to tell us when and where the meeting was due to take place. That the chancery was issuing such letters does not seem doubtful to me; but no copies were retained, and the originals had no permanent value which would have made them worth preserving.

A team of three scribes, sent out from the treasury, arrives to meet up with the commissioners; they too will need board and lodging. They bring with them the B text for this county, and also at least one other piece of documentation – a checklist of the discrepancies between the B text and the geld account, such as survives in the original for four counties (above, pp. 61–6). The treasury has discovered these discrepancies; the commissioners are to investigate them. The commissioners and the scribes are both small groups, and their movements would be relatively easy to coordinate. Logistically it would be a much larger problem to bring together all the local participants. This, it seems, must all have been the sheriff’s responsibility. Once the proceedings began, he was himself an essential participant; but it was also his job, sufficiently far in advance, to make sure that the word was spread, that juries were selected from every hundred, and that everyone knew exactly where and when they should assemble.

Even so, the path of the juggernaut is approximately known (Fig. 12). In his imaginary journey around the country,55 55

Large numbers of people are involved. In Cambridgeshire, the county of which we know most, there were fifteen hun-

The sequence of DB booklets shown here is slightly different from that which I suggested originally (Table 3). Looking at the evidence again, I have changed my mind about Cambridgeshire (below, p. 141).

118

The fieldwork phase START

Yo 













Ch







Nm

Dy



Li

















St





Sh



Le










Nk



Hu











Wo

Nn 

Wa

He

Ca




Bd 











Sk



Ox









Bu 

Gl

Be



Wi 



















Ha

 





Mx Sy













Dn

Ex














So

Ht 





Do

Ke





Sx





Co





FINISH

Figure 12. The sequence of DB booklets construed as an approximate map of the progress of stage 2 of the survey. bring their retinues with them.57

dreds, one of which was counted as a double hundred. Each was to be represented by a jury of eight men – four Frenchmen, four (French-speaking) Englishmen – and twice that many for the double hundred; so already we have 128 people to organize.56 Some counties had many more hundreds than this. In Wiltshire, for example, there were 40 hundreds, and so more than 300 jurors had to be brought to the meeting. However many of them there were, their attendance was compulsory, and we can be sure that anyone who failed to show up would have been in very serious trouble.

Once everyone has been brought together – commissioners, scribes, jurors, barons, sheriff – what happens next? How are the proceedings conducted? Suppose that the meeting has just finished dealing with one hundred and is ready to start dealing with the next one. The jurors for this hundred have already been marshalled by the sheriff’s officers: now they are brought forward, and their names are called out and written down. It is necessary for everyone to know who they are; it is necessary for the jurors to know that their names are being recorded. They have, I suppose, been thoroughly coached in advance. The section of the B text to which they will have to swear has already been read out to them (in French), behind the scenes, by the scribe who is managing the rehearsal, so that they can settle any doubts or disagreements among themselves before they appear in court.58 We do not want to waste time, in the formal sessions, watching the jurors squabble. Now they are put on

The king’s barons were present too. It is not clear that they were under any obligation to attend; their absence, whether voluntary or involuntary, was not going to bring the proceedings to a halt. But they had to be notified, and they had to be given the chance to attend, if they wished to do so. Probably most of them did wish to, and did attend, especially if they saw any risk of their rights being called into question by someone else. In Cambridgeshire there were about 40 people who would be entitled to attend because they were the king’s barons. Again in some counties the number was much larger than this – more than 150 in Wiltshire. Perhaps the sheriff might have a word with some of the smaller people, letting them know that the king would not be offended if they found themselves unable to come. But still there are numerous important people – bishops and abbots, earls and barons – who cannot possibly be discouraged from participating; and they, when they come, will

57 Not all of them were men. In Cambridgeshire, for example, the list includes the abbess of Chatteris, countess Judita, and two other women. As far as I can see, there is no clear proof that any woman attended in person. 58 Here we can find an explanation for the fact that the team consists of three scribes. If rehearsal takes longer than performance, as it presumably does, more than two scribes will be needed to keep things moving. If rehearsal is not allowed to take more than twice as long, three scribes will be enough. At any moment, one scribe will be servicing the formal session, one will be ending his rehearsal of the hundred which is next in line, and one will be beginning his rehearsal of the hundred after that.

56

It is possible in one or two cases that the same man may have served on two juries. I am not convinced that this was so, and in any case it does not make much difference.

119

The survey of the whole of England their oath to do their duty by God and the king. The relevant section of the B text is then read out again aloud (again in French), so that everyone present will know exactly what the sworn record of the survey is going to say. We cannot let anyone have the chance to plead ignorance, after the event.

them the edited version of the B text. The commissioners and the other barons head off in various directions, wherever the king’s orders or their own interests may take them next. And the sheriff breathes a sigh of relief – unless he, too, is hurrying off to another meeting.

By and large, it was being assumed that the first team of commissioners had done their job carefully and conscientiously, and the jurors were not expected to have much to say. As a matter of course, they would have to be asked whether they were aware of any mistakes in the text which ought to be put right, or of any facts, omitted from the text as it stood, which the king ought to know about. If their hundred included any land of the king’s, they would be required to state what encroachments, if any, they knew about: where the king was concerned, some fair amount of detail might be demanded. The commissioners would have some questions of their own to put to the jurors – some which arose on the spot, some which the treasury had told them to ask, because of conflicting entries in the geld account. That was the main business of the meeting, transacted between the commissioners and the jurors. But there must also have been some opportunity for other people present to put in a claim or make a protest on their own account. The commissioners, it is clear, were under instructions to take note of all cases of disputed possession, and to get some preliminary statement of the facts from the local jury. They were not empowered to settle disputes, but they were required to make a record of them; and that record, once made, was intended to be definitive. We can be sure, I think, that anyone who failed to speak up now would have little hope of being listened to in the future. After that, the jurors were thanked and allowed to depart, and the commissioners (unless they needed a few minutes’ recess) were ready to deal with the next hundred straight away.

We know little about the aftermath. One of the documents which was on its way to Winchester was a new version of the geld account – such as survives in the original for one county (above, pp. 67–8) – replying to the treasury’s questions. Prompted by the treasury, the commissioners had discovered numerous cases of geld evasion. Quite frequently it turned out that people had been failing to pay all or some of the geld which was due from land that they owned – by falsely claiming to hold it directly from the king, by falsely claiming to hold it in domain, sometimes simply by not admitting its existence. The peasants had paid; when did peasants ever have any choice but to pay? But the lord of the place, instead of forwarding the money to the collectors, had kept it for himself. In due course justice would have to be done; but the missing money could be collected at once. From a cryptic remark by bishop Rotbert (above, p. 108), it seems that the sheriff was left with instructions to take whatever action might be needed to get hold of this money – from the owner if possible, from his unlucky peasants (who had already paid once) if necessary. There is one other point which I will mention but not pursue. Given that their time was short and strictly limited, it seems likely that the commissioners would occasionally come across some problem which they were unable to resolve satisfactorily. (Perhaps the original investigators had botched some part of their report; perhaps the local jurors were raising questions which could not be answered on the spot.) If this happened, for whatever reason, the commissioners would presumably notify the king, make their excuses, and recommend that some further investigation should be made, concentrated on this single topic. To what extent was the work of the survey continued and completed by special-purpose inquiries? Was there a stage 3? I ask the question without trying to answer it. But I note that two of the surviving D booklets each end with a block of text which is not listed in the index – in one case an astonishingly detailed survey of the town of Colchester (D-Ex104r–7v), in the other a list of the lands in dispute between the bishop of Bayeux and Robert Malet’s mother (D-Sk450r).

After the last hundred jury had been discharged, there was still some further business to be transacted. In every county, there was at least one town of which some sort of survey was needed, and apparently a jury of the men of the town was made to swear to that. In every county again, some statement was recorded of any customary arrangements existing in the time of king Edward which ought to be of profit to the present king, and a county jury was required to swear to that. The indications are that matters of this kind were recorded in annexes to the B text; but these annexes, because they fell outside the feodal frame, were roughly handled during the compilation phase (the DB scribe is not alone to blame), and it is sometimes difficult to make much sense of them, in their surviving form. In the main text, the compilation phase did not do so much damage. Even in counties for which nothing but DB is available, those passages which originated in stage 2 of the survey are generally quite easy to recognize, once one has learnt what clues are to be looked for.

From Easter onwards, the king’s movements are recorded (but only vaguely recorded) by the English chronicler. From Winchester, we are told, the king ‘travelled so that he was at Westminster’ for Whitsun (24 May), and after that ‘he travelled about so that he came to Salisbury’ for Lammas (1 August). It was normal for the court to meet at Whitsun, but a meeting at Lammas is something out of the ordinary. The chronicler says nothing to suggest how these movements might have been linked with the survey: in fact

Then finally the meeting is over. The local people disperse, variously happy or unhappy with the way that things have worked out. The scribes return to Winchester, taking with

120

The fieldwork phase he says nothing at all about this second stage. Apparently the king made a point of letting himself be seen; but he did not interfere with the work of his commissioners. He did not attend any of the county meetings: if he had done so, we should know it. By Whitsun (seven weeks after Easter) the survey would be roughly at the halfway mark, and the juggernaut, it seems, would have been somewhere in southeastern England at the time. By the beginning of August the survey was already completed, and the meeting which brought the king to Salisbury (and which must have been planned many weeks in advance) was presumably intended to be seen as its culmination. If the chronicler had understood the purpose of this meeting correctly – his account of it has already been quoted (above, pp. 79–80) – the king was asserting his right to the personal loyalty of all his English subjects, a loyalty which transcended the whole hierarchy of feodal relationships mapped out in the written record of the survey. After this, presumably as soon as the meeting was over, ‘he travelled into Wight because he wanted to go into Normandy’; having stayed in Wight for some time, perhaps only while he was waiting for a favourable wind, ‘he travelled into Normandy’. Though nobody knew it at the time, though the chronicler still did not know it when he wrote these words, the king would never be seen in England again.

repair (again), and the people who ought to carry out these repairs were dragging their feet (again). Whoever these people were, it would have suited them if the abbot, losing patience, had done the work himself; but abbot Simeon had no intention of falling into that trap. Rather than set a precedent, he would live with the inconvenience; but the inconvenience could not last indefinitely. Third – and this is where our interests overlap with his – Simeon was displeased with the results of the survey. He seems to have thought that he and Saint Audrey had been fairly thoroughly traduced by the local juries. An emissary of Simeon’s set off in pursuit of the king, caught up with him somewhere, recited the abbot’s complaints, and in due course came back to England with a writ from the king addressed to archbishop Lanfranc.59 The writ ended up in the archive at Ely, and this is what it said: Willelm king of the English to archbishop Lanfranc greetings. I want you to look at the abbot of Ely’s charters, and if they say that the abbot of the place is to be blessed wherever the king of the country orders it to be done, I command you to bless him yourself. Also see to it that Ely bridge is repaired without delay by those who usually repair it. Find out from the bishop of Coutances, from bishop Walchelin, and from the others who saw to it that Saint Audrey’s lands were written down and sworn to, how the swearing was done, who did it, who heard it done, which the lands are, how large, how many, what they are called, and who holds them. Once these things have been noted point by point and written down, see to it that I am promptly informed of the truth of the matter by a report from you; and the abbot’s emissary is to come with it.

There were, no doubt, many good reasons for the king to return to Normandy – but perhaps the thought crossed his mind that it would also be desirable to put some distance between himself and any people who might have cause for complaint about the survey. Already in Salisbury, he had had to deal with a protest from bishop Walchelin, who thought that he had been unfairly treated by the commissioners who visited Somerset (above, pp. 72–3). Walchelin’s brother, abbot Simeon of Ely, was also feeling hard done by; but he was slower off the mark, and before he had acted the king was already overseas.

We do not know the sequel. We do not know for a fact that the writ was ever delivered, or, if it was, that Lanfranc had time to act on it before the king died. If he did, presumably he would have written to abbot Simeon, asking for the charters to be brought to him; but nothing came of that, and Simeon did not get blessed by the archbishop.60 As for the bridge, Lanfranc would presumably have forwarded the king’s order to the sheriff of Cambridgeshire and then forgotten the matter. The third and largest question would demand more thought than that, but what would the outcome be? Suppose that Lanfranc wrote to the bishop of Coutances: what could the bishop say in reply except that he and his companions had followed their instructions, that the proceedings had been properly conducted, and that the details which Lanfranc was asking for could be found (so far

Not that the survey was Simeon’s only concern. Three things were troubling him. First, though he had been in office for some years, strictly speaking he was still not the abbot. (A century later, when people were more punctilious about these things, he would have had to call himself the elect of Ely.) To become the abbot, Simeon would have to receive benediction from a bishop; and that was a contentious issue. Ely was in the diocese of Lincoln, nobody doubted that; but Simeon refused to be blessed by the bishop of Lincoln, and the bishop of Lincoln refused to let him be blessed by anyone else. (This is bishop Remigius, the man whom we met in Worcester.) Though not of any practical significance, the impasse was an annoyance: Simeon now thought that he had found a way of escaping from it. The abbey’s charters, it seemed to him, could be construed to mean that it was up to the king to decide where the benediction should take place. That seems a rather risky line of argument – what would Simeon have done if the king had agreed with it and then ordered him to be blessed by the bishop of Lincoln? – but Simeon was willing to try it. Second, the bridge at Ely was in need of

59 The writ is printed by Bates (1998, no. 127); it can also be found in Hamilton (1876, p. xxi) and Blake (1962, p. 206). Towards the end, I follow Round’s lead in starting a new sentence with His, not with fac. The dating was settled by Round (1895, p. 133). Davis (1913, p. 42), knowing that Simeon was appointed in 1082 and assuming that he was blessed immediately, dated the writ to that year. Galbraith (1942, p. 167) stood by Round’s dating at first; later on he changed his mind. A mischievous paper of Miller’s (1947) did some harm in its day, but I think it can now be forgotten. 60

The end of the story was that Simeon eventually backed down, on the advice of his brother, and allowed himself to be blessed by bishop Remigius (Blake 1962, pp. 201–2).

121

The survey of the whole of England as they could be found anywhere) in the written report submitted some months earlier to the treasury? Suppose that Lanfranc wrote to the treasury officials: what could they do? Given the name of a hundred, they could copy out the names of the jurors; given the name of a manor, they could copy out all or any of the particulars that were recorded. But how could they comply with instructions as vague as these? The king had said that he was expecting a messenger to arrive bringing the archbishop’s report, accompanied by the abbot’s representative; then he would decide what should be done.61 But no further writ was issued, and in the absence of that it is impossible to know what action was taken in response to the first writ. At some uncertain date, a scribe from Ely was given access to the B text and allowed to copy any paragraphs of interest to him (above, p. 107); but that is another story.

suspected of being a copyist’s addition; but here we have the clearest proof that this was the survey’s official Latin name. Abbot Gislebert’s complaint had another visible result, a note added in the margin of Db-Sy-32r, by the DB scribe: Modo geld’ pro viii hid’ (Chaplais 1987, pl. IV (b)). Complaints from bishops and abbots are the only ones which are likely to have left any trace;64 but there were, one imagines, earls and barons who also had reasons for feeling disgruntled, and who also had access to the king. If the records were more complete, no doubt it would appear that the king was kept busy fending off complaints of this sort, from people who wanted the written record set straight. Such evidence, however, would still not let us know what we should like to know most of all – what plans the king himself had in mind. If the king had come back to England towards the end of 1087, if he had met with his council at Gloucester again, as he had done two years earlier, the decisions made at that meeting might have told us what the survey was all about. But the king was dead by then.

At Westminster, abbot Gislebert was also feeling aggrieved. His church owned a manor in Surrey which lay partly within the king’s forest of Windsor. The manor, Pyrford, was assessed at 16 hides; that seems to have been agreed.62 The question was: how many hides had been taken into the forest? Even though they still belonged to the monks of Westminster, these hides had become, in a special sense, part of the king’s domain; and therefore they were exempt from paying geld. The commissioners surveying Surrey seem to have reported that there were 3 hides inside the forest (that is the number recorded in DB): it followed that geld was due from 13 hides (minus whatever deduction the abbot could claim with respect to his own domain). We do not know quite what action Gislebert took, but we do know what the outcome was: a writ from the king notifying the sheriff of Surrey that he has ‘granted eight hides quit’, because they are in his forest. In other words, he is telling the sheriff that geld is due only from the other 8 hides. This writ is of exceptional importance because it survives in the original.63 If only copies survived, the last four words – post descriptionem totius Angliae – would doubtless have been

Stage 2 had brought about some important changes in the content of the B text. The jurors’ names, the corrections and additions that were found to be necessary, the notes of current disputes, the new information about the towns and the customs of the county – all of this had to be written somewhere. Thus I revert to a question that I posed before (above, p. 66): in the form in which it was returned to the treasury after stage 2, was the B text the original, with some accretion of new material, or was it a new copy? All that we can do is balance the probabilities. It is possible that the scribes were under instructions to produce a clean manuscript, unmarred by corrections, uncluttered with marginal additions. But we may doubt whether that was necessary. The additions that had to be made did not amount (so far as we can judge) to more than a small proportion of the total text. Given that, and given that time was short, the annotated original – perhaps with some inserted leaves and an extra quire at the end – would probably be thought good enough. The B text was not intended to be kept for all time. It was required immediately as a source (via C) for D; it might have to be preserved for some time, until it was no longer of any conceivable use; but it did not need to be put into a form for posterity to admire. On the contrary, it might be positively desirable if the additions made in stage 2 could be distinguished at a glance, by their

61

There is no question of a new survey: the king is asking for a copy of the passages from the B text which relate to the lands of Saint Audrey. An annalistic text from Ely – excerpts from a copy of the Worcester chronicle augmented with passages of local interest (Blake 1962, p. 410) – includes an entry which presents the facts in that light: Willelmus rex fecit describi omnem Angliam, quantum quisque terre . . . possidebat, atque tunc nostras possessiones . . . describi iussit, petente Symeone abbate (Blake 1962, p. 430). The first half of the sentence is derived from the Worcester chronicle; the second alludes to this writ. In the first half describi means ‘to be described’; in the second it merely means ‘to be copied out’. 62

The local jurors thought it their duty to report that Pyrford had once been assessed at 27 hides, and that they were doubtful whether the reduction – made while the manor belonged to earl Herald – had ever been properly authorized (DB-Sy-32rb). But neither the king nor the abbot seems to have cared about that. 63

64 A late narrative from Gloucester preserves the faint echo of another complaint. Three hides at Nympsfield belonging to the monks of Gloucester were, without the abbot’s knowledge, written down as part of the king’s manor of Berkeley (cf. DB-Gl-163ra). The abbey’s chronicler says this: Anno domini millesimo octogesimo septimo, Rogerus senior de Berkelee in descriptione totius Anglie fecit Nymdesfeld describi ad mensam regis, abbate Serlone nesciente (Hart 1863, p. 101). (This is the abbot Serlo whom we met in Worcester.) Despite the written record – DB was not corrected – these hides was not lost.

It is reproduced by Bishop and Chaplais (1957, pl. xxiv) and by Chaplais (1987, pl. IV (a)); the text is printed by Bates (1998, no. 326). The date can be bracketed quite closely – not earlier than about May 1086, not later than September 1087 – but not closely enough for the writ to be meshed with the progress of the survey. (I do not think it can be decided whether the king was in England or abroad at the time.) The two men who witness this writ – Willelm bishop of Durham and Ivo Tailgebosc – would not have been disqualified from serving on the team of commissioners which visited Surrey; one cannot say anything more positive than that.

122

The fieldwork phase placement on the page and by the writing.65 It would then be very easy, for example, for the C scribes to compile a record of all pending litigation, such as survives in the original for three counties (above, p. 40): they would merely have to scan through B and copy the marginal entries that were relevant.

evidence, it probably did not move to Ely until the 1130s. Some extracts from B survived in the archives of the churches for which they were made. I mention two examples – the only two of which I feel able to speak both briefly and with confidence, but perhaps not the only examples to be found.

Of course it must be true that each B text was taken back to Winchester instantly, as soon as it was ready. There it was used for the compilation of the C text, and possibly also, at a later stage, for some checking of the D text. Then what? Though I was slow to realize this, I see now that the B text did not become redundant once D was in existence. On general grounds, if one stops to think, it seems very unlikely that B was discarded at once; we would surely want to keep it, for the time being, in case any query came up. It is B, after all, not D, which is the sworn record of the survey. More specifically, there is some information in B which has not been transferred into D. The names of the jurors are only to be found in B; and therefore B will have to be kept, for as long as that information has significance – i.e. for as long as there is any chance that the jurors may be accused of perjury. Now, if B is allowed to survive at all, there is a chance that inertia will take over, and that B will be kept indefinitely – no longer because there is any particular reason for keeping it, but because there is no particular reason for throwing it away.

(1) Compiled for the monks of Ely, the text which I call xEl (above, pp. 97–8) is – demonstrably so for three counties and probably so for all six – a concatenation of excerpts from the B text, overlaid with passages reflecting the results of some more detailed survey. There does not seem to be any good evidence for dating either the extracts or the interpolations, but probably neither layer of text is very much later than 1086.66 If xEl was compiled in Simeon’s time, as seems to have been either known or supposed by a later generation of monks, it (or some version of it) is earlier than 1093. (2) Compiled for the monks of Saint Augustine’s, the text which I call xAug is largely composed of extracts from the B text for Kent.67 In the only known copy it carries the title Excepta de compoto solingorum comitatus Cancie secundum cartam regis, ‘Excerpts from the enumeration of the sulungs of the county of Kent according to the king’s official record’.68 This copy is late and full of errors; it was printed, not very accurately, by Ballard (1920). There are, as Ballard saw, some anachronistic passages which have to be regarded as interpolations – some which are glaringly obvious (one of these dates from the middle of the thirteenth century), perhaps some others which are not so easy to recognize. Despite these complications, we can be certain that this text derives from B-Ke, because (with one small exception) the order of the entries is perfectly cadastral – perfectly cadastral twice over, in fact, because Kent was divided into lathes (of which there were seven) as well as into hundreds (of which there were more than sixty).69 It

Sooner or later, all at once or piece by piece, the B text did disappear, and it is only by accident that we know anything about it. The only strictly contemporary evidence is an epitome of B-Yo, a fair copy of which was (for some reason yet to be explained) made by the DB scribe, in a quire of similar size and similar format to the quires that he was using for the DB text (PRO, E 31/2, fos. 379ra–82rb). Since I have no understanding of the motives at work here, I do not take this to imply that similar epitomes existed for every county. But possibly they did, and possibly some of them may have left some trace.

66

It was Round’s idea that xEl was compiled in response to the writ addressed to archbishop Lanfranc (above, p. 121). As Miller (1947) saw, that is certainly wrong; but Round’s error affects the dating of xEl, not the dating of the writ.

The complete B text for Cambridgeshire (and apparently also for Hertfordshire) survived long enough to be carried off to Ely (above, p. 107). Even after that, a sequence of happy accidents had to follow before we could know anything about it. An Ely scribe decided to make a copy; the circumstances which led to the loss of part of this copy did not lead to the loss of the whole of it; Arthur Agarde got hold of the surviving portion and passed it on to Robert Cotton; Cotton decided to shelve it under the bust of Tiberius (where it would only be charred around the edges in 1731) rather than under the bust of Otho (where it would have been burnt to ashes). But first of all the original had to survive in the treasury for some length of time. As I read the

67

PRO, E 164/27, fos. 17r–25r. This book is a miscellaneous register from Saint Augustine’s, mostly written circa 1320; but the opening section (fos. 2r–48r) may perhaps be somewhat earlier than the rest. The xAug text is the work of the scribe who wrote the first part of this section (fos. 2r– 27r); neither he nor the next scribe (fos. 27r–48r) occurs elsewhere in the manuscript.

68

There is no need to emend Excepta to Excerpta: the verb excipere was used frequently in this sense – five times on one (printed) page of the Dialogus de scaccario (ed. Johnson 1950, p. 70), for example. (More important is the phrase compotus solingorum, the Kent equivalent for compotus hidarum.) The title continues: uidelicet ea que ad ecclesiam sancti Augustini pertinent et est in regis Domesday. (A later scribe has added W. conquestoris. It was Ballard’s policy to omit all annotation by other hands; so he does not deserve to be reproached for omitting these two words.)

65

To make this concrete, I offer some more empty predictions. If B-Wi survived, it would show a fair amount of marginal annotation, relating to the business transacted in stage 2. This annotation would be the work of scribe ksi and two companions (above, p. 68), and their contributions would rotate, hundred by hundred (above, note 58). Concerning the primary text I have no predictions to make.

69

Luckily for us, the abbot owned some property in every lathe; so the sequence is fully attested: Sutton, Aylesford, Milton, Wiwar, Borwar, Eastry, Limwar. At the time when it was passing through the C scribes’ hands, the B text was differently arranged, with Borwar ahead of Wiwar. I am not sure what conclusion should be drawn from that.

123

The survey of the whole of England is, however, far from certain when the excerpts were made. More or less in agreement with Ballard, I am inclined to think that the likeliest date would be circa 1110, when a new abbot, Hugo I (1107–26), was trying to restore some order to the abbey’s affairs after a 14-year vacancy. If that is right, it will mean that B-Ke was still in existence more than 20 years after the survey. But an earlier date is not out of the question. As these examples prove, there was some period of time during which (if one knew whom to ask) excerpts from B could be obtained – during which, perhaps, they were still preferred to excerpts from D or DB. But the length of this period is, and seems likely to remain, a matter of much uncertainty.

124

Chapter 11 The conduct of the survey: the compilation phase

Before we could think that we had a good grasp on the compilation process, we should need to have the time-scale worked out. That means that we should need to know, for each county, six specific dates: the date of the inception of the C text, the date of its completion, the date of the inception of the D text, the date of its completion, the date of the inception of the DB text, the date of its completion. If these dates were all known (not to the nearest minute, but preferably within a few days), we could make up a table with 33 rows (one for each county) and six columns (one for each category of date). Given the information contained in this table, we could then watch the compilation process unfolding day by day, from the inception of the first C booklet to the completion of the last DB booklet.

Even when it comes to actual dates, we are not totally ignorant. There are at least a few bounds to be set here and there. For reasons already explained, it is probably safe to assume that none of the dates in the first column (the C-inception dates) was earlier than the beginning of April; if that is true for the first column, it is true for the whole table. Moreover, I think we can be sure that none of the dates in the first column was later than the end of July. By the beginning of August, the last C text, C-So, had not only been begun: it had been finished – or at least it was thought to have been finished, until it turned out that one last-minute alteration had to be made, at the instance of the bishop of Winchester. On the other hand, the fact that it was still worth altering C-So implies that D-So had not yet been started. D-So must be later than the beginning of August; DB-So must be later again; and there are two other counties of which we know that the date in column 5 (the DB-inception date) must be later than it is for Somerset (above, pp. 79–80).

It is well that we should be aware of the depth of our ignorance, as long as the awareness does not induce despair. The fact is that we do not know – and have no hope of ever getting to know – a single one of these 198 dates. If we ask, not for actual dates, but for information regarding the sequence of events, we do not know much; but we are not entirely at a loss. It goes without saying that in every row the dates have to increase from left to right: the C text cannot be finished till after it has been started, the D text cannot be started till after the C text has been finished, and so on.1 In column 5, the order of the entries is mostly known: for thirty counties, with only a few uncertainties, the inception sequence for the DB booklets is reasonably well established. In column 1, the order is partly known: for nine of the same counties, the inception sequence for the C text is also tolerably certain. To the extent that they can be compared, there is a strong (but not perfect) correlation between the order of the entries in columns 1 and 5; for that to be true for two columns so far apart, there must have been a strong correlation between each of the intervening pairs of columns (1 and 2, 2 and 3, 3 and 4, 4 and 5). We shall not go far wrong if we assume that the order of the entries in each column is determined by the order of the entries in the preceding column.2

For the C text alone, bounds can be set on both its inception and its completion. The entire C text was written, it seems, between early April and late July – that is, within a period of roughly 100 days. We can take it for granted that the work was started at the earliest possible moment, as soon as the first of the B texts began arriving in the treasury; we are allowed to think that the C text had only just been completed when the bishop of Winchester began lobbying to have some part of the record rewritten. Throughout that period, the scribes would no doubt have been made to work as hard as they possibly could, from sunrise to sunset, day after day, week after week, until the job was done. Since it seems to have been the policy for five scribes (not counting mu) to be participating at any given moment, the labour available amounts, at most, to around 500 man-days. Since the C text can be estimated to have consisted of rather more than 3000 leaves, the rate of production that would have to be achieved would be something in excess of six leaves per scribe per day. For that average to be kept up, the two scribes who bore the brunt of the work (alpha and beta) would presumably have to be writing faster than that. There is scope here for some experimentation; but it does not seem to me at all unlikely that some of the C scribes were producing an average of ten or twelve leaves per day. They had a mountain to climb, but it was not Mount Everest: they were not making a copy of the Bible.

1 Here I am ignoring the possibility that a stretch of text which was thought to be finished might afterwards be found to need some further work. In many of the DB booklets, it is plain that additions had to be made, sometimes even after the text had been rubricated. There are two reasons why such changes might have become necessary: the DB scribe may have been correcting some error of his own; or he may have been revising the text to bring it into line with a revised version of the D text. 2 It is only with column 4 that this assumption looks a little risky. If (as I suppose) the D booklets were written in parallel, two or three at a time, the order of completion might differ significantly from the order of inception. To put it simply, the booklet for a small county might be started later but finished sooner than the booklet for a large county.

The rate of production of the D text is determined by the rate of production of the C text. It cannot have been faster than that; on the assumption that no avoidable delay was

125

The survey of the whole of England allowed, it is unlikely to have been much slower. Hence we would expect the production of the D text to occupy a similar span of time, roughly 100 days, offset by however long it took for the first collection of C booklets to become available. To judge from the few surviving booklets, it is unlikely that there were more than three D scribes at work at any given moment; and that would imply that the labour available for the compilation of the D text amounted at most to around 300 man-days. Because the D scribes had a much easier job than the C scribes, that number seems plausible to me. If they were under pressure to complete the job before the king left the country, they would have had to work as fast as possible; if they were not, they could have afforded to slow down a little. Though the DB scribe was coming along behind, one man could not keep up with two or three.3

If the reader thinks that these estimates are crude, I can only say that I agree. But any attempt to understand the compilation process will necessarily involve some estimates of this kind, and it seems to me a step in the right direction if the estimates are made explicit. Any proposed interpretation of the evidence has to stand up to this test: the amount of labour assumed to be available must be roughly the same as the amount of labour assumed to be required. Putting ourselves in the king’s place, we want the job to be done as quickly as possible. We accept the fact that to get the job done more quickly means employing more hands to do it, but we do not want to hire hands which cannot be kept busy. If the scribes complain of being overworked, that will not trouble us much. They may grumble as much as they please; they are not going to go on strike.

For the DB text a similar allocation of labour would be needed. Once it had been decided that one scribe should do all the work, there was no getting away from the fact that 300 divided by 1 is 300. The scribe could be prevented from slacking, but he could not be made to do the impossible. Working at maximum speed, perhaps he might have been able to produce – this is just a guess – as much as twelve columns per day; but he can hardly have sustained that rate throughout. From discontinuities in the manuscript itself, it seems clear that the work was done in a somewhat spasmodic fashion. There were long periods when the scribe was writing hard; but there were periods as well when he was checking back, planning ahead, or perhaps just scratching his head, wondering how best to escape from some impasse. Without knowing what share of his time was spent on incidental activities like these, it is hard to see how an estimate of his top speed could be converted into an estimate of his average speed.4 Still, if we suppose that he was writing fairly hard for more than half the time, 300 days would probably be enough.5 At the earliest, he could not have started work before about June 1086; so it seems that the work would have had to continue over the winter, perhaps reaching completion in the spring of 1087. 3

1 Unlike D and DB, the C text was not intended to be kept for ever. The scribes who wrote it assumed that it would be discarded, piece by piece, as its usefulness was exhausted; and most of it was discarded. By accident part of it survived – the part which was still potentially useful at the moment when use ceased being made of it. By luck it survived for long enough to be carried off to Exeter, where it found a safe home in the church’s library. On the C scribes’ account as well as ours, that was a fortunate event. Their handiwork did not all vanish. Because a portion of the C text exists in the original, we know how the survey data were taken out of the cadastral frame and put into a feodal frame. We are not reduced to guessing how this was done, from the evidence of D and DB; we can actually watch it happening. From the evidence of this surviving fragment, some conclusions follow quite directly. The C text originated in Winchester. It was written in the king’s treasury, the staff of which, because of the extra work involved, had been expanded far beyond its normal size. Three scribes – I call them mu, alpha and beta – appear to have worked on every section of the C text. Since the same three scribes were responsible for handling the geld accounts, they appear to represent (two of them at least) the treasury’s permanent staff. But those three were assisted from time to time by numerous other scribes, and these appear to have been temporary employees, hired only for the duration of the work. They were organized in teams of three, and we can watch these teams (five of them at least) rotating in and out of the treasury, as the operation proceeds (above, pp. 51–2). At any moment, therefore, there would normally be six scribes at work, the three treasury scribes plus one of the teams of hired scribes. But occasionally other scribes pitched in as well, perhaps because they had finished some task of their own and had a little time to spare.

A rather obvious point, but one which I failed to see (above, p. 23).

4

((Thorn and Thorn (2001, p. 72) seem to think that this obstacle can be surmounted, but I remain despondent.)) 5 This estimate is of the same order of magnitude as the one arrived at on the basis of Fairbank’s experiments (above, p. 14). As Jenkinson put it, ‘we might hope for a speed of about three folios (twelve columns) in two days’ (i.e. 6 columns per day); the wording seems to imply that this estimate is on the optimistic side. If the length of the written text is taken to be approximately 1440 columns, it will follow that DB represents ‘about 240 days’ work’, hardly less but quite possibly more (Jenkinson 1954, p. 34). But there is an undeclared parameter here. If Fairbank’s speed was measured in minutes per line or lines per hour, as it presumably was, some conversion would have had to be made from man-hours to mandays, and we are not told how this was done. For a civil servant employed in the Admiralty (like Fairbank) or the British Museum (like Jenkinson), there might only be eight hours in a day; but the DB scribe, when he was working flat out, may well have kept going for longer than that, perhaps for twelve hours, perhaps even (light permitting) for sixteen hours. Are we willing to believe that the DB scribe might have reached a maximum speed of around 12 columns (3 leaves) per day? If so, the equivalent maximum speed would be around 30 pages (15 leaves) per day for one of the C scribes, around 24 pages (12 leaves) per day for one of the D scribes.

Of the permanent employees, the one who writes least, scribe mu, is probably the man in charge: the tasks which he undertakes seem to be the tasks which he is not willing 126

The compilation phase to delegate.6 He supervises the compilation of the C text; he is presumably also responsible (somebody must be) for making sure that the treasury’s normal functions continue running smoothly meanwhile. The second, scribe alpha, is a man whose cursive style of writing seems to prove that he has made a career for himself as an administrator.7

seems to me, is whether the treasurer would, in normal circumstances, have been so busy that he needed a full-time secretary. In 1086 the circumstances were so very far from normal that it was necessary to hire a large number of temporary employees. It is a fact of some significance in itself that one could – if one were the king – requisition the services of at least fifteen additional scribes, all of them adequately competent, and some of them quite proficient. Who these men were, where they were found, how far afield one had to go in search of them, are questions for the experts, not for me. Two of them, perhaps three, are known to have been employed, at some point in their careers, copying manuscripts for the bishop of Salisbury;10 it is permissible to hope that more identifications of this kind may eventually be made, perhaps across the Channel as well as in England. It seems to me best to work on the hypothesis that these scribes were professionals – the sort of specialist who might sometimes find regular employment in the household of an individual patron but might sometimes sell his skill on the open market. On this occasion they were invited or impressed into the king’s service, and paid (I suppose) by the day.

These two can reasonably be identified with officials whose duties are described for us, some ninety years later, in the Dialogus de scaccario (ed. Johnson 1950). The treasurer was the man responsible for all aspects of the department’s business which involved the use of writing: scribe mu seems to be that man.8 Because his duties would sometimes require him to travel, the treasurer had a deputy (called the treasurer’s clerk), permanently based in Winchester, who had day-to-day charge of the department: scribe alpha seems to be that man. By the 1170s, the treasurer also had a secretary (called the treasurer’s scribe), who had to be capable of writing neatly, at the treasurer’s dictation, but was not required to do anything on his own initiative. Perhaps scribe beta was that man, but I do not press the point; it is not impossible that he was a temporary employee who, unlike the others, had been assigned to work at headquarters alongside scribe alpha. There is, of course, some risk in extrapolating backwards so far from the 1170s; but the risk should not be exaggerated. The treasury was there long before the exchequer existed; cash was flowing into the treasury (and out of it) long before this new apparatus was put in place to monitor the inward flow. In the 1080s, just as in the 1170s, there must have been someone keeping the door (the usher), and there must have been someone standing guard overnight (the nightwatchman). By the 1170s there were two parallel systems at work – one based on tallies, the other on written records – which regulated one another. If we extrapolate backwards far enough, perhaps we reach a period when the treasury was being managed by two illiterate officials (the chamberlains), who conducted all their business by means of tallies;9 but that period had ended by the 1080s. The geld accounts are proof – the clearest proof that could be asked for – that by this time it was a matter of routine for the treasury’s transactions to be recorded in writing. That means that there must have been a literate official (the treasurer) jointly in charge of the department; and this official would have had to have a deputy (the treasurer’s clerk), just as his illiterate colleagues did. The only doubtful point, it

From the surviving booklets, we can form some idea of the method by which the C text was compiled. Because the collections that survive were the last to be compiled, we cannot be sure how far the method had been worked out theoretically in advance, how far it was evolved by trial and error as the work proceeded. Even in the early stages, however, there cannot have been time for dithering: somebody had to know what needed to be done, and how the scribes assigned to the task were expected to go about doing it. The basic requirement was for the operation to be organized in such a way that several scribes could work simultaneously, without obstructing one another, and we understand how that objective can be achieved. In principle, the quires of B are made to form themselves into a reception line, and the quires of C are then made to move along this line, shaking hands with each quire of B in turn (above, pp. 45–7). Once it is started, the compilation process develops a momentum which will carry it forward without much thought on the part of the scribes involved, without much supervision. It is, I think, worth asking the question whether the treasury officials might have had any previous experience of working with an algorithm of this kind. Only one suggestion occurs to me: they could conceivably have done something similar with the geld accounts for some previous year. Suppose that they took a collection of geld accounts and spread them out along a table, in the way that I imagine them spreading out the quires of B. They could then scan through the accounts, excerpting the entries which would

6 Such as the additions made in the second version of the Wiltshire geld account (above, p. 68). 7 This is the man who wrote the geld accounts for Dorset, Devon, Cornwall and Somerset (above, pp. 61–2). 8

If that is right, we know his name, Henric, and we know that he had been given some land in Hampshire; but (as far as I am aware) we know nothing more than that. Whether his handwriting tells us anything about his career is a question that I leave for the experts. 9 It was one of the duties of the treasurer’s clerk to write inscriptions on the tallies. Do we have any idea how the tallies might have been distinguished from one another in this hypothetical period when they had no writing on them?

10

Unlike Webber (1992), I am assuming that these books were originally made for Osmund’s private library, and only later given (perhaps bequeathed) to Salisbury Cathedral.

127

The survey of the whole of England tell them – for each baron, for each county, for each hundred – how many hides of land were held by this man, and how many of these hides were held in domain. By working on tasks like this – relatively small tasks, perhaps of no great urgency – the treasury might have developed the skills that it needed when it came to compiling the C text. This line of thought is all guesswork; but it seems attractive to me – more attractive than the idea that the algorithm was invented out of nothing, on the spur of the moment.

This argument is tenuous in the extreme, and the reader is under no compulsion to give it credence. Except in an abstract way, we are never going to know how the C text was compiled; beyond a certain point, that does not matter. We can see that the job was doable; we can see that it was done. The facts to hold onto are these: that the compilation of the C text was a large and complicated business, that a method was devised for dealing with it, and that the job was carried through successfully.

The mechanical procedure that I have described will certainly get the job done; but it is wasteful of time for the C scribes to have to scan through the quires of B over and over again. If that was not obvious beforehand, it would soon become obvious once the work had started. Much time could be saved if some reliable person would read through each quire of B in advance and make an index of its contents. Then, whenever one of the C scribes came looking for a task, he could tell at once, by a glance at the index, whether or not it was worth his while to scan this particular quire.

If we could have looked into the office, at a time when the work was running smoothly, we would have seen five scribes in action. Each scribe had part of the B text. He also had one of the booklets which, when finished, would form a component of the C text; this booklet was intended to cover the lands of some given baron within the given county. The scribe’s instructions were to scan through this portion of B, stopping when he found a relevant paragraph, and then to copy this paragraph into the C booklet, with some necessary changes. Because the paragraph was being removed from its old frame and inserted into a new one, the opening words had to be modified. We might guess that this was so; in fact we know that it was so. On occasion, the scribe who had just completed a paragraph continued by writing the formula which would begin the next paragraph: ‘(The same man) has a manor which is called’ – at which point there is a change of hand. The scribe was expecting to find another paragraph which would need to be copied; but he had not found it yet, and never did. In the event it was one of his colleagues who found the next paragraph. This formula, then, is a feature of the text which originated with the C scribes: it replaced whatever formula they found in the B text.12 Once the opening sentence had been reorganized, however, presumably the rest of the paragraph was copied word for word. (Is there any reason why it might not be?)

Perhaps coincidentally, perhaps not, one of the surviving quires of C has an index of the sort that we are thinking might have been made for the quires of B. The quire in question (fos. 456–63) forms one of the omnibus booklets – its title is Terrae francorum militum in Deuenesira – used for several short stretches of the C-Dn text: the contents correspond with eleven separate chapters in DB-Dn, plus a paragraph in one of DB-Dn’s omnibus chapters. Near the bottom of the first page (456r), in small script, a line of text was added by scribe mu. It comprises these five names: Osbernus, Giraldus, R. paganellus, Guillelmus de ou, et ansgerus de monte acuto. Though in fact this index covers only the first three pages, as far as 457r15, the et before ansgerus seems to prove that it was complete at the moment when scribe mu made it. As soon as the next stretch of text was written (457r16–v11), the index became incomplete. By the time that this quire had been filled up, as far as 462v6, there were seven more names asking to be added to the index; but nobody took any notice.11

Finally the stage was reached when every paragraph from B had been decanted into one of the C booklets, and that meant that work on this county was concluded.13 The logical thing to do at this point would be to set the collection of C booklets aside (until the D scribes requisitioned them) and to start a new collection for the next county. But we know, from the surviving portion of the C text, that this is not what the C scribes did for the last five counties in the queue. Instead of making five collections of booklets, they made only two – one for Wiltshire and Dorset, the other for Devon, Cornwall and Somerset. To put it more exactly, they made a collection of C booklets for Wiltshire, and then, as far as possible, they used the same booklets for Dorset as well, not starting a new booklet except when they had no choice (i.e. when they encountered a baron in B-Do whom

It is tolerably certain, therefore, that this index was made by one of the C scribes, while the C text was in process of being compiled. The only question is why. From the fact that it was not continued and completed, we may gather that it did not serve any important purpose, as far as the C scribes were concerned. My suggestion is that scribe mu gave this quire of C the same treatment that had been given as a matter of course to the quires of B. He had indexed the quires of B; perhaps for some good reason, perhaps absentmindedly, he indexed one quire of C in the same way.

12 Which may or may not have been the same as in B-Ca: ‘And of these ten hides . . . ’.

11 At a later stage, the words in deuenesira were added by somebody else (not one of the C scribes) at the end of this line. The hand may be the same which inserted the significant words de episcopo baioccensi in the Somerset section of the same booklet, at 467r5 (above, p. 78).

13 I say nothing about the C scribes’ treatment of the entries that they found in B relating to towns. That is a question which might be worth looking at in detail.

128

The compilation phase they had not encountered in B-Wi). Similarly, they made a collection of C booklets for Devon, and then, as far as possible, they used the same booklets for Cornwall as well, and then again for Somerset as well.

omicron) was prone to be careless: in two instances that we know about (above, p. 44), he failed to follow the rule. In one instance the defective heading was completed by someone else; in the other instance it was not, and a paragraph which belonged in Dorset ended up in the wrong county. Accordingly, when we find two stray entries in DB-Wi (69ra, 73va) which properly belong in Dorset, we are justified in suspecting that this was due to some deficiency in the C text, even though the matching C entries do not survive.15

The packing of two or three counties into a single collection of booklets is a very striking feature, but I do not think that it has any deep significance. Though I have continued to ponder over this question, I cannot think of any explanation beyond the one which I suggested before – that the scribes were trying to cut down on their consumption of parchment (above, pp. 42–3). Furthermore, for reasons which will shortly appear, I am satisfied now that it is an adequate explanation. For this option to be available to them, the C scribes must have been well ahead of the D scribes: they had to be sure of being able to complete C-Do before the D scribes were ready to start work on D-Wi, sure of being able to complete C-So before the D scribes were ready to start work on D-Dn. Apparently they were safe in making these assumptions. But the C scribes did not expect this ad hoc arrangement to last. They were taking it for granted that the counties would be unpacked again, when the D text was written; and that is indeed what happened.14 Taking DB as a proxy for D, we find almost no trace of the peculiar arrangement which the C scribes imposed on the text.

Even if the county headings are all in place, a blunder by one of the D scribes may produce a similar result. If this scribe overlooks a heading and continues copying regardless, a block of text will be entered in the wrong county. In this case, however, we would expect the block to be duplicated: we would expect it to appear (wrongly) at the end of a chapter in one county and also (rightly) at the beginning of a chapter in another county. From evidence of this sort, if we could find enough of it, we might be able to work out the order in which the counties were dealt with by the C scribes.16 There is a reason why I risk labouring the point. From the evidence of DB it is possible to prove the existence of another collection of C booklets spanning more than one county – four counties, in fact, C-NnWaStOx – in which the county headings were occasionally omitted or misplaced; and the observable consequence of that is what we should expect it to be, that a few of the entries turn up in the wrong DB booklet.17 The booklet which is most affected, DB-Nn, contains some entries which properly belong in each of the other three. In a vague sort of way these facts are known – they are mentioned, for instance, by Finn (1957, p. 63, 1961, p. 168) – but I am not aware that they have ever been explained. Now that we understand what method was used for compiling the C booklets, now that we understand what risks were associated with that method, the explanation is obvious.

With this policy in place, it is crucially important for the beginning of each new county to be marked correctly in every single booklet. If any county heading is omitted or misplaced, the D scribes will not be able to unpack the counties without making mistakes: any block of text which lacks its proper heading will seem to them to be part of the preceding county. To prevent mistakes of that kind, the C scribes have to be given some rule such as this: Before you copy any part of the B text into a C booklet, scan backwards through the C booklet until you find a county heading. If the heading that you find is a heading for the current county, all is well: go back to the end and continue. If it is not, a new heading is needed, and you are responsible for making it: go back to the end, write the heading, and then continue.

To make sense of the evidence, we simply need to get it tabulated correctly. The overall pattern is shown in Table 36; the details are as follows:

And they have to be told to obey the rule without fail. Even one mistake will be fatal. If a block of text is added by a scribe who forgets the rule, the county heading may be absent altogether; if a block of text is added afterwards by a scribe who remembers the rule, the county heading will be present, but not in the right place. Unless these errors are corrected in time (the C text does seem to have been checked to some extent), the D scribes are doomed to go wrong, and the DB scribe is doomed to follow them.

entries in DB-Nn derived from B-Wa: the last entry in chap15 The Werrington case (above, p. 51) is different: what we see there is a portion of the B-Co text being purposely transferred into C-Dn. 16

The C scribes, not the D scribes. This evidence is not going to tell us whether the D booklet in which the error occurs was written before or after the other one: the observable result will be the same in either case.

17 For example, there was a hundred in Staffordshire called Cuttlestone hundred. (The spelling varies. Probably it ought to be Cudoluestan or something similar, but more often than not that elides into Coluestan.) In DB as a whole, there are thirteen blocks of text which carry the heading ‘In Cuttlestone hundred’. All of them ought to occur in DB-St, but in fact only ten of them do. Three have gone astray: one turns up in DB-Nn (222vb), and two turn up in DB-Wa (243ra, 243rb). It is obvious, therefore, that things went wrong during the compilation phase: the hundred headings were in place, but some of the county headings were not. From this evidence alone, we could tell that the C scribes dealt with these counties either in the order NnWaSt or in the order WaNnSt.

Almost without exception, the scribes who worked on the surviving C booklets, C-WiDo and C-DnCoSo, understood the rule and were punctilious about obeying it. But one of the scribes who worked on the C text for Dorset (scribe 14

The next person who edits the C text will, I hope, follow the same policy and deal with each county separately.

129

The survey of the whole of England B-Nn

B-Wa

B-St

B-Ox

DB-Nn

219–29

222va 224r 226rb 227va

222vb 226rb

221ra 224vb 226rb

DB-Wa

—

238–45

239rb 243ra 243rb

238va 244ra

DB-St

—

—

246–51

250rb 250rb

DB-Ox

—

—

—

ous enough to make a pattern. The negative indications are equally significant: there are no entries in DB-Wa derived from B-Nn, none in DB-St derived from B-Nn or B-Wa, none in DB-Ox derived from B-Nn, B-Wa or B-St. For these four counties, I think we can say with certainty that the DB text was derived, through D, from a collection of C booklets which covered precisely these counties, in precisely this order. By the criterion explained above, the errors which involve duplication are likely to be the fault of the D scribes; but most of the errors appear to have been due to deficiencies in C itself.20 Because the county headings were omitted or misplaced more frequently here than they were in C-WiDo or CDnCoSo, it seems likely that a change of policy had occurred, the implications of which were at first not fully understood by the people who had to implement it. Up until now, it would appear, the C scribes had been dealing with each county by itself: they compiled a collection of C booklets for a single county, and then they set it aside until the D scribes were ready to deal with it. They were under instructions to make sure that each booklet began with a heading which gave the name of the baron and the name of the county concerned, and usually they remembered to do so. On that policy, however, it would not matter much if some of the headings were missing: the name of the baron could be found in the first paragraph, the name of the county in the accompanying booklets. Now it would matter greatly. Once it had been decided to pack two or more counties into a single collection of C booklets, the headings became indispensable.21 Mostly the scribes understood that; but some either failed to grasp it, or failed to remember it when the moment came.22

154–61

Table 36. Entries misplaced in DB because of headings omitted in C-NnWaStOx. ter 10 (222va); the last two entries (added at the foot of the page) in chapter 19 (224r); the last entry in chapter 36 (226rb); the last entry in chapter 46 (227va) entries in DB-Nn derived from B-St: both entries in chapter 16 (222vb); the next-to-last entry in chapter 36 (226rb) entries in DB-Nn derived from B-Ox: the last seven entries in chapter 4 (221ra); the last four entries in chapter 23 (224vb); the last entry in chapter 35 (226rb) entries in DB-Wa derived from B-St: the last four entries in chapter 12 (239rb); the last entry in chapter 27 (243ra), duplicated in DB-St (250ra); the last entry in chapter 28 (243rb)

If we look more closely, we can prove that there was indeed a change of plan, and that it occurred while the C scribes were at work on the text for Warwickshire. Consider the case of Hugo de Grentemaisnil. Hugo owned land in three of the counties in question – nothing in Staffordshire, but several manors in each of the others. Three blocks of text would have existed in C, and the corresponding blocks in DB occur at 224va (for his lands in Northamptonshire),

entries in DB-Wa derived from B-Ox: the next-to-last entry in chapter 3 (238va); the last entry in chapter 37 (244ra) entries in DB-St derived from B-Ox: the next-to-last entry in chapter 12 (250rb);18 the last entry in chapter 12 (250rb), duplicated in DB-Ox (160va) Though some of these items of evidence may seem ambiguous, taken one by one,19 they are more than numer-

DB-Ox) at the end of a chapter. 20 To see how this works out in detail, the reader will need to reconstruct the C-NnWaStOx text for thirteen individual tenants: the bishop of Worcester, the bishop of Coutances, the church of Thorney, the church of Saint-Rémi, earl Roger, the count of Meulan, Turkil, Hugo de Grentemaisnil, Willelm Pevrel, Willelm son of Ansculf, Willelm son of Corbucion, Gislebert de Gand, and Osbern son of Ricard.

18

This entry and the next one should each be a separate chapter, but DB does not number them (nor index them) as such.

19

There are, for example, three entries relating to the village of Mollington in Oxfordshire – one entry in DB-Ox (157rb), but also one each in DBNn (226rb) and DB-Wa (244ra). Since the centre of the village was, in the nineteenth century, only about three miles from the border with Northamptonshire and only about one mile from the border with Warwickshire, it is conceivable that DB gives an accurate description of the situation, as it existed in 1086, and that the county borders here were realigned at some later date. As a matter of fact, it is known that part of this parish – the part which in 1086 belonged to Osbern son of Ricard (244ra) – was reckoned to be part of Warwickshire in the thirteenth century (Book of fees, pp. 508, 948) and later; but then the possibility arises that the border may have been adjusted precisely in order to make it conform with the evidence (supposedly infallible) of DB. These are questions for local historians to answer; but I hope it will be remembered that there are two stray entries to be accounted for, not just one, and that they both occur (unlike the entry in

21 It is not impossible that the same sort of thing had been done before to some extent; it is more likely than not, for example, that Derbyshire was packed into the same collection of C booklets as Nottinghamshire. We are not going to know that this happened unless something went wrong – unless the C scribes or the D scribes bungled the job. 22

If we were inclined to put all the blame on one scribe, the culprit would have to be one of the major scribes (because more than one successive county is affected); but I see no reason why we should so inclined. It seems to me more likely that the blame was shared by a number of minor scribes, who (like omicron) had not absorbed their instructions fully before they joined in with the work.

130

The compilation phase 242ra–b (for his lands in Warwickshire), and 224vb (for his lands in Oxfordshire). Because of the omission of the county heading, the third block of text has got itself displaced into the wrong DB booklet. If all three blocks had occupied a single C booklet, the absence of a heading would have caused the third block to attach itself to the second one, and the Oxfordshire lands would be included in DBWa. But in fact they are included in DB-Nn. The bishop of Coutances is in the same case as Hugo: the block of text covering his Oxfordshire lands (221ra) has attached itself to the block for Northamptonshire (220va–1ra), even though the bishop owned a manor in Warwickshire too (238vb).

of the rule which had been followed from the beginning for making sure that the hundred headings were in place. While the C scribes were working on C-Wa and C-St, they were expected to apply both rules; when they started work on C-Ox, for a while the same was still true. But then the procedure was simplified. From now on, the scribes were permitted to drop all the hundred headings (and perhaps in return they had to promise to be especially careful about the county headings). The upshot is that in DB-Ox there are very few hundred headings, and those only for a few of the hundreds that existed.23 In DB-Wi and the booklets after that, hundred headings are altogether absent; and we know that their absence here is not the fault of the D scribes or DB scribe: it is a feature derived from the C text.24

It follows that the C scribes must have created two booklets for each of these men: they made a C-Nn booklet; then they made a C-Wa booklet; and then they turned the C-Nn booklet into a C-NnOx booklet. When they compiled the C-Nn text, they were still making a separate collection of C booklets for each county; and at first they thought that they were doing the same with C-Wa. (That is why they started fresh booklets for the bishop and for Hugo.) But then the policy changed. The C-Nn booklets (which had been set aside but not yet removed from the office) were reactivated; and the scribes were instructed that they should, from now on, make use of an existing booklet whenever possible, rather than starting a new one. That policy was followed for the rest of C-Wa, and then for C-St, and then again for C-Ox.

It seems to me that we can say for certain which counties, in which order, were the last nine to be dealt with by the C scribes:

C:

(1) (2) | | ..... Nn Wa St Ox Wi Do Dn Co So

The two changes in procedure affecting the C text can be mapped onto this sequence: (1) the decision to save parchment by packing more than one county into a single collection of booklets; (2) the decision to save time by omitting the hundred headings. This second change was, I suppose, a delayed consequence of the first one. Because of the extra time that the scribes were now having to spend on the county headings, the work would certainly have slowed down to some extent, perhaps to an worrying extent; if they were told to forget about the hundred headings, that would speed things up again.

It seems that somebody (perhaps scribe mu) had done some calculations and come to an alarming conclusion: at the current rate of consumption, the C scribes were going to exhaust their allocation of parchment before they reached the end of their task. The existing procedure was visibly wasteful – in many booklets there were several blank pages at the end – and that wastage would have to be reduced. By packing counties together, parchment was saved, but only at the risk of dire consequences if any county heading was omitted. In fact, a number of mistakes did occur – not a large number, but enough that we can understand what was happening. Unaware of these failings (the fact that the errors were not corrected implies that they were not detected), the C scribes applied the same procedure to the next five counties – not to all five at one go, but first to two of them (C-WiDo) and then to the final three (C-DnCoSo).

There is a coda to the story of the C text. The coda may not be of much importance in itself; but without it there would have been no chance for any C booklets to survive. The D scribes, when they had finished with some instalment of the C text, did not throw it away: they returned it to the treasury scribes, who had some further use to make of it. If this final operation (whatever it was) had been completed, the C booklets would all have been discarded; because it was interrupted, some of the C booklets had some chance of surviving, and did indeed survive (above, pp. 53–5). More

One other change in procedure occurred during the compilation of the C-Ox text – apparently very soon after the start of this job. Up until now, the C scribes had been under instructions to include any hundred headings that were needed. Apparently they were inclined to be slack about this (though some of the headings which have gone missing may have been included in C and then dropped at a later stage), but in principle the rule was followed. When one of the C scribes found an entry in B, he did not copy it at once. He scanned backwards through B, looking for the hundred heading which governed this entry; then he copied the heading; and only then did he copy the entry itself. The new rule which had been introduced for making sure that the county headings were in place (see above) was a modified version

23 I take it that these were the hundreds which came at the beginning of the B text; but I do not have the local knowledge needed for working this out in detail. 24

The solitary heading that occurs in DB-Wi is only there because it had got itself welded into the text: the passage vi hid’ in RUSTESELLE appears in blundered form as vi hid’ in HUND’ de WRDERUSTESELLE (69rb). The solitary heading that occurs in DB-Dn is a mirage: in fact these words are part of the preceding sentence (Huic m’ pertin’ ii v’ t’r˛e et dimid’ In TAUETONE HUND’), but the scribe, arbitrarily, wrote them in capital letters, with the result that they look like a heading for the next paragraph (101va). A similar mirage appears in DB-Do (75vb).

131

The survey of the whole of England precisely, the facts are these. By the time that it was interrupted, the operation had been completed on all counties except the last five, and was in the process of being performed on Wiltshire; no C booklets survive for any of these counties (29 in number). The four counties at the end of the queue had not yet been touched; the C booklets survive for these, except for any C-WiDo booklets which contained (as many did) some portion of the C-Wi text. With regard to the last three counties, it had already been decided that Somerset should be dealt with before Devon and Cornwall, and the C-DnCoSo booklets had been sorted into two stacks accordingly. But then the operation came to a stop and never started again.

Though I do not have any clear idea how this checking operation worked, I suspect that it petered out when it did because the absence of hundred headings from the last collections of C booklets made it qualitatively more difficult. Perhaps the checker managed to fight his way through Oxfordshire but admitted defeat in Wiltshire. Perhaps he gave up on Oxfordshire, tried the next county in the hope that it might be easier, discovered that it was not, and gave up altogether. At first, no doubt, it was possible to believe that the work had only been suspended, not abandoned; but the survival of these booklets is the proof that it did not start up again.

The surviving geld accounts make a pattern correlated with this one. For all counties except the last five, no geld account survives. For Wiltshire uniquely, a version of the geld account survives (in three successive copies) which is later than stage 2 of the survey. For each of the last four counties, a version of the geld account survives (the version that I call ch) which dates from between stage 1 and stage 2 – which was created specifically to inform the commissioners responsible for finalizing the B text what discrepancies the treasury had found between the original geld account and the interim version of the B text (above, pp. 65–6). That correlation is our only clue to the nature of this final operation on which the treasury scribes were now engaged. They were making use of the C text; they were also making use of this (ch) version of the geld account.

2 There are three surviving D booklets, covering one county each: Essex, Norfolk, Suffolk. They add up to 451 leaves, probably about one-sixth of the total number. It is possible to estimate the sizes of the missing booklets, though only rather crudely (Table 5); for three counties it would be possible to reconstruct the D text, after a fashion, by interpolating between C and DB; but nothing that we can do will alter the fact that only three booklets survive in the original. This is not a satisfactory sample. With four or five booklets – especially if they were scattered across the map, and therefore presumably distributed over several stages of the D scribes’ imaginary journey – we should not have to feel inhibited from generalizing. With only three booklets, for three counties juxtaposed on the map, it is hard to be sure whether something that is true for all or some of these three was true for D as a whole.

Because it involved some version of the survey text and some version of the current geld account, I think we have to infer that some further, final checking of the geld account was being carried out.25 As far as the survey was concerned, the treasury’s participation had come to an end as soon as the C text was completed; but the treasury could never lose interest in the geld. Somebody (scribe alpha, at a guess) was going through the accounts again, looking to see if anything had slipped through the cracks when the data were translated from the cadastral frame into a new one. What he was doing, I suppose, was drawing up a list, for each county in turn, of the number of hides for which each of the king’s barons owed geld. He was assuming, in other words, that the collection of the geld was about to be feodalized. The old system was to lapse. From now onwards the king’s barons would be responsible for paying the geld, and they would pay it through the sheriff. The treasury had one last chance to make sure that the king would not be losing anything, when the new system started up.26 25

On the other hand, we have no reason to think that there was, originally, anything extraordinary about these particular booklets (or, further back, about these particular counties). These booklets survive by virtue of an accident which, on any straightforward view of the case, did not occur till later, perhaps much later: they survive because part of DB was either never completed or else completed but subsequently lost (below, p. 142). At the moment when they were writing these booklets, the D scribes had no way of knowing that. They could not be aware that in a hundred years’ time the D booklets would mostly have vanished, and that only three – these three – would still exist. In the absence of any evidence to the contrary, I think we have to assume that the surviving booklets are approximately representative, and that these three counties took their places in the queue approximately where we would expect to find them – not far from Cambridgeshire, not far from Kent – at every stage, at least up until the completion of the D text.27

26

So far as we are willing to assume that the entire D text was thought of as a unit, some degree of uniformity is to

But that will not explain one fact which seems to need explaining. How did it happen that two C booklets were within arm’s reach of the DB scribe while he was writing DB-So (above, pp. 76–9)? As was pointed out long ago by Ballard (1906, pp. 249–50), there is no doubt but that the system did change: the only question is when. In 1130, and at least for a few years previously, the geld accounts appearing in the exchequer rolls were the product of this new system. County by county, each baron was liable for some known number of hides: he was responsible for paying the money to the sheriff (unless payment had been forgiven), and the sheriff was responsible for forwarding it to the treasury.

We hear nothing of the sort of arrangements which existed in 1086. 27 If that is assumed to be true for the C text, one modest but not quite vacuous prediction follows: the D text ought to include the hundred headings. It does.

132

The compilation phase be expected. To the extent that they resemble one another (and, furthermore, to the extent that they resemble the C booklets), the three surviving D booklets should be typical of the whole collection. Nothing very exciting follows from that. All three booklets are largely composed of quires of eight; the dimensions are the same throughout; the ruling, though not very carefully done, was generally for 24 lines. The main text, beginning on the verso of the first leaf, was written in single column; there is no pronounced variation of format within or between booklets. When the main text was complete, the first page was used for adding an index, laid out in three or four columns. Then, with red ink, the chapters were numbered off: matching numerals were inserted against each item in the index and (always in the outer margin) against the first line of the chapter itself. The other rubrication may perhaps have been added at a slightly later stage;28 the colophon at least was presumably a finishing touch, added just before this batch of three booklets was sent off to be bound.

D scribes might need to do a large amount of shuffling before they could put the booklets into a satisfactory sequence. (But even then some pieces of the previous ordering might survive, just because there was no particular reason for changing them.) On the other hand, if the C scribes had already sorted the booklets into an order which seemed sensible to them, the D scribes might not feel obliged to do much shuffling; in the extreme case, they might simply accept the order proposed by the C scribes. However that may be, it was the D scribes who took the action which fixed the order. By writing out the text, they gave practical effect to the decisions which had been made, regardless of who had made them. Anyone acquainted with the structure of the DB text, replicated in booklet after booklet, will be surprised, perhaps even shocked, by the comparatively careless and inconsistent way in which the D text is organized. The king comes first – but on that point we would not expect to encounter any difference of opinion. There is a tendency for important people – bishops and abbots, counts and earls – to occur near the front of the queue, but they are not grouped together and sorted into order with anything like the same punctilio as in DB. In DB the rule is clear that clerics take precedence over laics. Among the clerics, the ordering is, up to a point, predictably hierarchical: archbishops come before bishops, bishops comes before abbots, abbots come before abbesses, and so on. Among the laics, counts and earls (in Latin the title is the same for both) come first. The structure of the DB text is a topic which will need to be discussed in greater detail, when the time arrives (below, p. 137), and I am trying not to anticipate that discussion more than can be helped. For the moment these comments should suffice.

In principle, the production of the D text was an easy job. The C scribes had done the hard work; the D scribes just had to make a fair copy. They took the C booklets, sorted them into the order that they thought best, and then copied out the text continuously, chapter after chapter, into a single booklet. Just by accident, it would occasionally happen that the start of a new chapter coincided with the start of a new quire; but in general the structure of the manuscript would not show any congruence with the structure of the text. Finally one of the scribes compiled the index, copying it onto the first page of the first quire, and with that the job was done.29 The ordering of the chapters originated with the D scribes, and so did the index which codified that ordering; but the main text, as far as we can judge, would have been something close to a verbatim transcript of the C text. There was no motive for omitting anything; there was no opportunity for adding anything (unless the D scribes sometimes referred back to the B text, as quite possibly they did).

Here, for example, are the first twenty chapters in D-Nk, as they are listed in the index: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

The basic decision which had to be made by the D scribes was the one which determined the order of the chapters. If we could work this out in detail, no doubt it would appear that the order was created partly by the C scribes, partly by the D scribes. Merely by forming the C booklets into a stack, the C scribes were imposing an order; they could not help but do that. The only question would be whether the order was haphazard (the accidental order, perhaps, in which the booklets had been completed) or whether some thought was put into it. If the order was haphazard, the 28

The rubrication of D-ExNkSk is described by Jenkinson (1954, pp. 45– 6), with the suggestion (based on some apparent variation in the quality of the ink) that ‘it may have been done in two stages’. Rumble (1987, pp. 80–1) regards the running heads (all of the ones on verso pages, most or all of the ones on recto pages) as the work of the same scribe who wrote the colophon, his scribe 7; unless I have missed it, he does not say who did the numbering. 29

The D indexes are (as one would expect them to be) almost perfectly correct. There would be no need to say this were it not for the fact that the same cannot be said of the DB indexes (below, p. 137).

133

Willelmus rex Ep’c baiocensis Com’ de maurit’ Com’ Alanus Com’ Eustachius Com’ Hugo Rob’ malet Willelmus de War’ Rog’ bigot Willelmus ep’c Osbertus ep’c Godricus dapifer Hermerus de ferer’ Abb’ de Sc’o E’ Abb’ de eli Abb’ Sc’i B’ de ramesio Abb’ de Hulmo Sc’s Stephanus de cadomo Willelmus de escois Radulfus de bello fago

The survey of the whole of England Putting ourselves in the DB scribe’s place, we are not going to feel at all satisfied with this arrangement. According to our sense of protocol, the two English bishops (chapters 10–11) should come directly after the king, and the abbeys (chapters 14–18) should follow the bishoprics. In DB-Nk, if it was ever written, the chapters would presumably have been rearranged at least to this extent: 1 10 11 2 14 15 16 17 18 3 4 5 6 7 8 9 12 13 19 20

organized with respect to one another. And that is bad news for anyone who might think of producing a new edition of part of the C text – of C-Dn, for example. Once the separate stretches of text have been transcribed from the manuscript, how should they be arranged? There is no given order, no order which is obviously the right one. If we knew, more or less exactly, the order in which the booklets were copied by the D scribes, it would, I think, be perfectly fair for an editor to put them back into that order. Since we do not know this, nor have any hope of knowing it, what other policy might this editor follow? Is there any satisfactory answer to that question?31

Willelmus rex Willelmus ep’c Osbertus ep’c Ep’c baiocensis Abb’ de Sc’o E’ Abb’ de eli Abb’ Sc’i B’ de ramesio Abb’ de Hulmo Sc’s Stephanus de cadomo Com’ de maurit’ Com’ Alanus Com’ Eustachius Com’ Hugo Rob’ malet Willelmus de War’ Rog’ bigot Godricus dapifer Hermerus de ferer’ Willelmus de escois Radulfus de bello fago

According to Rumble (1987), five scribes participated in the writing of these booklets.32 Their contributions are listed in Table 37.33 There is, I emphasize, nothing original here: though I have arranged it differently, the information is all taken straight from Rumble’s description. Two of these men, scribe 1 and scribe 2, who both copied out long stretches of the text, seem also to have shared the responsibility for putting it into its final form; it was these two who wrote the indexes. As for scribe 1, who made numerous additions and corrections throughout, he was probably the man in charge; but his status (unlike scribe mu’s) was not so elevated that he could not pitch in and write many pages himself. The other three scribes, by contrast, seem just to have been following instructions, and probably we should regard them (like most of the C scribes) as temporary employees.

The conclusion seems clear enough. As long as the king came first, the D scribes did not care much about the order of the chapters; nor did they care much whether the order in one booklet was consistent with that in another. The DB scribe thought differently. He saw it as part of his task to put the chapters into a better sequence – a sequence which would satisfy his sense of decency, and which would also be consistent from county to county. By and large he was successful in doing that. In order to do it, however, he had to be prepared to rearrange the text in a very drastic fashion. The ordering of the chapters in DB is an ordering created by the DB scribe; the ordering in D may have been – not only may have been but probably was – considerably different from this.30

None of the scribes who worked on D can be identified with any of the scribes who worked on the compilation of the C text: of that I feel fairly confident. But there are numerous additions and annotations in C which I have not looked at closely, and perhaps it may turn out that some of these were written by scribes who wrote some part of D. It is much to be hoped that someone will look at the evidence again with that thought in mind; I have not attempted it myself. The person who does it will have to be someone who (unlike me) is so intimately acquainted with the D scribes’ work that he or she can hope to recognize the hand even from a

I see no escape from this conclusion, except by resorting to the gratuitous assumption that the surviving D booklets may not be representative; and that is precisely what we cannot allow ourselves to do. But I concede that the conclusion is unwelcome. It means that we have no basis for reconstructing the order of the chapters in D, even with respect to those counties for which the C text survives. We can go a long way towards reconstructing the individual chapters as they appeared in D, by collating C with DB. But even in these counties – even where the evidence is at its best – we are, I fear, never going to know how the chapters were

31 There is an unsatisfactory answer. What Ellis did, in a desultory fashion, was to arrange the C text in parallel with the DB text; and this hypothetical editor may have no choice but to do the same with C-Dn (though do it more systematically). That arrangement will be convenient, for the purpose of comparing C and DB, but it will not have any authenticity. In principle one ought to be rearranging DB so as to line it up with C, not vice versa. 32 I do not count Rumble’s scribe 4, who makes only one small contribution (D-Nk-110v8–10), filling in a space left vacant by scribe 2 (Rumble 1987, pl. VII (a)). By the standard that I have applied to the C booklets, this does not entitle him to a number. In any case it cannot be said for certain that he was part of the original team: this addition could have been made at some later date, as long as either B-Nk or C-Nk was still in existence. 33 This table omits the first page of each booklet. The index to D-Ex (laid out in four columns) was supplied by scribe 1 (1r), the index to D-Nk (three columns) by scribe 2 (109r); the index to D-Sk (three columns) was begun by scribe 2 but mostly written by scribe 1 (281r).

30

As the reader may notice, this means that I am, with some disappointment, rejecting a suggestion of Galbraith’s which I was formerly inclined to accept (above, p. 43). I discuss the point elsewhere (below, p. 139).

134

The compilation phase few scribbled words. I confess to having hoped that there might be some intersection, enough to indicate that both texts were compiled in one place. But such evidence, if it existed, would not exactly prove the point (because scribes have legs and can move from place to place); and certainly the absence of such evidence does not disprove it. Because only small fractions of either text survive, a scribe might have written both a large part of C and a large part of D without our knowing anything about it. The circumstantial evidence is strong enough by itself: since C was written in the treasury, since D was intended to be kept in the treasury, we can take it for granted that D was written in Winchester. What we have to infer, it seems, is that the D text was compiled in a different office from the C text: there was interchange of documentation between these offices, but (as far as we can tell) no interchange of personnel.

D-Ex 1v–8v

scribe 1

9v–16v

scribe 1

17v–99r 99r 99v–103v 104r–7v

scribe 1 scribe 3 scribe 1 scribe 3 D-Nk

109v–56r

scribe 2

157r–9v 159v–72v

scribe 3 scribe 5

173r 173r–6r 176r–v 176v–8v 178v–9v 179v–81v 182r–v 183r–90v

scribe 1 scribe 2 scribe 5 scribe 2 scribe 5 scribe 2 scribe 1 scribe 6

191r–3r 193r–208v

scribe 1 scribe 5

209r–22r 222v–34v

scribe 1 scribe 6

235r–73v 273v–9r 279r–v 279v–80r

scribe 6 scribe 1 scribe 2 scribe 1

Having come so far, it seems to me that we can go a little further. The compilation of the D text was, I suggest, the responsibility of a different official, of higher status than the treasurer; and presumably that has to mean the chancellor (whose name was Girard).34 It is not to be thought that the chancellor would have participated personally in the writing of the text: his duties required him to stay close to the king, and his supervision would have had to be exerted from a distance. What happened, it seems, is that the conduct of the work was delegated to two reliable employees, scribe 1 and scribe 2, detached from the king’s household for the duration of the job. They were provided with accommodation in Winchester, separate from but in proximity to the office where the C text was being compiled; they were also provided with some extra help, in the shape of a few hired scribes. We might perhaps hope to recognize the chancellor’s hand, if it appeared, as a hand which made comments or corrections overruling scribes 1 and 2; and no such hand occurs. But I do not think it unlikely that the rubrication was added by the chancellor himself, who would thus have been putting his stamp of approval on every finished page. And is it not positively likely that the chancellor himself was the man who added the colophon?

D-Sk 281v–88v 289r–90r 290r–v 291r–4v 294v 294v–7r 297r 297r–8v 298v 298v–354r 354r 354r–6r 356v–78v 378v 379r–88v 388v 389r–449v 450r

scribe 2 scribe 1 scribe 5 scribe 2 scribe 1 scribe 2 scribe 1 scribe 2 scribe 1 scribe 2 scribe 1 scribe 2 scribe 5 scribe 1 scribe 5 scribe 1 scribe 5 scribe 2

As far as we can imagine it, the division of labour among the D scribes involves no such complications as those with which the C scribes had to contend. In the simplest case, each county is dealt with by a single scribe. Two or three scribes can be at work side by side, each on a different D booklet.35 Assigning a task means giving the stack of C 34 Though the evidence is thin, it is sufficient to prove that Girard served as chancellor from 1082 7 till 1087 91 (Galbraith 1931, Bates 1998, pp. 101–2, Burton 2004). This is the same man who became bishop of Hereford in 1096, archbishop of York in 1100; he died in 1108. Two people who are known to have taken an early interest in the work of the survey – Walkelin bishop of Winchester (above, pp. 72–3) and Simeon abbot of Ely (above, pp. 121–2) – had one thing in common beyond the fact that they were brothers: they were also both uncles of Girard’s.

Table 37. Scribal stints in the surviving D booklets, as they were identified by Rumble (1987).

35 Whether it would be feasible to seriate the surviving booklets, with respect to their order of inception or completion, is not clear to me. Perhaps

135

The survey of the whole of England booklets for a specified county to a specified scribe. There is no interference between one task and another. Of course it is not compulsory for each task to be completed by the same scribe who started it: at any moment, for one reason or another, a task may be handed over to a different scribe, or one task may be exchanged for another: the task is still the same task, even if someone else is now responsible for it. Unless such contingent changes of hand supervene, the entire D booklet will be written by one scribe, except for any corrections and additions which may subsequently need to be made.

As soon as he did that, the disjunction was hidden, and as if by magic two booklets were turned into one. The remaining booklet, D-Nk, is more of a puzzle. Put bluntly, the question is this: why did the D scribes make such heavy going of the Norfolk text? It might be worth someone’s while to look at this booklet more closely, but I have nothing constructive to say about it.37 Overshadowed by DB, this surviving fragment of the D text has never received its fair share of attention, and I am conscious of having failed to do it justice. If things had worked out as the D scribes expected them to, the D text would have become the basis for all future discussion, and we should not be taking much notice of DB. As things are, anyone aiming to write a balanced account of the records resulting from the survey will need to make a special effort to give D the largest share of space, DB the smallest share. In comparison with D, for as long as D survived, DB had no more value than an epitome of DB has now in comparison with DB. Just as an epitome of DB – such as PRO, E 36/284 (Hallam 1986, pls. 15–16) – might be handsomer than DB, so DB might be handsomer than D; but it did not have any authority.

Having completed one such task, the scribe takes possession of the next available stack of C booklets and starts writing the D text for that county. If no new stack is available, he does not have to stay idle: he can join forces with one of his colleagues. All that is necessary is for the stack of C booklets to be split into two portions: the scribe who was already at work on this county continues with his portion of the stack, and the scribe who has now joined him begins copying the other portion. Unless the first scribe takes some preventive action, that plan will produce two booklets, perhaps with several blank pages at the end of this scribe’s last quire. The second scribe does not need to worry about that; in fact he will probably prefer to leave some blank pages at the end, in case there are any additions which may have to be made.

The D text was intended to be the permanent record of the survey. Completed, rubricated, bound up in (so it seems) six massive volumes, it was expected to inspire the admiration of future generations. On the last page of the last county in each volume,38 written in red ink and capital letters, possibly by the hand of chancellor Girard himself, this inscription was addressed to posterity:

Of the surviving booklets, D-Ex is the most straightforward. The first two quires are disjunct; but from the third quire onwards (17v–103v) the text was copied out continuously, nearly all of it by scribe 1. With D-Sk a different plan was followed. The first portion of the text (281v–356r) was mostly written by scribe 2, with occasional interventions by scribe 1; the second portion (356v–449v) was mostly written by scribe 5, again with occasional interventions by scribe 1.36 Apparently the stack of C booklets was divided roughly in half, so that two scribes could work simultaneously, each making a separate booklet. But the discontinuity between these booklets has been adroitly disguised. Scribe 5 began at the beginning of a chapter (which, when the chapters were numbered, became chapter 14), on the first leaf of a quire; but he started on the verso (356v), leaving the recto blank. (Scribe 2 did the same in his first quire, but not for the same reason: there the first recto was being reserved for the index.) With the help of some luck, this made it possible for scribe 2, as he approached the end of the preceding chapter, to lay off the text in such a way that he had something left to write onto this recto page (356r).

ANNO MILLESIMO OCTOGESIMO SEXTO AB INCARNATIONE DOMINI, VIGESIMO VERO REGNI WILLELMI, FACTA EST ISTA DESCRIPTIO, NON SOLVM PER HOS . . . COMITATVS SED ETIAM PER ALIOS.

It will be recalled that a contemporary writer, speaking of the survey, used very similar language (above, p. 108). But if posterity lapsed into ingratitude, even for an instant, the monument that had been so laboriously constructed might vanish. Sooner or later, that moment came, and somebody decided to discard, not all six, but five of these volumes. Only one volume was kept, and that one only by default. If the corresponding DB booklets had existed at that moment, this volume would have vanished too; because they did not exist, it had to be allowed to survive. We are left with a fragment of the D text, and we have to be grateful

we could risk assuming that Norfolk and Suffolk were kept together at every stage; for these two counties the question would then be whether D-Nk is earlier or later than D-Sk, and for Essex the question would be whether D-Ex is earlier or later than the other two. Possibly those questions are answerable. But I see no hope of being able to decide whether we are dealing with a tight sequence (such as .. NkSkEx..) or with a loose one (such as .. Ex.. SkNk..), and that is probably the question which matters most.

37

It might be suggested, for example, that the C scribes had packed these two counties into a single collection of booklets (with Sk before Nk, hypothetically), and that the D scribes had no trouble dealing with the first county but found it hard work to unpack the second one. I do not find this suggestion convincing myself – the sort of difficulties which might be expected to result, if this hypothesis were true, do not seem to match up with the sort of difficulties which the D scribes were encountering – but perhaps there may be something in it.

36

It looks as if scribe 2 and scribe 5 were under instructions to omit any paragraph which they thought problematic, leaving a suitable number of blank lines (they could calculate this number from the number of lines in C) so that the missing text could be inserted later by scribe 1.

38 The colophon to D-ExNkSk (450r) is reproduced by Rumble (1987, pl. V (b)). ((Also by Thorn and Thorn (2001, ill. 31).))

136

The compilation phase for that. For the rest we are left with a condensed version of the survey text – rearranged, abbreviated, reworded – which was not originally intended to stand alone.

in D, tightly and consistently organized in DB. Since this question is directly relevant, and since it has seldom been discussed (and never with any insight), I propose to work through it in some detail; and the example that I take is the next-to-last index written by the DB scribe, the index for DB-Dn (Table 38).42 For the moment it will do no harm to assume that the index is perfectly correct, i.e. that the items in the index correspond one-to-one with the chapters in the main text. (In fact the index has some defects, but that will not become relevant till later.)

3 In the writing of C and D, speed was the prime consideration. Teams of scribes were set to work, in large enough numbers to get the job done within the shortest practicable span of time. DB is a different case. This is a calligraphic manuscript, designed and executed by a single scribe.39 He was working by himself, not in a busy office. He had time to think, and made the most of it. In many respects, the DB text is a vastly more impressive piece of work than D, more ambitious, more thoroughly thought out, more successfully executed. As soon as it came to be understood that DB was the work of an individual, it ought to have become clear too that he was a man possessed of much originality; but that realization, it seems to me, has not sunk in fully even now.40 One of his creations was a system of phonetic spelling, based on the system used for French, which could be applied to English names. (The French system could already cope with Breton names; the new system had to be capable of coping with Welsh and Cornish names too.) He designed it, not for his own benefit, but for the benefit of those of his colleagues who did not know much English. By using this system, he hoped, they would, when reading out some document, be able to pronounce the names well enough to make themselves understood – well enough not to be sniggered at by the natives. Though possibly none of the elements was new, the system as a system was; and government scribes in England spent the next hundred years debating its merits and demerits.41

Up to a point, the organizing principles behind this index are obvious. The most basic distinction is the one between the king and everybody else. Line 1 is reserved for the king; his name alone is written in capitals. Lines 2–52 are available for everybody else. Unlike the D scribes, the DB scribe has some definite rules in mind: he has applied them in other counties, and now he applies them here. Clerics are given precedence over laics, bishops over monasteries, English bishops over Norman bishops, English monasteries over Norman monasteries,43 male monasteries over female monasteries.44 (Few women appear in the indexes: those who do are normally put in the lowest possible place.) These rules do not determine the order fully – they do not explain why Tavistock comes before rather than after Buckfast, for instance, or why Le Mont-Saint-Michel comes before rather than after Saint-Étienne – but in broad terms they work well enough (lines 2–13). Among the laics, earls and counts are given precedence (lines 14–15); then we come to a large crowd of lesser barons. One is a woman; predictably she is placed last (line 45). Some of the others are important men, of a status not much inferior to an earl’s; but of course their importance is not necessarily reflected by the extent of their holdings in this or any other single county. This part of the index (lines 16–44) has a carefully worked-out structure, and no trace of anything like it can be found in the D indexes. The items are grouped by initial letter, in what seems to be an arbitrary order (B, I, W, G, R, T, A, O), and they are also grouped by name (Willelm, Walter, Robert, Radulf, Alured).45 Towards the end

Another strong contrast between D and DB results from the same sort of systematizing urge, felt and given effect by the DB scribe. It has already been said (above, p. 133) that the order of the chapters is loosely and inconsistently organized 39

I refer to the DB scribe as ‘he’ for convenience only; I do not mean to say that ‘he’ must have been a man.

42 The numbers are mine: they refer to the lines, not the items, but that distinction makes no difference except in line 47. All the numbering in DB, both of the items in the index and of the chapters in the main text, was added afterwards, in red, and therefore I ignore it: I am discussing the index as it was originally written. Errors affecting the numbers are a separate issue: they are annoying, especially for an editor, but I do not see any significance in them.

40

The discussion of the place-name evidence by Sawyer (1956) is flawed for this reason. There are some references to ‘the scribe of DB’ or ‘the DB scribe’ (pp. 488–9), but these should apparently be taken to mean ‘the scribe who wrote the matching entry in DB’: in general it seems to be assumed that DB was written by several ‘scribes’ who all used the same ‘set hand’ (pp. 495–6). (Similarly, in his review of Jenkinson (1954), Sawyer (1954) makes one cryptic allusion to Fairbank’s proposal – ‘the scribe(s)’ – but then implicitly rejects it with the remark that ‘it is very difficult to detect changes of hand’ in DB.) Yet the distinction is obviously crucial. If DB was written by several scribes, its orthography must be, in Galbraith’s terms, a curial orthography evolved over the previous twenty years. If DB was written by one man, its orthography must be novel. The same point was made by Rumble (1985, pp. 48–9); but he agreed with Sawyer that there was something ‘Old English’ about this new system, and that seems mistaken to me. As far as I can judge, the DB scribe knew English well, but only as a spoken language.

43

There is a query about line 10. As a matter of historical fact, it is clear that the lands described in this chapter belonged to the cathedral church in Rouen, which was served by canons, not monks. If that had been what the DB scribe was trying to say, however, we would expect the item to be worded differently (perhaps Canonici rotomagenses) and placed further down the list. He may possibly have thought – what Ellis (1833, vol. 1, p. 481) understood him to mean – that the church in question was the monastery of Notre-Dame-du-Pré, on the southern outskirts of Rouen; I cannot say whether that is likely or not. 44 There is only one female monastery represented here, La Trinité in Caen (line 13).

41 One person who disapproved of it was the man who compiled an epitome of DB-Ke (below, p. 144). He made a point of giving the place-names a much more English look – and by doing so declared that in his opinion the DB scribe had simplified the spelling too far.

45

Again the order seems arbitrary: why not put Walter before Willelm (as is done, for instance, in DB-So), Radulf before Robert (as again is done in DB-So)? Note also that there is one exception to this rule: the two Ricards

137

The survey of the whole of England (lines 46–52), the logic at work is less obvious; but we may doubt whether it is accidental that the three G’s come together, two of them sharing a line. In Devon, as in many other counties, a number of tenants occurred who held one or a few small manors directly from the king. They were, in a technical sense, the king’s barons, but they were not important people. The D scribes seem not to have objected to making as many short chapters as might be needed, with the result that the D indexes tend to be very long; the DB scribe seems to have more inclined to cut things short by making one or two omnibus chapters at the end, and that is what happens here (lines 51–2).46 These differences (especially the last one) could be said to be matters of degree: the DB scribe did not always follow his own rules; the D scribes did vaguely defer to a similar etiquette. Nevertheless, the differences add up to a very pronounced contrast.

HIC ANNOTANTVR TENENTES TERRAS IN DEVENESCIRE. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

REX WILLELMVS Ep’s de Execestre Ep’s Constantiensis Eccl’a Glastingberie Eccl’a de Tauestoch Eccl’a de Bucfesth Eccl’a de Hortune Eccl’a de Creneburne Eccl’a de Labatailge Eccl’a de Rotomago S’ Mar’ Eccl’a de Monte S’ Michael’ Eccl’a S’ Stefani de Cadom Eccl’a S’ Trinit’ de Cadom Comes Hugo Comes Moritoniensis Balduinus uicecomes Iudhel de Totenais Willelmus de Moion Willelmus Chieure Willelmus de Faleise Willelmus de Poilgi Willelmus de Ow Walterius de Douuai Walterius de Clauile Walterius Goscelmus Ricardus filius Gisleberti comit’ Rogerius de Busli Robertus de Albemarle Robertus Bastard Ricardus filius Torulf Radulfus de Limesi Radulfus Pagenel Radulfus de Felgheres Radulfus de Pomerei Ruald Adobed Tetbaldus filius Bernerii Turstinus filius Rolf Aluredus de Ispania Aluredus brito Ansgerus Aiulfus Odo filius Gamelin Osbernus de Salceid Vxor Heruei de Helion Giroldus capellanus Girardus Godeboldus Nicolaus Fulcherus Haimericus Will’s et alii seruient’ regis Coluin et alii taini regis

Suppose that the index for D-Dn was about as haphazard as the index for D-Nk (above, p. 133). Starting with this index, how does the DB scribe go about reorganizing it? In principle, he has to follow some such procedure as this. He takes a strip of parchment, long enough for the purpose, wide enough for two columns. In column 1 he makes a copy of the existing index, exactly as it stands in D-Dn (except that he spells the names as he prefers to spell them).47 In column 2 he is going to compile his own draft index: each time he adds a name to this new list, he will cancel it from column 1. He starts with the king, of course; then he deals with all the bishops; then with all the monasteries; then with all the earls and counts. By this time he has made 15 entries in column 2 and has cancelled 15 entries here and there in column 1. Now he is ready to start dealing with the barons. The first uncancelled item in column 1 is Balduinus uicecomes; so he adds that name to his list and cancels it from column 1. Next, he scans through the rest of column 1, looking for another Balduinus; failing to find one, he scans through column 1 again, looking for another B; failing again, he moves on. He finds the next uncancelled item in column 1, Iudhel de Totenais, adds it to column 2, cancels it from column 1, and then scans twice through the rest of column 1 as before, looking first for another Iudhel and then for another I. Both scans having failed, he moves on again. By now the first uncancelled item in column 1 is Willelmus de Moion; so he adds this name to column 2 and cancels it from column 1, in the usual way. Then he scans through the rest of column 1, looking for another Willelmus, and this time the scan pays off. In fact he finds four more Willelms, and each of them in turn is added to column 2, cancelled from column 1. Having disposed of all the Willelms, he scans through column 1 again, looking for another W, and once again the scan pays off. The W he finds are not brought together. 46 The C scribes, for practical reasons, had resorted to making a number of omnibus booklets (above, p. 43). The DB scribe, for reasons of his own, was adopting a similar policy; but the contents of his omnibus chapters do not coincide in detail with those of the C booklets.

Table 38. The index for DB-Dn. (In the manuscript the names are arranged in two sub-columns, the second of which starts with line 27.)

47 I take it that the DB scribe was forbidden to make marks in D: therefore he needed a copy.

138

The compilation phase

1 15 18 19 20 16 22 14 17 21 23 24 2 3 4 5 6 36 40 44 7 8 29 38 11 33 43 50 25 34 39 9 26 30 47 41 45 10 31 49 12 13 32 27 28 35 37 42 48 46

is a Walter, Walterius de Douuai; when the time comes to look for another Walter, again he will succeed. So the first Walter will move up to follow the last Willelm, and the second Walter will move up to follow the first Walter. And so on. Finally, when every entry in column 1 has been cancelled, the new index is complete.48 (It should be of the same length as the old one: if it is not, something has gone wrong.) Applied to the first fifty items (i.e. the first two columns) of the index for D-Sk, this algorithm produces the result shown in Table 39.

WILL’ REX anglorum L. archiep’c Ep’s tedfordensis Feudum eiusd’ ep’i Ep’s rouensis Ep’s baiocensis Ep’s ebroicensis Abb’ sc’i Etmundi Abb’ de ramesia Abb’ de eli Abb’ de bernai Abb’ de ceterith Comes moritoniae Comes Alamus Comes Hugo Comes Eustachius Rob’ malet Rob’ greno Rob’ filius corbutionis Rob’ de todeneio Rog’ Bigot Rog’ pictauensis Rog’ de otburuill’ Rog’ de ramis Rad’ de bella fago Rad’ baignardus Rad’ de Limeseio Rad’ de felgeriis Ricardus filius comitis gisl’ Ranulfus piperellus Ranulfus f ’r ilgeri W. de Scoies W. de uuarenna W. f ’r eiusd’ W. de archis Galt’ diaconus Galt’ gifart Hermerus de ferreris Hugo de monteforti Hugo de grentemesnil Frodo f ’r abb’is Godricus dapifer Goisfridus de magna uill’ Suenus de essessa Eudo dapifer Albericus de uer Petrus ualoniensis Tihellus de herion Drogo de beureria Comitissa de alba marla

It is not likely that the DB scribe proceeded in this purely mechanical fashion. The task is a small one, easier to perform than to describe, made easier still by practice; no doubt he allowed himself to take short cuts. To the extent that he did work mechanically, the order of the items in column 2 will be dictated by the order of the items in column 1. It would be possible for us, given the index of D-Dn, to construct our own version of an index for DB-Dn, just as we have done hypothetically for DB-Nk (above, p. 134) and DB-Sk (Table 39), in the expectation that our version would not be grossly different from the DB scribe’s version. But the process cannot be reversed. Given the index of DBDn, we cannot reconstruct the index of D-Dn. Some partial sequences can be recovered, but they cannot be fitted back together again to make a single sequence. Among themselves, the five Willelms ought still to be in the same order as in D; among themselves, the four Radulfs similarly. But it is quite impossible to say where any Willelm stood with respect to any Radulf. There is a reason why it is important to grasp this point. It often happens that the DB index does not tally exactly with the DB text.49 These discrepancies are slight, and to me they seem epiphenomenal. If the DB scribe was juggling with 40 or 50 chapters, it is not surprising that he sometimes lost track of what he ought to be doing. Now and then, he made a change that he had not been intending to make, or omitted to make a change that he had been intending to make – and failed to adjust his draft index correspondingly. There is, in my view, nothing more to it than that. For Galbraith, however, these discrepancies formed the basis for a different conclusion. They meant, he thought, that the DB scribe was copying the indexes from D, rather than making new indexes of his own (Galbraith 1961, pp. 192–3). At one stage I thought that this suggestion (despite being open 48 The D scribes could have achieved the same result with less effort, simply by shuffling the C booklets. (The reader may wish to work out an algorithm for them.) They could have; but they did not. 49

In DB-Dn, the anomalies are as follows. (i) There is a short chapter – Quod tenent clerici de rege (104rb) – which is neither numbered in the main text nor listed in the index. (ii) The last paragraph in chapter 24 is a special case – a manor held jointly by Walter de Clavile and Goscelm (above, pp. 43–4) – and the DB scribe was in two minds how to deal with it. The wording is the wording that he uses for a new chapter (Walterus et Goscelmus ten’ de rege . . . ), but he makes no heading for it. Line 25 in the index seems to refer, inaccurately, to this paragraph. (iii) The last paragraph in chapter 25 is another special case (above, p. 44). Again the wording, but only the wording, suggests that this is a new chapter (Goscelmus de Execestre ten’ de rege . . . ). There is no matching item in the index.

Table 39. The first fifty items from the index for D-Sk (281r) put into a DB-like order. (One of the Willelms (line 30) is the brother of one of the Rogers (line 29).)

139

The survey of the whole of England to various objections) might possibly have some truth in it; but I do not think that now.

were (so it seems to me) far more drastic than Galbraith imagined. The DB index is a copy of the map which he had made to guide himself through the maze. But then, when he came to write the text, it occasionally happened that he took a slightly different route from the one which he had mapped out for himself in advance.

From Galbraith’s point of view, one of the best examples is the index that accompanies DB-Yo (Table 40, col. 1).50 If we compile our own index for DB-Yo in the normal way, by scanning through the text and making an item for each chapter in turn (that is, if we do what we might expect the DB scribe to have done),51 the result, even supposing that we model it as closely as possible on the actual index, will diverge considerably from it (Table 40, col. 2). Some of the differences are discretionary (such as whether to refer to the count of Mortain by name, or whether to spell this name Rotbert or Robert), and by calling them that we are saying that they do not signify. But there is also some substantive variation. Two chapters are not in the positions where the index tells us to look for them (items 25–6); two chapters occur in the text which are missing from the index;52 and one chapter advertised in the index (item 4) is missing from the text.53

There does not seem to be any excuse that we can make, on the scribe’s behalf, for the omission of chapter 4, ‘Land of the abbot of York’. Galbraith (1961, pp. 198–9) suggested that this chapter was eliminated deliberately, because it turned out that the abbot held none of his lands directly from the king; but that is proved wrong by evidence put on record by the DB scribe himself. His epitome of B-Yo (PRO, E 31/2, fos. 379ra–382rb) confirms what we are told in DB-Yo, that the abbot held some of his land from Berengar de Todeni (314ra);55 but it also reports the existence of five manors held by the abbot directly from the king (explicitly so in three cases, implicitly so in the others), and these manors ought to form the contents of the missing chapter.56 There are some mitigating circumstances – the task was a difficult one (far harder than Galbraith supposed), and this was the scribe’s first attempt at it – but the fact remains that he managed to lose a whole chapter, and that any checking which he may have done afterwards failed to detect the omission.57 Even if every chapter of DB was checked against the corresponding chapter of D, it might still escape attention – and in this case apparently it did escape attention – that a chapter of D had been dropped.

Given the existence of serious discrepancies like these, Galbraith inferred that the DB indexes were not indexes compiled in the normal way, after the completion of the text. Each had to be a copy of a preexisting index; divergences between the index and the text had to be the consequence of changes introduced by the DB scribe during the writing of the DB text. With all of that I agree; I do not see how the facts could be explained in any other way. But Galbraith then jumped to the conclusion that the preexisting index could only have been the index which accompanied the D text, and with that I do not agree.54 Despite their differences in detail, the two indexes shown in Table 40 do not differ in basic design: the structure prefigured by the DB index and the structure displayed by the DB text both show the tight organization which is characteristic of DB, uncharacteristic of D. The changes made by the DB scribe

In principle at least, the method employed for rearranging the contents was reliable. Once he had his draft index to guide him through it,58 the DB scribe could begin on the body of the text. Now the hard work started. For each paragraph in turn, he used the information that he found in D to fill in the blanks of a template that he carried in his head; and then he transferred this filled-in template to the page in front of him. The facts all came, all had to come, from D. The choice of which facts to include, the decision as to how to organize them, the creation of suitable formulas – these are all due to the DB scribe himself. In detail he changed his mind from time to time, and that is what makes it possible to reconstruct the sequence in which the

50 This is the first index written by the DB scribe, and it differs in some details from what was going to become the normal pattern: (i) it occupies the whole width of the column, rather than being divided into two sub-columns; (ii) the first item is the same as the chapter heading, TERRA REGIS, rather than REX WILLELMUS; (iii) the other items are in the genitive case, rather than in the nominative. With respect to point (i) DB-Yo is unique; with respect to points (ii) and (iii) DB-Yo is resembled only by DB-Li. These peculiarities do not affect the issue under discussion. 51

The reader who repeats this experiment will find that it is perfectly straightforward. There is only one place where some pause for thought may be needed. Chapter 2 begins properly at the top of 302va, but a section of this chapter has been supplied on the previous page.

55

The land which he held from a tenant of earl Hugo’s, Willelm de Perci (305ra), is listed as the earl’s (380va), without mention of the abbot or of Willelm.

52

One of these chapters has the look of being an appendix: if this were the only discrepancy, no doubt we should write it off as a special case.

56

In Lestingham (Abb’) ii c’. . . . In Apeltun (Abb’) ii c’. . . . In Spantune . . . (Rex et ab’ de eo) i car’. . . . In Apeltun (Abb’ de rege) ii c’. In Normanebi . . . (Abb’ de rege) iii c’ (380vb).

53

On top of these discrepancies, the numbering of the chapters in the text is doubly defective: it jumps from XIII to XV and from XXIIII to XXXV. The net result is that (apart from chapter 1) there are only five chapters which have the same number in the text as in the index.

57 In DB-Nm he blundered even more seriously than here; but there he noticed his mistake, and put it right, after a fashion (Galbraith 1961, p. 195).

54 Even on Galbraith’s own terms, the conclusion does not really seem to fit. For him, I should have thought, it would make more sense to suggest that the indexes in the surviving D booklets were, like the rubrication, only added as an afterthought – after it had been decided that these three booklets (originally intended to be discarded with the rest) were going to have to be kept.

58 I do not think it important to decide when exactly a fair copy of the index was written into DB – whether it was written prior to the main text, or whether it was inserted afterwards, into a space left vacant for the purpose. Whatever may be true for the fair copy, the draft would have been in existence before the main text was started.

140

The compilation phase 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

TERRA REGIS Archiep’i eboracensis, & canonicorum ac hominum eius Ep’i dunelmensis & hominum eius Abbatis de eboraco Hugonis comitis Rotberti comitis de moritonio Alani comitis Rotberti de Todeni Berengerii de Todeni Ilberti de Laci Rogerii de busli Roberti malet Willelmi de warenna Willelmi de perci Drogonis de heldrenesse Radulfi de mortemer Radulfi pagenel Walterii de aincurt Gisleberti de gant Gisleberti tison Hugonis filii Baldrici Erneis de burun Osberni de arcis Odonis balistarii Ricardi filii Erfasti Goisfridi Alselin Alberici de coci Gospatric Terra tainorum regis

1 2 3 5 6 7 8 9 10 11 12 13 14 15 16 17 — 26 18 19 20 25 21 22 23 24 27 28 29 —

TERRA REGIS Archiep’i eboracensis Ep’i dunelmensis Hugonis comitis Comitis moritoniensis Alani comitis Roberti de Todeni Berengarii de Todeni Ilberti de Laci Rogerii de Busli Roberti malet Willelmi de Warene Willelmi de Perci Drogonis de Beurere Radulfi de Mortemer Radulfi pagenel Goisfridi de Lauuirce Goisfridi alselin Walteri de Aincurt Gisleberti de Gand Gisleberti tison Ricardi filii Erfasti Hugonis filii Baldrici Erneis de Burun Osberni de Arches Odonis arbalistarii Alberici de Coci Gospatric Tainorum regis Rogerii pictauensis

Table 40. Two indexes for DB-Yo: the index provided by the DB scribe (298vb) and an imaginary index, modelled on the first one, representing the actual contents of the text. booklets were written (Table 41).59 Even so, he achieved a high degree of consistency, from paragraph to paragraph, from chapter to chapter, from county to county. And he did it all himself. Once, just once, another scribe had the temerity to add a short paragraph (above, pp. 30–1); but he was promptly chased away and never allowed to touch the manuscript again (at least not till after the DB scribe had finished with it). The more one understands what the DB scribe’s job involved, the more respect one has to feel for him. Above all, one admires his stamina. Month after month after month, he kept working away at his task, until it was nearly finished.

scribe’s control. In several booklets, for example, more or less frequently, we find blanks in the text where we expect to be given the potential number of ploughs – the number of ploughs that would exist on this manor if it were being exploited to the maximum. Where this information is missing, the presumption is that it was missing from D, and perhaps from B as well. The fact that the DB scribe left a blank – advertising the deficiency rather than disguising it – tends to prove two things: that he (like scribe mu) attached some especial importance to this number; and that he entertained some idea that the missing information might become available. Apparently he thought it possible that some further inquiry might take place. Then again, it has often been remarked – originally, I think, by Maitland (1897, pp. 177–8) – that there is blank space at the beginning of DB-Mx which looks as if it was intended for a description of London, blank space at the beginning of DB-Ha which looks as if it was intended for a description of Winchester. It is possible that the DB scribe ought to have written these descriptions but delayed doing so until he ran out of time. But it is also possible that he had been told to leave spaces here – that the original survey had failed to produce an adequate description of London or of Winchester, and that some further inquiry was under contemplation. In the end, those signs of incompleteness which are certainly the

Nearly but not quite finished. There are various indications which seem to prove that the scribe did not succeed in completing the manuscript to his own satisfaction. Some of these deficiencies were, very probably, beyond the DB 59

Though I have continued trying, I have still not been able to work out a satisfactory seriation for the aspect 2 booklets (above, p. 17). One feature which I have looked at is the spelling of the [ly] sound in French, particularly in the name tailgebosc. The spellings that occur in DB-Ca are the same that occur in DB-Bd, but DB-Ht and DB-Bu use the -ilg- spelling which is standard from this point onwards (as in batailge). It seems from this that we should seriate either MxBdCaHtBu or MxBdCaBuHt; hence the adjustment which I have made in this table.

141

The survey of the whole of England aspect 1 1 1 2 2 2 2 2

3 3 3 3 3 4 4 4 4 4 5 5 5 5 5 6 6

counties Yorkshire–Lincolnshire Nottinghamshire–Derbyshire Huntingdonshire Middlesex Bedfordshire Cambridgeshire ? Hertfordshire ? Buckinghamshire ? ? ? Kent Sussex Surrey Hampshire Berkshire Gloucestershire–Worcestershire Herefordshire Shropshire-Cheshire Staffordshire Warwickshire Northamptonshire Leicestershire Oxfordshire Wiltshire Dorset Somerset Devon–Cornwall

binding sequence

17th-century foliation

alternative foliation

26 25, 24 17 10 18 16 11 12

297–372 280–96, 272–9 203–8 126–31 209–18 189–202 132–42 143–53

1–76 77–101 102–7 108–13 114–23 124–137

1 2 3 4 5 14 15 23 22 21 19 20 13 6 7 8 9

0–15 16–29 30–6 37–55 56–63 162–78 179–88 252–71 246–51 238–45 219–29 230–7 154–61 64–74 75–85 86–99 100–25

160–75 176–89 190–6 197–215 216–23 224–40 241–50 251–70 271–6 277–84 285–95 296–303 304–11 312–22 323–33 334–47 348–73

Table 41. Revised seriation of the DB booklets. DB scribe’s fault turn out to be very few: an entry in DB-Do which breaks off in mid sentence (above, p. 31), marginal additions here and there which lack their finishing touches of red ink. Small imperfections of this kind – such as spaces for coloured initials which were never supplied – occur in many medieval manuscripts, and generally no one would think that they carried much significance. Nevertheless, the flaws do visibly exist; and they mean that it must have been doubtful for a time, to anyone who looked at DB, whether the manuscript was still a work in progress, or whether it had already been finished, as far as it ever would be.

reference to Essex, for example, in DB-Ke-9ra – we might perhaps infer that the scribe was at least intending to deal with them too; but we would probably be willing to take that much for granted, without asking for evidence. The question is whether the intention was fulfilled. The more I think about this, the more inclined I am to believe that all three booklets were written, but that one or all of them went astray, during the span of time that elapsed before DB was bound.60 Unfortunately, the more I think about it, the less able I am to see how the point could be proved either way. It is unimaginable how we could prove that the booklets never existed. It would take a miracle to prove that they did once exist. There might survive – from Ely or Bury or some other monastery – a copy of some extract from DB-Sk, which in substance would agree with the corresponding paragraph in D-Sk, but which in form and

The chief deficiency – the absence of booklets for three whole counties – is of a different order of magnitude. It does, to be sure, demand some explanation. If we came across a manuscript copy of the Bible which omitted Exodus, Isaiah and Luke, we should certainly want to know why. The answer might turn out to be bathetic, but the question would have to be put. Here, however, we do not even know what question we ought to ask. The fact that the booklets do not exist does not prove that they never existed. Are we to explain why these booklets were not produced? Or are we to explain why they were not bound up with the other booklets, when DB was made into a book? From references to these counties in some of the other booklets – a

60 It is not inconceivable that the difficulties encountered in seriating the aspect 2 booklets (above, note 59) are partly due to the loss of the following booklets. For the sake of argument, suppose that DB-Ht was the last of the aspect 2 booklets to be written, and that the next booklet to be written was DB-Ex (which, for one reason or another, does not survive). Comparing DB-Ht with DB-Ex, it might be fairly easy to identify some features of the text which start appearing in the former and persist into the latter. With the loss of either booklet, it would become much harder to see which features carry the information that we need.

142

The compilation phase wording would proclaim itself to be the work of the DB scribe. But any such extract, if it existed, would surely have been discovered and published by now. The miracle is not going to happen, and I think we have to accept the fact that we are doomed to everlasting ignorance. There are two options. Wailing and gnashing of teeth is one; the other is to think ourselves into the frame of mind where this becomes just one more of the many things that we should like to know but never shall. I recommend option B.

Hence it appears that the man in charge of the D text held off from having the booklets bound until after the DB scribe had finished with them. It was only then that he sorted the booklets into six stacks, putting them into a new order, very different in some respects from the order of completion, and sent these stacks off to the binder. If it is safe to infer, from the wording of the colophons in the D volumes (supposing the sole survivor to be typical), that these colophons were written and the volumes bound while the king was still alive, that will give us a bound on the date of DB: it will follow that DB was completed sooner than September 1087. By this argument, the whole operation, from the inception of C to the binding of D, would have been completed within less than eighteen months, perhaps not much more than twelve months; and that span of time would have included – would have purposely been prolonged to include – as much time as was needed for the writing of DB.

Towards the beginning, the DB scribe may have been hard on the heels of the D scribes, taking delivery of each new D booklet almost as soon as it were finished. Towards the end, he must have been well behind. The last D booklets to pass through his hands had probably been finished several months before. Even so, the order in which the scribe dealt with these counties is correlated closely with the order in which they had been dealt with by the C scribes:

C: D: DB:

Though the argument is tenuous, the conclusion is unobjectionable, as far as I can see. The compilation process was the expression of innumerable small decisions, but these were all consequential on two large ones. At some moment it was decided that the results of the survey would need to be translated, county by county, into a feodal frame. At some moment (perhaps the same, perhaps later) it was decided that the feodalized results would need to be condensed into a single volume. In the end it can only be a guess that these decisions, the two large ones and all the small ones which followed from them, were concatenated into a single sequence of events; but is this not a better guess than any other?

..... Nn Wa St Ox Wi Do Dn Co So So .. St Wa Nn Le Ox Wi Do So Dn Co

and for that to be true it must also have been correlated, not less but perhaps more closely, with the order in which they had been dealt with by the D scribes. To put it briefly, the completion order for the last D booklets cannot have been very different from the inception order for the last DB booklets.61

Consider the alternatives. First, one might agree that the compilation process was a single sequence, each stage being integrated with the stage before it, and yet disagree about the time-scale. I doubt whether anyone will think it possible that the process might have been completed in a shorter time than I suggest, but it might have taken much longer.63 Second, more radically, one might question the assumption of concatenation. Instead of a single process, one might argue that there were two or three separate stages, each of which was thought for a time to be the final stage. The survey was conducted, the B text was compiled, and that was the end of the matter. Some time later (and possibly this would mean in the reign of Willelm II), it was decided that the information contained in B was of little use as it stood, but would still be useful, despite the lapse of time since its collection, if it could be put into a different frame. By way of the C text, the D text was compiled, and again that was the end of the matter. Some time later again (and by now perhaps we are in the reign of Henric I), it was decided that D was of little use as it stood, and that the information, so far as it still had value, should be condensed into a single volume. The DB text was compiled, and once again that was the end of the matter (until peo-

However, we have reason to think that the D booklets did not retain that order when they were bound (above, pp. 29– 30). Five of these booklets went to make up one volume by themselves (D-WiDoSoDnCo); four were kept together and included in another volume (D-CaHuBdNnLeWaStShCh); but D-Ox became separated from the rest, becoming part of a third volume (D-MxHtBuOxGlWoHe). At the moment when it came into the DB scribe’s hands, D-Ox was still keeping company with the D booklets completed at about the same time as it: it had not yet been bound, nor even put into proximity with those other booklets with which it was going to be bound. If that was true for D-Ox, it must also have been true for six other booklets at least (D-Mx, D-Ht, etc.); and if that much is true, it does not seem risky to assume that the D booklets were all still unbound when the DB scribe got hold of them.62 61 It need not be assumed that the correlation was perfect, only that it was close. In fact the signs are that D-So was the last booklet to be completed by the D scribes, just as C-So had been the last text to be begun by the C scribes, and that the order here was changed by the DB scribe (above, p. 79). 62

For Galbraith, the D booklets were not intended to be bound at all (they were meant to be discarded as soon as their contents had been distilled into DB); so for him it went without saying that the DB scribe, when he dealt with D, was dealing with loose quires. This is the closest that I can come to ending (as I should like to do) on a note of agreement with Galbraith.

63

Of course there is more at issue here than just the date of DB. Would the machinery of government have come to a stop, as Galbraith was inclined to assume, as soon as a king died? Or would it have continued functioning, to some degree, under its own momentum?

143

The survey of the whole of England ple started making epitomes of DB). Is there anything here which seems at all attractive?

There is evidence from Kent which seems to me to prove that DB existed (or, if one wishes to put it very cautiously, that at least one DB booklet had at least been begun) within three years of the survey. The evidence comes from that same collection of documents (above, p. 111) which originated (so I suppose) in archbishop Lanfranc’s chancery. One of them, perhaps just a small slip of parchment, had been used for writing out a few short paragraphs which can be identified – beyond doubt, I think – as edited extracts from the first few pages of DB-Ke (above, p. 34). It seems safe to conclude that DB (or part of DB) was already in existence, no later than May 1089, and that it was already possible for someone with the right connections to gain access to the original and to copy extracts from it.

On one point I refuse to budge. It seems to me quite clear, from the evidence of the surviving C booklets, that concatenation is a valid assumption for C and D. On the one hand, we can be sure – we can almost see it happening – that the production of C was coordinated with the progress of the fieldwork stage of the survey, and with the writing up of the accounts for the current geld. On the other hand, the production of C makes no sense in itself, only as a step towards the production of D. If that much is agreed, the scope for disagreement is reduced to two points. First there is a quantitative question. Counting from the moment when it was decided that the survey should be started, how long did it take before the D text was completed – so far completed that it was ready to be delivered to the binder? Was the work done in the shortest practicable time, i.e. within a matter of months, or was it allowed to drag on for several years? One question arising from that is whether we accept the inference that the colophon of D-ExNkSk, because it seems to see no need to distinguish between Willelm I and Willelm II, was written no later than September 1087.64 Of course there is room for doubt (an anticoncatenationist might think of suggesting that this colophon was copied from the matching B volume), but do we feel at all inclined to take advantage of that room?

Another document included in the same collection is also derived from DB. As it survives (Lit. E 28, fos. 5vc–7ra), it is a copy of an epitome of DB-Ke reorganized to take account of the fact that the bishop of Bayeux has forfeited his lands. Therefore it is later than May 1088. The author of this text was not trying to bring the survey up to date; only a new survey could do that. Instead he is asking the question what DB-Ke would have looked like, in outline, if the bishop had lost his lands before the survey, rather than shortly afterwards. More specifically, he is trying to work out what assets have fallen into the king’s hands in consequence of the bishop’s failed rebellion – what assets have thus become available for redistribution. None of the property repossessed by the king in 1088 remained his for very long. Within twenty or thirty years at the most, it had all been granted out again. It is clear enough who the new men were who profited from the fall of the bishop of Bayeux – Goisfrid Talebot, Willelm Pevrel, Willelm de Albigni, Walter Tirel – but little is known about the circumstances in which they came into their reward.66 If one reads this document closely, however, one can see, from the way in which the paragraphs are grouped, that the author already had some idea how the loot was going to be shared out; if it is right that a copy of this text was in archbishop Lanfranc’s hands by May 1089, it will seem likely that Goisfrid Talebot and his friends were not kept waiting long.67 In all this there are many points which need to be argued out in detail, and I hope to have a chance to do that in the future; but the conclusion seems solid enough that I am willing to anticipate it here. After but probably very soon after May 1088, somebody tried to work out what consequences followed from the forfeiture of the bishop of Bayeux; and the source-text to which he turned for that purpose was DB-Ke.68 In my

Then there is a qualitative question: was the production of DB concatenated with the production of D, or was it a separate undertaking, decided on and executed at some later (perhaps considerably later) date?65 There is, as I have said, some slight reason for thinking that the DB booklets were written before the D booklets were bound; but that evidence will not convince anyone who is predisposed to doubt it. More impressive is the fact that the DB scribe is known to have laid his hands on some of the C booklets, and to have occasionally used a blank page in C for writing a trial version of some paragraph which he was about to transfer from D into DB (above, pp. 76–9). But of course we do not know exactly when or exactly how that came to pass. It is conceivable, I suppose, that the C booklets were, by this time, ten or twenty years old and just happened to be lying around. But do we have any motive for preferring that to the more obvious interpretation, that all three versions of the text were, in the DB scribe’s own view of the matter, links in a single chain?

66 One by-product of these changes is a writ of Willelm II (Davis 1913, p. 133) responding to a complaint from the abbot of Saint Augustine’s; it dates from 1088 93. Unfortunately only one of the culprits is mentioned by name, Anskitil (de Ros), and he was not a newcomer.

64

The phrase at the heart of the colophon – facta est ista descriptio – cannot be made to mean anything more than what Round took it to mean: ‘This survey was carried out’. It gives a date for the B text; it does not give a date for the D text.

67

The English chronicler takes it for granted that confiscation was followed by immediate redistribution. After the rebellion had collapsed, ‘many French men relinquished their lands and travelled across the sea; and the king gave their lands to the men who were loyal to him’ (Swanton 1996, p. 225).

65 ((The idea that the date of DB is given away by a slip of the pen in DB-Sx is too silly to be worth discussing; I am reluctant even to mention it. The obvious explanation – as Thorn and Thorn (2001, p. 71) point out – is that the scribe momentarily forgot which chapter he was in.))

68

Unlike this epitome, DB-Ke itself has no anachronistic features. Having read it closely, many times, I detect not the slightest hint that the DB scribe

144

The compilation phase judgment it is safe to say that DB was in existence by the summer of 1088.69 For most of the people who have occasion to look at DB, the date when it was written does not signify. What they want to know is precisely what the DB scribe was trying to tell them: how things stood at the time of the survey, in early 1086. They read it, in short, as the best available substitute for B. In the end, however determined we are to try to look on the bright side, we cannot quite resist the thought that Maitland was justified in feeling disappointed. What we have got is not what we would have chosen. If we had been told that only one version of the survey text was going to survive and that we could choose which one, there is no doubt what decision we would have made. Without hesitation, we should have chosen the B text. Given B, we could, if we wished, construct our own feodalized version of the survey to take the place of D. Because we have improved technology, and because we are not in a hurry, our version of D would actually be better than the version produced by the D scribes. Given our own version of D, we could, if we wished, construct not just one but any number of shortened versions, each designed for some particular purpose. We could decide for ourselves what information should or should not be included, and what format will best suit our purpose, rather than being forced to accept the decisions made by the DB scribe. But these are all dreams. We have not got the B text. Instead we have got small parts of C and D, which are fairly good substitutes for B, plus a large part of DB, which is not. Nor can we comfort ourselves, as Maitland did, by imagining that it may become feasible, at some future date, for B to be reconstituted. That will never happen. What was done cannot be undone, and we have to make the best of it. But if the time and effort invested in the making of D and DB had rather been spent making multiple copies of B, if those copies had been distributed among the bishops and abbots who, as far as was humanly possible, could guarantee their safety, if even one complete contemporary copy of B had managed to survive till now, how much happier should we not be?

was in possession of any knowledge which he could not have possessed in 1086–7. 69

The evidence from Kent thus seems to me to bear out a knot-cutting comment made by Stenton: ‘On general grounds, there is an overwhelming probability that the [two] volumes [DB as well as D-ExNkSk] were written before the information which they contain was seriously out of date; that is, before, at latest, the confiscations after the revolt of 1088’ (1943, p. 647 = 1971, p. 655).

145

Appendices

Appendix I An outline reconstruction of B-Ca

If we are willing to resort to numerological arguments of the kind developed by Round (1895), we can advance a little further. Round himself worked things out for Longstow hundred (1895, p. 48), and the reconstruction below is based on his; but in fact there is more than one way of arranging these villages so that they fall into blocks of 25 hides, and Round’s way is (as far as I can see) not demonstrably the right way.1 If Round’s arrangement is accepted, the order of the villages is fixed (within one block by chapter 26, within the other block by chapter 25, between blocks by chapter 14), except that Eltisley cannot be ordered with respect to Croxton.2

The following list is an epitome of the surviving portions of the B text for Cambridgeshire combined with a reconstruction of the missing portions. By and large it agrees with the listing given by Hart (1974, pp. 47–67). All the way through, the paragraphs are equated with the corresponding paragraphs of DB-Ca, as they were numbered by Rumble (1981); the identifications of the places named are all taken straight from that edition. A very few entries seem to have been lost somewhere along the line between B and DB (i.e. in C, D or DB): the notation ‘10/—’ means that a matching paragraph ought to occur, but apparently does not occur, in DB’s chapter 10. Square brackets denote those manors belonging to the king which were omitted or cancelled from B (paras. 1/1–8, presumably also 1/9). Bold type distinguishes the entries which are represented in xEl, sometimes in a form which agrees word for word with B, but sometimes only in the form of an excerpt, more or less heavily edited. Of the fifteen hundreds in Cambridgeshire, the monks of Ely had some interest in every one except Whittlesford.

The same sort of argument works well on Chesterton hundred, which can be shown to consist of four blocks of 30 hides each (Hart 1974, p. 66). Chesterton itself, belonging to the king, makes one block of 30 hides; presumably it ought to come first, but there is no proof of that (and Hart preferred to put it last). Cottenham adds up to 26 hides – the abbot of Ely has 10 (para. 5/42), the abbot of Crowland has 11 (para. 9/1), one of Picot’s men has 5 (para. 32/40) – and these combine with the 4 hides in Westwick to make a block of 30; Cottenham comes before Westwick on the evidence of chapter 32. Drayton and Childerley, 20 and 10 hides respectively, combine to make another block of 30; Drayton comes before Childerley on the evidence of chapter 41. Histon adds up to another block of 30 hides; 3 it comes after the Cottenham–Westwick block on the evidence of chapter 5, before the Drayton–Childerley block on the evidence of chapter 3.

Stretches of text that are missing from V are printed in italics. Bold type continues to denote the entries represented in xEl; plain italics denote the entries that are only to be found in DB. The reconstruction is based on the evidence of these two texts, xEl and DB, and is, at best, only as reliable as they are. In some parts and in some respects it is fairly solid; in other parts, in other respects, it is not. I explain very briefly how the reconstruction was put together, so that readers can judge for themselves how far it is to be trusted.

Papworth hundred is the most difficult case, and numerology does not seem to help.4 DB’s chapter 23 fixes the order for five of the missing villages: Papworth, Elsworth, Conington, Boxworth, Swavesey. Except for Swavesey, this sequence is confirmed by chapter 26. Graveley belongs before Elsworth, on the evidence of chapter 7, but cannot be ordered with respect to Papworth; Knapwell belongs between Elsworth and Boxworth, on the evidence of the same chapter, but cannot be ordered with respect to Conington. To complicate things further, there is also one stray hide here (para. 41/11).

(1) The order of the first thirteen hundreds is known from V. As for the last two, it is clear from xEl, and from chapters 5 and 9 in DB, that Chesterton should come before Ely. (2) Within a given hundred, the order of the villages is reconstructable so far as it is vouched for either by entries in xEl or by strings of entries in single chapters of DB. For Ely hundred, where the monks of Ely owned almost everything (and were interested in knowing about the rest), xEl supplies all the answers. For Northstow hundred, where sheriff Picot happened to own part of every village, DB’s chapter 32 contains as much information as is needed. That leaves three hundreds which are not so easy to deal with: Longstow (of which the beginning survives), Papworth (of which the end survives), and Chesterton. In all three, the evidence is too sparse to fix the order satisfactorily.

1

As far as the arithmetic goes, Gamlingay (20 hides) might be combined with Caldecote and Longstowe (5 hides) just as well as with Hatley.

2

For a travelling salesman, Eltisley would come between Caxton and Croxton; but I am doubtful whether this argument is allowable.

3 The discrepancy reported here by Hart was of his own making: he was misreading the fraction 2/3 as 3/4. 4

Hart’s (1974, pp. 63–4) attempt at a reconstruction leaves me unconvinced.

146

Appendices (Cherry) Hinton: 14/2 Teversham: 14/3, 14/4, 35/2, 5/13 Horningsea: 5/14

(3) Within a given village, the order of the manors is reconstructable so far as it is vouched for either by entries in xEl or (but this does not happen often) by strings of entries in single chapters of DB. Except in Ely hundred, the evidence is generally inadequate. In Oakington, for instance, five manors are listed. Three of them are mentioned in xEl, and for these three the order is fixed. But the other two are mentioned only in DB, in different chapters, and these two manors cannot be ordered with respect to one another, nor with respect to the first three. This means that there are 5! / 3! = 20 ways of ordering these five entries, and we have no means of knowing which is the right one. I list the entries missing from xEl at the end of the line, in the sequence in which they occur in DB; but this, in relation to B, is an arbitrary order, and nothing can be done to improve it.

CHILFORD HUNDRED (78vb–81vb)

Camps: 29/7 (lines 1–6), 29/7 (lines 7–8), 21/1 Horseheath: 14/5, 29/8, 26/9, 14/6, 19/2 Hildersham: 29/9 Abington: 29/10, 1/16 (West) Wickham and Streetly: 14/7, 18/6, 5/15, 26/10, 19/3 Barham: 14/8, 14/9, 5/16 Linton: 14/11 (Little) Linton: 14/13 Abington: 14/14 Babraham: 14/15, 14/16, 38/1, 1/15, 29/11, 26/11 + 5/17, 25/1, 41/5 Pampisford: 5/18, 14/17, 32/3, 5/19, 25/2, 26/12, 41/6

STAPLOE HUNDRED (71ra–3rb)

Kennett: 18/8 Badlingham: 14/67 Chippenham: 22/6 Snailwell: 28/2 Exning: 14/68, 1/12 Burwell: 7/9, 14/69, 14/70, 11/2, 26/2 Soham: [1/1], 5/8, 14/73, 6/3 Fordham: [1/2], 14/71 Isleham: [1/3], 4/1, 28/1, 14/72 Wicken: 14/74

WHITTLESFORD HUNDRED (81vb–3ra)

Whittlesford: 41/7, 14/18 Sawston: 25/3, 22/2, 12/1 Hinxton: 32/4, 3/1, 1/10, 26/14 Ickleton: 15/1, 26/15 Duxford: 15/2, 14/19, 20/1, 26/16, 21/2 THRIPLOW HUNDRED (83ra–5va)

Thriplow: 5/20, 5/21, 5/22, 22/3 Fowlmere: 21/3 Foxton: 11/1, 22/4–5, 14/20 Harston: 32/5, 21/4, 14/21 Hauxton: 5/23, 26/17 Shelford: 5/25, 26/18, 14/22, 1/17 Stapleford: 5/28 Trumpington: 18/7, 32/6, 15/3, 38/2, 41/8, 1/—

CHEVELEY HUNDRED (73rb–4ra)

Silverley: 29/3 Ashley: 29/1 Saxon (Street): 29/2 (Wood)ditton: 14/61 (Wood)ditton: 1/11 Kirtling: 41/1 Cheveley: [1/4], 14/62

ARMINGFORD HUNDRED (85va–90va)

(Steeple) Morden: 2/1, 13/1, 26/19 Tadlow: 32/7, 41/9, 42/1 (Guilden) Morden: 32/8, 13/2, 22/7, 26/20 Clopton: 2/2, 25/4, 32/9 (East) Hatley: 32/10, 14/23, 25/5 Croydon: 14/24, 14/25, 32/11, 32/12, 25/6, 26/21, 13/3 Wendy: 14/26, 26/22 Shingay: 13/4 Litlington: 1/18, 26/23 Abington (Pigotts): 2/3, 1/19, 26/24, 13/5, 32/13, 1/20 Bassingbourn: 14/27, 2/4, 26/25 Whaddon: 26/26, 19/4, 26/27, 14/28, 14/29, 14/30, 14/31, 26/28 Meldreth: 13/6, 14/32, 31/1, 26/29, 5/31, 26/30 Melbourn: 5/34, 31/2, 14/33, 26/31, 13/7

STAINE HUNDRED (74ra–5va)

Bottisham: 17/1 Swaffham: 17/2, 14/63, 29/4 Swaffham: 5/9, 5/10, 17/3, 26/1, 14/64 Wilbraham: [1/5], 14/65, 29/5 Quy and Stow: 14/66, 29/6, 5/11, 32/1 RADFIELD HUNDRED (75va–7vb)

Dullingham: 10/1, 26/3, 14/75, 41/2 Stetchworth: 5/1, 5/2, 14/76, 10/— Burrough (Green) and Westley (Waterless): 5/3, 41/3, 26/4, 14/77, 14/78 Carlton: 41/4 (lines 1–5), 18/1, 18/2, 26/5, 14/79, 41/4 (lines 5–7) Weston (Colville): 18/3, 14/80, 26/6, 26/7, 18/4 (West) Wratting: 5/4, 5/5, 26/8, 14/81, 18/5 Balsham: 5/6, 5/7, 14/82

WETHERLEY HUNDRED (90va–5rb)

Comberton: [1/6], 32/14, 37/1, 44/1 Barton: 37/2, 12/2, 31/3 Grantchester: 31/4, 15/4, 38/3, 14/34, 12/3, 32/15 two mills: 14/35

FLENDISH HUNDRED (77vb–8vb) Fulbourn: 14/1, 35/1, 5/12, 22/1, 1/14

147

The survey of the whole of England Haslingfield: [1/7], 32/16, 14/36, 14/37, 22/8, 31/—, 14/38 Harlton: 17/4, 32/17 Barrington: 21/5, 11/3, 32/18, 17/5, 14/39 Shepreth: 26/32, 11/4, 22/9, 14/40, 26/33 Orwell: 13/8, 26/34, 22/10, 14/41, 17/6, 21/6, 31/5, 11/5 Wratworth: 13/9, 26/35, 14/42, 32/19, 31/6 Whitwell: 32/20, 13/10, 26/36, 14/43 Wimpole: 14/44, 25/7 Arrington: 13/11, 14/45

ELY DOUBLE HUNDRED

Whittlesey: 5/44, 8/1 Doddington with March: 5/45, 6/1 Chatteris: 5/46, 7/11 Littleport: 5/47 Stuntney: 5/48 Little Thetford: 5/49 Stretham: 5/50 Wilburton: 5/51 Linden (End): 5/52 Hill (Row): 5/53 Haddenham: 5/54

LONGSTOW HUNDRED (95rb–6vb . . . )

Eversden: 31/7, 14/46, 27/1, 26/37 Kingston: [1/8], 32/21, 14/47, 13/12, 25/8, 26/38, 26/39 Toft and Hardwick: 5/36, 14/48, 32/22, 44/2, 5/37 (Little) Gransden: 5/38 Bourn: 32/23, 14/49, 7/1, 33/1 Caldecote: 14/50, 26/40, 39/1 Longstowe: 7/2, 14/51, 26/41 Caxton: 26/42 Croxton: 26/43, 39/2 Eltisley: 16/1 Gamlingay: 25/9, 34/1, 38/4 Hatley (St George): 14/52, 25/10, 32/24–5, 41/10

Wisbech: 5/55, 7/12, 9/4, 18/9, 6/2, 5/56 Ely: 5/57 Hainey: 5/58 Downham: 5/59 Witchford: 5/60 Wentworth: 5/61 Witcham: 5/62 Sutton: 5/63 After Ely hundred, xEl continues with some excerpts from a survey of the town of Cambridge, distantly parallel with DB-Ca-189ra, and concludes with a single sentence which seems to derive from some statement of the profits accruing to the king as king (rather than as the owner of certain manors) from the county in question.5 DB has a statement of this kind for some counties, but not for Cambridgeshire.

PAPWORTH HUNDRED (. . . 97ra–8ra)

one hide: 41/11 Papworth: 30/1–3, 19/1, 14/53, 23/1, 26/44, 32/26 Graveley: 7/3 Elsworth: 7/4, 23/2, 26/45 Conington: 21/7, 23/3, 26/46 Knapwell: 7/5 Boxworth: 7/6, 14/54, 21/8, 23/4, 26/47 Swavesey: 14/55, 21/9, 23/5 (Fen) Drayton: 14/56, 7/7, 32/27, 1/21, 23/6 Over: 7/8, 26/48, 11/6, 32/28, 41/14 Willingham: 5/39, 14/57, 32/29

Appendix II Summaries compiled in the Treasury The statistical summaries discussed in chapter 9 (above, pp. 103–6) were first put into print by Ellis (1816). For the benefit of readers who do not have easy access to a copy of that edition, I print the two relevant stretches of text again, by permission of the Dean and Chapter of Exeter and the Master and Fellows of Trinity College, Cambridge, respectively.

NORTHSTOW HUNDRED (98ra–vb . . . )

(Long)stanton: 24/1, 14/58, 36/1, 32/30 Rampton: 32/31 Lolworth: 32/32 Madingley: 32/33, 3/2, 41/12 Girton: 7/10, 12/4, 32/34 Oakington: 43/1, 5/40, 32/35, 9/1, 41/13 Impington: 5/41, 32/36 Milton: 32/37 (Land)beach: 14/59, 32/38 (Water)beach: 32/39, 40/1

Editorial interference has been kept to a minimum. The first document, written by scribe mu, can answer for itself. The second, surviving as a later copy made by a scribe at Ely, is not in such good shape. I have marked two significant omissions, one in the middle of paragraph 3 and the other at the end of paragraph 12.6 But what will catch the eye most is the fact that in many places (and not just in the paragraphs 5

In prouincia Granteb’ reclamat abbas quartum nummum, ut carte sue testantur, et homines de scira (xEl / T-110v, ed. Hamilton 1876, p. 121).

CHESTERTON HUNDRED

Chesterton: [1/9] Cottenham: 5/42, 32/40–2, 9/2 Westwick: 39/3, 32/43 Histon: 3/3–5, 5/43, 12/5 (Dry) Drayton: 26/49, 9/3, 14/60, 38/5, 41/15 Childerley: 3/6, 32/44, 41/16

6 The omission in para. 3 was caused by a jump car’ . . . car’. In U a corrector has added this passage in the margin: xxx (uill’) lx (b) v (s) h’nes x carr’. (There is no note of this in Hamilton’s edition.) At first sight this seems to prove that the omission originated in T, and that the corrector had access to a better copy (such as T’s exemplar) from which he could supply the missing words. But does it really prove that? These look to me like imaginary numbers, too round to be genuine. The corrector, I think, found himself in a quandary. He could see, just as we can, that there was something missing here, and did not want to let the text be copied in this defective form. But he had no way of mending it unless he resorted to

148

Appendices

d’nio & xvi uill’ & iiii bord’ & iiii serui habentes v carr’. H˛ec t’ra appreciata e’ iiii lib’. H˛ec t’ra sufficit vii carr’.

directly of interest to Ely) alternative readings are given for the number of manors belonging to a certain category. The scribe writes a number, cancels it, and writes another number above it. Though I cannot explain this, I have two comments to make. (1) It is, I think, quite clear that these are not corrections originating with scribe T1: he was reproducing what he found in his exemplar (and probably it puzzled him as much as it puzzles us). (2) One of the alternative readings, usually the second one, is a dummy number, x or xxx or iiii. Though scribe T1 regarded these as numerals, I would guess that they were originally something else – cancellations made with either criss-cross or vertical lines. The question to be asked, then, is why it was decided that these numbers should be cancelled.

Exeter Cathedral Library 3500, fos. 527v–8r [1] Æccl’a Glastiniensis h’t xi mansiones IN WILTESCIRA ccc & xxxvii hidarum, in his sunt xlii carruc˛e in d’nio & c & xii uillani & septies xx & ii bordarii & ii burgenses & xxxviii coliberti & xxxv serui habentes lx & ix carr’ & dimid’. H˛ec terra appreciata e’ clxix lib’ & x sol’. / De eadem æccl’a tenent milites iiii mansiones de xxvii hid’, & preterea xxvii hid˛e in mansionibus abb’tis. In his s’t xvii carr’ in d’nio & xxxiiii uillani & xviii bord’ & ix serui habentes xi carr’. H˛ec t’ra militum ualet lv lib’ & v sol’. H˛ec terra sufficit septies xx & xvi carr’ & dimid’. H˛ec terra emendata e’ liii lib’ & x sol’. / De predicta terra h’t i tagnus hidam & dimid’ & i carrucam & ual’ xx sol’. [2] DORSETA. Eccl’a ˛ glastiniensis h’t ii mansiones d’nicas in dorseta de xxii hid’ & iii uirgis & dimid’. Ibi s’t xxii carrucat˛e terr˛e non Gheldantis. Ibi s’t ix carruc˛e in d’nio & xliii uillani & lxxii bordarii & xix serui & xiii coliberti habentes xx carr’. H˛ec t’ra appreciata e’ xlv lib’. / Milites abbatis h’nt in dorseta v mansiones de xxxi hid’ & uirga & dimid’. In his s’t xiiii carruc˛e in d’nio & xxxvii uillan’ & xlvi bord’ & xix seru’ habentes xxi carr’. H˛ec t’ra appreciata e’ xxix lib’. / Duo tagni tenent de predicta t’ra i manerium v hidarum. In his s’t ii carr’ in d’nio & xii uill’ & xxvi bord’ & iiii serui habentes iii carr’. H˛ec t’ra appreciata e’ iiii lib’ & x sol’. H˛ec t’ra sufficit c & v carr’ & e’ peiorata de xl sol’. [3] DEVENESIRA. Æccl’a glastiniensis h’t i mansionem de vi hid’ in Deneuesira. In his s’t ii carr’ in guesswork – unless he made up some plausible numbers of his own. In the end he did not quite commit himself. He wrote this passage in the margin; but he did not go so far as to mark up the text, telling the copyist where it should be inserted. As things turned out, the fair copy of xEl was never made, perhaps because the same sort of quandary occurred too often. The text was in bad shape; there was no proper way of repairing it; so was there any point in making a fair copy? Scribe U3 did not think so; nor did the scribe of HEI / W.

[4] SOMERSETA. Æccl’a glastiniensis h’t xx mansiones d’nicas in somerseta nouies xx & xiiii hidarum & iii uirgarum. Preterea s’t ibi xxiii hid’ & xl carrucat˛e t’r˛e non gheldantis. In his s’t lxxv carruc˛e & dimid’ in d’nio & xlvii uill’ & ccc & xxv bord’ & c & viii serui & xix coliberti & x piscatores habentes clx carr’. H˛ec t’ra appreciata e’ cc & quat’ xx & viii lib’ & iiii sol’. / Milites uero e˛ ccl’˛e h’nt in eod’ comitatu l mansiones de clx hid’ una minus. In his s’t lxiiii carruc˛e & dimid’ in d’nio & cc & xix uill’ & cc & xlix bord’ & c & i seruus & ii coliberti habentes c carr’ dimid’ carr’ minus. H˛ec t’ra appreciata e’ cxlvi lib’. / De predicta terra tenent tagni xi mansiones de xxx & i hid’ & dimid’ & ii carrucatis t’r˛e non gheldantis. In his s’t xii carr’ & dimid’ in d’nio & xxi uill’ & xlviii bord’ & xvi serui habentes xi carr’. H˛ec terra appreciata e’ xx lib’ & xv sol’. H˛ec t’ra sufficit D & liiii carr’ & dimid’. H˛ec t’ra emendata e’ in manu turstini abbatis c & xxviii lib’. Cambridge, Trinity College O. 2. 41, fos. 110v–13r [1] Abb’ eli in Granteb’gescira in d’nio h’t xxxiiii Man’ de c & quat’ xx & xiiii h’; quat’ xx & v car’ & dm’ in d’nio; D & xxii uill’i; cc & quat’ xx 7 xviii bord’ c lxi ser’ h’ntes c & quat’ xx & v car’ & dm’. Hoc totum ualet ccc & xviii lib’ & iii sol’. H˛ec t’ra sufficit cc & quat’ xx & xviii car’, emendata in manu abb’is Sy liiii lib’. [2] IN eod’ comitatu h’nt Pi uicecom’ & Hard’ de escalariis & Wido de Rainbucurt & alii ho’es de Thainlandis e˛ ccl’i˛e eli de quibus seruiunt abb’i concessione regis (xl)c (xiijij)i Manerium de xxxii h’ & dm’ v’, xiii car’ & dm’ in d’nio; xxxvi uill’ xxxvi b’ ix ser’ h’ntes xi car’ & dm’, & ualent xxvi lib’ & v sol’ & viii dn’. H˛ec t’ra suffic’ xxviii car’ & ii bobus. [3] Istimet in eodem comitatu & alii plures h’nt de socha e˛ ccl’iæ eli: (lviiito)c (xv)i Man’ de xxxi h’ & dm’, xiii car’ * * * & e’ appreciata xxx lib’. H˛ec t’ra suffic’ xxviii car’, & illi qui hanc t’ram de soca tenuer’t t R ead’ uendere potuer’t, sed sac’ & soc’ & commend’ & seruicium semper remanebat e˛ ccl’˛e Eli, & de hac iterum seruiunt abb’i: concessione & iussu regis Willelmi. [4] IN hertfortscira h’t abb’ eli in dominio; iii Man’ de xlix h’, viii car’ in d’nio, l uill’i lv b’ xx ser’ h’ntes xxxiiii car’, appretiatur l lib’. H˛ec t’ra suffic’ liii c’.

149

The survey of the whole of England

[5] Idem abb’ in Essesse h’t in d’nio v man’ de xlix h’ & dm’, xiiii c’ in d’nio, c & ii uill’ xlv b’ xliiii ser’: h’ntes xxxix c’, & hoc totum ualet lxiiii lib’ & x sol’. Milit’ eius in eod’ comitatu h’nt ii Manerium de v h’, iii c’ in d’nio, vi uillani vii b’ vii ser’ h’ntes vi c’, precium viii lib’. H˛ec t’ra sufficit lxi c’, emendata de ix lib’: in manu Symeonis abb’is. [6] Idem abb’ in Nortfolch h’t in d’nio xv Manerium de lxvii c’ t’r˛e 7 xxxiii ac’; xxxiiii c’ 7 dm’ in d”nio, cc & xv uill’i; ccc & xxx b’ liiii (s)i lvii sochem’ h’ntes lxviii c’, & hoc appreciatur c & v lib’ & vii sol’ & vi d’. Milit’ eius in eodem comitatu h’nt (ix)c (viii)i Man’ de xvi c’ terr˛e, vii c’ in d’nio, xxii uill’ lxi b’ liii soch’m’ h’ntes xiii c’, & appretiatur hoc xii lib’ & x sol’. H˛ec t’ra suffic’ c xli c’, emendata de xvii lib’: in manu Symeonis abb’is. [7] Idem abb’ h’t in Sudfolch in d’nio (xxx)c (xvi)i Man’ de lxix c’ terr˛e 7 xxxii ac’, xlvi car’ in d’nio, c lxxxviii liberos ho’es lxxx burgenses x sochem’ cc 7 xii (uill’i)i cc lx 7 xiii (b)i lv (s)i h’ntes c 7 xv car’, & appreciatur c lx xxix lib’ 7 xii sol’. Milites sui h’nt in eodem comitatu (xvii)c (xiii)i Man’ de xl c’ t’r˛e: & x ac’, xxxvi car’ in d’nio, c xiiii liberos homines xxvii sochem’ xxvi (uill’i)i xxxv b’ iiii ser’ h’ntes xxii c’ & dm’, appretiata e’ xx lib’ 7 viii sol’ & vi d’. H˛ec t’ra suffic’ cc xlviii c’, emendata de vii lib’ in manu Symeonis abb’is, 7 de quinque hund’ 7 dm’ de eod’ comitatu: x lib’ per annum. [8] Idem abb’ in huntendona syra h’t in d’nio ij(ii)c Man’ de xxxix h’ & dm’; x c’ in d’nio, c & xii uill’ xxvii b’ h’ntes xliiii c’, precium xl lib’, terra xlv car’.

H˛ec t’ra predicta de Thainl’ 7 soca scripta 7 appreciata in breue abb’is de eli, 7 illi qui hanc t’ram de soc’ tenuer’t t R ead’ uendere potuer’, sed sac’ 7 soc’ 7 comd’ seruitium semper ræmanebat e˛ ccl’˛e eli, 7 de hac seruit pi abb’i concessione regis. [12] Harduinus de escalariis h’t in Granteb’ syra in d’nio (lx)c (xxx)i man’ de xxix h’, xiiii c’ 7 dm’: in d’nio, xliiii uill’ lxiiii (b)i h’ntes xx c’ 7 dm’, precium xxxv lib’7 7 x sol’ 7 viii d’. Milit’ eius in eod’ comitatu h’nt (quat’ xx 7 i)c (xxx)i Man’ de xxxiii h’ 7 dm’ & dm’ v’, xix c’ 7 dm’: in d’nio, xviii (uill’i)i; quat’ xx 7 i b’, ix (s)i h’ntes xvii c’ 7 dm’, precium xxv lib’ 7 xv sol’ 7 vi dn’. * * * [13] IDEM harduinus h’t in hertfort syra in d’nio (xvii)c (x)i man’ de xxii h’ 7 iii v’ 7 dm’, viii c’ 7 dm’ in d’nio, xlii uill’ xxviii (b)i iiii (s)i h”ntes xvi c’ 7 dm’; precium xxiii lib’ 7 viii sol’. Milit’ eius in eodem comitatu h’nt (xxxvi)c (x)i Man’ de xviii h’ 7 dm’, vi c’ in d’nio, xvii uill’ lxxii (b)i x (s)i h’ntes xiii c’, precium xvi lib’ 7 v sol’, emendata de xv lib’ sub harduino, t’ra ad liii car’. [14] Harduinus predictus in granteb’ syra h’t de Thainland’ eli de quibus concordatus e’ cum abb’e concessione regis (xxviii)c (iiii)i man’ de viii h’ 7 iii v’, v c’ in d’nio, xvi uill’, viiito b’ iiii (s)i h’ntes vi c’ 7 dm’, pretium x lib’ 7 xv sol’ 7 viii d’, terra xi c’, emendata de iii lib’. H˛ec t’ra e’ scripta 7 appr˛etiata in breue abb’is eli. [15] Idem harduinus in eod’ comitatu h’t de soca abbati˛e eli (xviii)c (iiii)i man’ de iiii h’ 7 iii v’, iii c’ in d’nio, vii uill’ xiiii (b)i h’ntes ii c’, precium vii lib’ 7 iiii sol’ 7 ii dn’, t’ra viii c’, 7 h˛ec e’ appretiata in breue abb’is eli. & illi qui hanc terram ten’ de soca tpr’ R ead’ uendere pot’, sed saca 7 soc’ 7 commend’ 7 seruitium semper remanebat e˛ ccl’i˛e eli. (7)i De hac seruit hard’ abb’i: iussu regis.

[9] Picotus uicecomes h’t in granteb’ syra in d’nio (lxxviii)c (xxx)i man’ de lviii h’ & dm’ v’, xii car’ & dm’ in d’nio, lxiii (uill’i)i c 7 iiii (b)i xii (s)i h’ntes xxxi c’ 7 dm’, appr˛etiatur lvii lib’ 7 viii sol’, & iio molend’ IN Burgo de granteb’ge de viii lib’. Milit’ eius in eod’ comitatu h’nt (quat’ xx & iii)c (xxx)i man’ de liiii h’ 7 iii v’, xvii c’ 7 dm’ in d’nio, xxxiiii uill’i c xlii b’ vii (s)i h’ntes xxvii c’, precium l lib’ xiii sol’ 7 iiii d’, emendata de xxvi s; t’ra c 7 xii car’.

[16] Wido de rambutcurt in Granteb’ syre h’t de Thainland’ e˛ ccl’˛e de eli i man’ de i v’ 7 dm’, pr˛ecium x s, & de soca h’t ipse: in eod’ comitatu: xviii Man’ de iiiior h’ 7 i v’ 7 dm’, iii c’ in d’nio, ii uill’ xvi (b)i h’ntes ii c’ 7 dm’: pretium v lib’: t’ra vii c’. H˛ec t’ra e’ scripta 7 appr˛etiata in breue abbatis eli. & inde Wido seruit abb’i iussu regis.

[10] Idem pi in eod’ comitatu h’t de Thainland’ e˛ ccl’˛e eli de quibus concordatus e’ cum abb’e concessione regis x Man’ de xxi h’ 7 i v’ 7 dm’, viii c’ in d’nio, xx (uill’i)i xxviii (b)i v (s)i h’ntes v c’, precium xv lib’, t’ra xvi c’ 7 ii bobus. [11] Idem pi in eod’ comitatu h’t de soca e˛ ccl’˛e (eli)i (xxii)c (x)i man’ de xix h’ 7 dm’ v’, vi c’ 7 dm’ in dominio, xviii (uill’i)i xxvii (b)i ii (s)i h’ntes v car’, precium xiiii lib’7 vi sol’ 7 viiito d’, t’ra xii c’ 7 dm’.

150

Bibliography

Alecto Historical Editions 1986 Alecto Historical Editions (publ.), Great Domesday (London, 1986). Alecto Historical Editions 1987–92 Alecto Historical Editions (publ.), Great Domesday, 31 vols. (London, 1987–92). This is the ‘county edition’; a ‘library edition’ and a ‘collectors’ edition’ of the entire manuscript were issued in parallel with it. Alecto Historical Editions 2000 Alecto Historical Editions (publ.), Little Domesday, 3 vols. (London, 2000). Alecto Historical Editions 2003 Alecto Historical Editions (publ.), Digital Domesday (London, 2003). Ballard 1906 A. Ballard, The Domesday Inquest (London, 1906). Ballard 1920 A. Ballard (ed.), An eleventh-century inquisition of St Augustine’s, Canterbury (British Academy, Records of the social and economic history of England and Wales IV-2, 1920). Baring 1912 F. H. Baring, ‘The Exeter Domesday’, English Historical Review, 27 (1912), 309–18. Barlow 1963 F. Barlow, ‘Domesday Book: a letter of Lanfranc’, English Historical Review, 78 (1963), 284–9. Bates 1998 D. Bates (ed.), Regesta regum AngloNormannorum: the acta of William I (1066–1087) (Oxford, 1998). Bishop and Chaplais 1957 T. A. M. Bishop and P. Chaplais (eds.), Facsimiles of English royal writs to A.D. 1100 (Oxford, 1957). Birch 1887 W. de G. Birch, Domesday Book: a popular account of the Exchequer manuscript so called (London, 1887). Blake 1962 E. O. Blake (ed.), Liber Eliensis (Royal Historical Society, Camden Third Series 92, 1962). Bond, Thompson and Warner 1884–94 E. A. Bond, E. M. Thompson and G. F. Warner (eds.), Facsimiles of manuscripts and inscriptions, 2nd series, 2 vols. (Palaeographical Society, 1884–94). Botfield 1849 B. Botfield, Notes on the cathedral libraries of England (London, 1849). Brown 1957 R. A. Brown, ‘ “The treasury” of the later twelfth century’, in J. Conway Davies (ed.), Studies presented to Sir Hilary Jenkinson (London, 1957), 35–49. Burton 2004 J. E. Burton, ‘Gerard (d. 1108)’, Oxford Dictionary of National Biography (Oxford, 2004), 21:928–9. Chaplais 1987 P. Chaplais, ‘William of Saint-Calais and the Domesday Survey’, in J. C. Holt (ed.), Domesday studies (Woodbridge, 1987), 65–77. Cheney 1983 C. R. Cheney, ‘Service-books and records: the case of the “Domesday monachorum” ’, Bulletin of the Institute of Historical Research, 56 (1983), 7–15. Clark 1958 C. Clark (ed.), The Peterborough chronicle,

1070–1154 (London, 1958). Clark 1970 C. Clark (ed.), The Peterborough chronicle, 1070–1154, 2nd ed. (Oxford, 1970). Clover and Gibson 1979 H. Clover and M. Gibson (eds.), The letters of Lanfranc archbishop of Canterbury (Oxford, 1979). Condon and Hallam 1984 M. M. Condon and E. M. Hallam, ‘Government printing of the Public Records in the eighteenth century’, Journal of the Society of Archivists, 7 (1982–5), 348–88. Darby and Finn 1967 H. C. Darby and R. Welldon Finn (eds.), The Domesday geography of south-west England (Cambridge, 1967). Darlington 1955 R. R. Darlington (ed.), ‘Wiltshire Geld Rolls’, in R. B. Pugh and E. Crittall (eds.), The Victoria History of the Counties of England: a history of Wiltshire, vol. II (London, 1955), 169–217. Darlington 1955 R. R. Darlington (ed.), ‘Summaries of fiefs in the Exon Domesday’, in R. B. Pugh and E. Crittall (eds.), The Victoria History of the Counties of England: a history of Wiltshire, vol. II (London, 1955), 218–21. Davis 1913 H. W. C. Davis (ed.), Regesta Willelmi conquestoris et Willelmi rufi 1066–1100 (Regesta regum Anglo-Normannorum I, Oxford, 1913). Davis 1954 R. H. C. Davis (ed.), The kalendar of abbot Samson of Bury St Edmunds and related documents (Royal Historical Society, Camden Third Series 84, 1954). Dethlefsen and Deetz 1966 E. Dethlefsen and J. Deetz, ‘Death’s heads, cherubs and willow trees: experimental archaeology in colonial cemeteries’, American Antiquity, 31 (1966), 501–10. Douglas 1933 D. C. Douglas, ‘Odo, Lanfranc, and the Domesday Survey’, in J. G. Edwards, V. H. Galbraith and E. F. Jacob (eds.), Historical essays in honour of James Tait (Manchester, 1933), 47–57. Douglas 1936 D. C. Douglas, ‘The Domesday Survey’, History, 21 (1936), 249–57. Douglas 1944 D. C. Douglas (ed.), The Domesday Monachorum of Christ Church Canterbury (London, 1944). Du Boulay 1966 F. R. H. Du Boulay, The lordship of Canterbury: an essay on medieval society (London, 1966). Ellis 1816 H. Ellis (ed.), Libri censualis vocati Domesday-book, additamenta ex codic[ibus] antiquiss[imis] . . . (London, 1816). Ellis 1817 H. Ellis, Introduction to the supplementary records of Domesday (London, 1817). Ellis 1833 H. Ellis, A general introduction to Domesday Book (2 volumes, London, 1833).

151

The survey of the whole of England Erskine and Williams 2003 R. W. H. Erskine and A. Williams (eds.), The story of Domesday Book (Chichester, 2003). Eyton 1877 R. W. Eyton, ‘Notes on Domesday’, Transactions of the Shropshire Archaeological and Natural History Society, 1 (1877–8), 99–118. Farrer 1925 W. Farrer, Honors and knights’ fees, vol. III (Manchester, 1925). Finberg 1944 H. P. R. Finberg, ‘The early history of Werrington’, English Historical Review, 59 (1944), 237– 51. Finn 1951 R. Welldon Finn, ‘The evolution of successive versions of Domesday Book’, English Historical Review, 66 (1951), 561–4. Finn 1957 R. Welldon Finn, ‘The immediate sources of the Exchequer Domesday’, Bulletin of the John Rylands Library, 40 (1957–8), 47–78. Finn 1959 R. Welldon Finn, ‘The Exeter Domesday and its construction’, Bulletin of the John Rylands Library, 41 (1958-9), 360–87. Finn 1960 R. Welldon Finn, ‘The Inquisitio Eliensis reconsidered’, English Historical Review, 75 (1960), 386– 409. Finn 1961 R. Welldon Finn, The Domesday Inquest and the making of Domesday Book (London, 1961). Finn 1964 R. Welldon Finn, Domesday studies: the ‘Liber Exoniensis’ (London, 1964). Flight 1997 C. Flight, The bishops and monks of Rochester, 1076–1214 (Kent Archaeological Society Monograph Series VI, Maidstone, 1997). Flower and Smith 1941 R. Flower and H. Smith (eds.), The Parker chronicle and laws (Corpus Christi College, Cambridge, MS. 173): a facsimile (London, 1941). Galbraith 1920 V. H. Galbraith, ‘Royal charters to Winchester’, English Historical Review, 35 (1920), 382– 400. Galbraith 1931 V. H. Galbraith, ‘Girard the chancellor’, English Historical Review, 46 (1931), 77–9. Galbraith 1942 V. H. Galbraith, ‘The making of Domesday Book’, English Historical Review, 57 (1942), 161– 77. Galbraith 1948 V. H. Galbraith, Studies in the public records (London, 1948). Galbraith 1950 V. H. Galbraith, ‘The date of the geld rolls in Exon Domesday’, English Historical Review, 65 (1950), 1–17. Galbraith 1957 V. H. Galbraith, [Review of Pugh and Crittall 1955], English Historical Review, 72 (1957), 94–6. Galbraith 1961 V. H. Galbraith, The making of Domesday Book (Oxford, 1961). Galbraith 1967 V. H. Galbraith, ‘Notes on the career of Samson, bishop of Worcester (1096–1112)’, English Historical Review, 82 (1967), 86–101. Galbraith 1970 V. H. Galbraith, ‘Afterthoughts’, in L. P. Curtis (ed.), The historian’s workshop (New York, 1970), 5–21. Galbraith 1974 V. H. Galbraith, Domesday Book: its place in administrative history (Oxford, 1974).

Gale 1691 T. Gale (ed.), Historiae britannicae, saxonicae, anglo-danicae, scriptores XV (Oxford, 1691). Gullick 1987 M. Gullick, ‘The Great and Little Domesday manuscripts’, in A. Williams and R. W. H. Erskine (eds.), Domesday Book studies (London, 1987), 93–112. Hallam 1986 E. M. Hallam, Domesday Book through nine centuries (London, 1986). Hallam 1987 E. M. Hallam, ‘Annotations in Domesday Book since 1100’, in A. Williams and R. W. H. Erskine (eds.), Domesday Book studies (London, 1987), 136–50. Hallam and Bates 2001 E. Hallam and D. Bates (eds.), Domesday Book (Stroud, 2001). Hamilton 1876 N. E. S. A. Hamilton (ed.), Inquisitio comitatus Cantabrigiensis, subjicitur inquisitio Eliensis (London, 1876). Hardy 1865 T. Duffus Hardy, Descriptive catalogue of materials relating to the history of Great Britain and Ireland, vol. II (Rolls Series 26/2, 1865). Hart 1974 C. Hart, The hidation of Cambridgeshire (Leicester, 1974). Hart 1863 W. H. Hart (ed.), Historia et cartularium monasterii sancti Petri Gloucestriae, vol. I (Rolls Series 33/1, 1863). Harvey 1980 S. P. J. Harvey, ‘Recent Domesday studies’, English Historical Review, 95 (1980), 121–33. Holt 1987 J. C. Holt, ‘1086’, in J. C. Holt (ed.), Domesday studies (Woodbridge, 1987), 41–64. Holt 2001 J. C. Holt, ‘Domesday studies 2000’, in E. Hallam and D. Bates (eds.), Domesday Book (Stroud, 2001), 19–24, 199. Holtzmann W. Holtzmann, Papsturkunden in England, vol. II (Berlin, 1935–6). Irvine 2004 S. Irvine (ed.), The Anglo-Saxon Chronicle – a collaborative edition – volume 7 – MS E (Cambridge, 2004). Jaffé 1885–8 P. Jaffé (ed.), Regesta pontificum Romanorum, 2nd ed., 2 vols., rev. S. Loewenfeld, F. Kaltenbrunner and P. Ewald (Leipzig, 1885–8). James 1902 M. R. James, The Western manuscripts in the library of Trinity College, Cambridge: a descriptive catalogue, vol. III (Cambridge, 1902). Jenkinson 1954 [H. Jenkinson (comp.)], Domesday rebound (London, 1954). Johnson 1906 C. Johnson, ‘Introduction to the Norfolk Domesday’, in W. Page (ed.), The Victoria History of the Counties of England: a history of Norfolk, vol. II (London, 1906), 1–37. Johnson 1950 C. Johnson (ed.), Dialogus de Scaccario (London, 1950). Johnson and Cronne 1956 C. Johnson and H. A. Cronne, Regesta Henrici primi 1100–1135 (Regesta regum Anglo-Normannorum II, Oxford, 1956). Johnson and Jenkinson 1915 C. Johnson and H. Jenkinson, English court hand A.D. 1066 to 1500 (Oxford, 1915). Ker 1948 N. R. Ker, ‘Hemming’s cartulary: a description of the two Worcester cartularies in Cotton Tiberius A. xiii’, in R. W. Hunt, W. A. Pantin and R. W. Southern (eds.), Studies in medieval history presented to Frederick

152

Bibliography Maurice Powicke (Oxford, 1948), 49–75. Reprinted in Ker 1985, pp. 31–59. Ker 1957 N. R. Ker, Catalogue of manuscripts containing Anglo-Saxon (Oxford, 1957). Ker 1976 N. R. Ker, ‘The beginnings of Salisbury Cathedral library’, in J. J. G. Alexander and M. T. Gibson (eds.), Medieval learning and literature: essays presented to Richard William Hunt (Oxford, 1976), 23–49. Reprinted in Ker 1985, pp. 143–73. Ker 1977 N. R. Ker, Medieval manuscripts in British libraries, vol. II (Oxford, 1977). Ker 1985 N. R. Ker, Books, collectors and libraries: studies in the medieval heritage, ed. A. G. Watson (London and Ronceverte, 1985). Larking 1869 L. B. Larking (ed.), The Domesday Book of Kent (London, 1869). Le Patourel 1948 J. Le Patourel, ‘The reports of the trial on Penenden Heath’, in R. W. Hunt, W. A. Pantin and R. W. Southern (eds.), Studies in medieval history presented to Frederick Maurice Powicke (Oxford, 1948), 15–26. Lennard 1946 R. Lennard, [Review of Douglas 1944], English Historical Review, 61 (1946), 253–60. Lennard 1959 R. Lennard, Rural England 1086–1135: a study of social and agrarian conditions (Oxford, 1959). Lloyd 1956 L. J. Lloyd with A. M. Woodcock, The library of Exeter Cathedral (Exeter, 1956). Love 2004 R. C. Love (ed.), Goscelin of Saint-Bertin: the hagiography of the female saints of Ely (Oxford, 2004). McGurk 1998 P. McGurk (ed.), The chronicle of John of Worcester – volume III – the annals from 1067 to 1140 with the Gloucester interpolations and the continuation to 1141 (Oxford, 1998). McKisack 1971 M. McKisack, Medieval history in the Tudor age (Oxford, 1971). Maitland 1897 F. W. Maitland, Domesday Book and beyond (London, 1897). Mason 1954 J. F. A. Mason, ‘The date of the geld rolls’, English Historical Review, 69 (1954), 283–9. Miller 1947 E. Miller, ‘The Ely land pleas in the reign of William I’, English Historical Review, 62 (1947), 438– 56. Morgan 1858 J. F. Morgan, England under the Norman occupation (Edinburgh and London, 1858). Morgan 1983 P. Morgan (ed.), Domesday Book – 1 – Kent (Chichester, 1983). Osley 1965 A. S. Osley (ed.), Calligraphy and palaeography: essays presented to Alfred Fairbank on his 70th birthday (London, 1965). Palmer, Palmer and Slater 2000 J. Palmer, M. Palmer and G. Slater, Domesday Explorer: Great Domesday Book on CD-ROM (Chichester, 2000). Pearson 2001 H. Pearson, ‘The Alecto Domesday project’, in E. Hallam and D. Bates (eds.), Domesday Book (Stroud, 2001), 151–8. Phillimore & Co. 1975–86 Phillimore & Co. (publ.), History from the sources – Domesday Book – a survey of the counties of England, 34 vols. in 39 (Chich-

ester, 1975–86). One supplementary volume was issued in 1982; three volumes containing indexes appeared in 1992. Plummer 1892–9 C. Plummer (ed.), Two of the Saxon chronicles parallel, 2 vols. (Oxford 1892–9). Pollock 1895 F. Pollock, [Review of Round 1895], English Historical Review, 10 (1895), 783–4. Prescott 2001 A. Prescott, ‘Sir Henry Ellis and Domesday Book’, in E. Hallam and D. Bates (eds.), Domesday Book (Stroud, 2001), 159–90, 216. Pugh and Crittall 1955 R. B. Pugh and E. Crittall (eds.), The Victoria History of the Counties of England: a history of Wiltshire, vol. II (London, 1955). Richardson 1928 H. G. Richardson, ‘Richard fitz Neal and the Dialogus de Scaccario’, English Historical Review, 43 (1928), 161–71, 321–40. Robertson 1956 A. J. Robertson (ed.), Anglo-Saxon charters, 2nd ed. (Cambridge, 1956). Robinson 1988 P. R. Robinson, Catalogue of dated and datable manuscripts c. 737–1600 in Cambridge libraries (Cambridge, 1988). Roffe 1990 D. R. Roffe, ‘Domesday Book and Northern society: a reassessment’, English Historical Review, 105 (1990), 310–36. Roffe 2000 D. R. Roffe, Domesday: the inquest and the book (Oxford, 2000). Roffe 2001 D. Roffe, ‘Domesday: the inquest and the book’, in E. Hallam and D. Bates (eds.), Domesday Book (Stroud, 2001), 25–36, 199–200. Round 1888 J. H. Round, ‘Danegeld and the finance of Domesday’, in P. E. Dove (ed.), Domesday studies, being the papers read at the meetings of the Domesday commemoration 1886, vol. I (London, 1888), 77–142. Round 1895 J. H. Round, Feudal England (London, 1895). Round 1903 J. H. Round, ‘Introduction to the Essex Domesday’, in H. A. Doubleday and W. Page (eds.), The Victoria History of the Counties of England: a history of Essex, vol. I (Westminster, 1903), 333–425. Rumble 1981 A. Rumble (ed.), Domesday Book – 18 – Cambridgeshire (Chichester, 1981). Rumble 1985 A. R. Rumble, ‘The palaeography of the Domesday manuscripts’, in P. H. Sawyer (ed.), Domesday Book: a reassessment (London, 1985), 28–49. Rumble 1986 A. Rumble (ed.), Domesday Book – 34 – Suffolk (Chichester, 1986). Rumble 1987 A. R. Rumble, ‘The Domesday manuscripts: scribes and scriptoria’, in J. C. Holt (ed.), Domesday studies (Woodbridge, 1987), 79–99. Sainty 1983 J. C. Sainty (comp.), Officers of the Exchequer (List & Index Society, Special Series 18, 1983). Saltman 1956 A. Saltman, Theobald archbishop of Canterbury (London, 1956). Sawyer 1954 P. H. Sawyer, [Review of Jenkinson 1954], English Historical Review, 69 (1954), 650–1. Sawyer 1955 P. H. Sawyer, ‘The “original returns” and Domesday Book’, English Historical Review, 70 (1955), 177–97. Sawyer 1956 P. H. Sawyer, ‘The place-names of the

153

The survey of the whole of England Domesday manuscripts’, Bulletin of the John Rylands Library, 38 (1955–6), 483–506. Sawyer 1968 P. H. Sawyer, Anglo-Saxon charters: an annotated list and bibliography (London, 1968). Southern 1978 R. W. Southern, ‘Vivian Hunter Galbraith 1889–1976’, Proceedings of the British Academy, 64 (1978), 397–425. Stenton 1943 F. M. Stenton, Anglo-Saxon England (Oxford, 1943). Stenton 1971 F. M. Stenton, Anglo-Saxon England, 3rd ed. (Oxford, 1971). Stephenson 1947 C. Stephenson, ‘Notes on the composition and interpretation of Domesday Book’, Speculum, 22 (1947), 1–15. Reprinted in C. Stephenson, Mediaeval institutions: selected essays, ed. B. D. Lyon (Ithaca, 1954), 184–204. Stevenson 1858 J. Stevenson (ed.), Chronicon monasterii de Abingdon, 2 vols. (Rolls Series 2, 1858). Stevenson 1907 W. H. Stevenson, ‘A contemporary description of the Domesday Survey’, English Historical Review, 22 (1907), 72–84. Stubbs 1913 W. Stubbs (ed.), Select charters and other illustrations of English constitutional history from the earliest times to the reign of Edward the First, 9th ed., rev. H. W. C. Davis (Oxford, 1913). Swanton 1996 M. J. Swanton (trans.), The Anglo-Saxon Chronicle (London, 1996). Tait 1908 J. Tait, [Review of Ballard 1906], English Historical Review, 23 (1908), 122–6. Taylor 1957 J. Taylor, ‘The French “Brut” and the reign of Edward II’, English Historical Review, 72 (1957), 423–37. Taylor 1983 S. Taylor (ed.), The Anglo-Saxon Chronicle – a collaborative edition – volume 4 – MS B (Cambridge, 1983). Thorn and Thorn 1979 C. Thorn and F. Thorn (eds.), Domesday Book – 10 – Cornwall (Chichester, 1979). Thorn and Thorn 1980 C. Thorn and F. Thorn (eds.), Domesday Book – 8 – Somerset (Chichester, 1980). Thorn and Thorn 1983 C. Thorn and F. Thorn (eds.), Domesday Book – 7 – Dorset (Chichester, 1983). Thorn and Thorn 1985 C. Thorn and F. Thorn (eds.), Domesday Book – 9 – Devon (Chichester, 1985). Thorn and Thorn 1982 F. Thorn and C. Thorn (eds.), Domesday Book – 16 – Worcestershire (Chichester, 1982). Thorn and Thorn 2001 F. Thorn and C. Thorn, ‘The writing of Great Domesday Book’, in E. Hallam and D. Bates (eds.), Domesday Book (Stroud, 2001), 37–72, 200–3. Urry 1967 W. Urry, Canterbury under the Angevin kings (London, 1967). Ward 1933 G. Ward, ‘The lists of Saxon churches in the Domesday Monachorum, and White Book of St Augustine’, Archaeologia Cantiana, 45 (1933), 60–89. Webber 1989 T. Webber, ‘Salisbury and the Exon Domesday: some observations concerning the origin of Exeter Cathedral MS 3500’, English manuscript studies 1100–1700, 1 (1989), 1–18. Webber 1992 T. Webber, Scribes and scholars at Salis-

bury Cathedral c. 1075–c. 1125 (Oxford, 1992). Whale 1905 T. W. Whale, ‘History of the Exon “Domesday” ’, Transactions of the Devonshire Association, 37 (1905), 246–83. Whitelock 1954 D. Whitelock (ed.), The Peterborough chronicle (Copenhagen, 1954). Williams 1968 A. Williams (ed.), ‘Dorset Geld Rolls’, in R. B. Pugh (ed.), The Victoria History of the Counties of England: a history of the county of Dorset, vol. III (London, 1968), 115–49. Williams and Erskine 1987 A. Williams and R. W. H. Erskine (eds.), Domesday Book studies (London, 1987). Reissued in a smaller format as Erskine and Williams 2003. Williams and Martin 1992 A. Williams and G. H. Martin (eds.), The Kent Domesday (London, 1992). Williams and Martin 2002 A. Williams and G. H. Martin (eds.), Domesday Book: a complete translation (London, 2002). Wormald 1946 F. Wormald (ed.), English Benedictine kalendars after A.D. 1100, vol. II (Henry Bradshaw Society 81, 1946).

154

Index

Abingdon, abbey of 34 Acta Lanfranci 111 Adam son of Hubert 111, 116–17 Adam son of Willelm 116 Agarde, A. 37, 81, 88, 123 Alecto Historical Editions 11, 15 Alfwin, monk of Saint-Rémi 116 Anglo-Saxon Chronicle, see English chronicle Anskitil de Ros 114, 144 Apton, see Bergh Apton

David, king of Israel and Judah 110 Davis, H. W. C. 121 Dialogus de scaccario 34–6, 123, 127 Domesday 1, 13, 25, 36–7 Domesday Book, see manuscripts, Public Record Office, E 31/1–2 Domesday Monachorum 111 Douglas, D. C. 3, 5, 7–9, 111–12, 117 Dover (Kent), church of Saint Martin 34 Durham, bishop of, see Rannulf Flambard, Willelm

Ballard, A. 5, 92, 112, 117, 123–4, 132 Baring, F. H. 4, 6, 10 Barlow, F. 109, 114 Barnes, R. 39–40, 52–3 Basire, J. 39 Bayeux, bishop of, see Odo Beda, monk of Jarrow 87, 108 Bergh Apton (Norfolk) 98–100 Bernard, monk of Cluny 7 Birch, W. de G. 69 Blake, E. O. 82–3, 85–8, 122 Bochena hundred (Dorset) 63 book of the treasury 34 book of Winchester 33–4 Boughton Monchelsea (Kent) 11 breves, brevia 72, 104 Burwell (Cambridgeshire) 90–1

Edwards, J. G. 2 Ellis, H. 4, 39–40, 44, 52–3, 55, 60–1, 66–7, 74, 76, 81–2, 95, 106, 134, 137, 148 Elmham, bishop of, see Willelm Ely, abbot of, see Simeon , bishops of, see Herveus, Nigel , monks of 2, 34, 81–9, 94–107, 116–17, 122–3, 149– 50 , , their cartulary 82–3, 85, 87–8, 96 , , their saints’ lives 87 , see also Historia Elyensis ecclesie, Libellus operum Aedeluuoldi episcopi English chronicle 7, 79–80, 109–10, 120, 144 Erchenger, baker 104 Eudo, king’s steward 95, 116 Evesham, abbot of, see Walter exchequer 34–7 Exeter, bishop of, see Willelm de Warelwast , deans of, see Lyttelton, Willoughby Exon Domesday, see manuscripts, Exeter Cathedral Library 3500 Eyre and Strahan 39 Eyton, R. W. 2, 4–5, 7, 117 Eyton’s conjecture 1, 5–6, 8, 10, 92, 117

Caen, abbey of La Trinité 137 Caley, J. 39 Canterbury, archbishop of, see Lanfranc , cathedral church called Christ Church or Holy Trinity, monks of 34, 111, 114 , see also Saint Augustine’s chancellor 36 , see also Girard Chaplais, P. 8–9, 31, 43, 71–4, 77 Cheney, C. R. 111 Chesterton hundred (Cambridgeshire) 146 circuits, see Eyton’s conjecture Colchester (Essex) 25, 120 colophon 29–30, 133, 135–6, 143–4 compotus hidarum 123 compotus solingorum 123 Cotton, R. 81, 88, 123 Coutances, bishop of, see Goisfrid Crowland Domesday 10 Cuttlestone hundred (Staffordshire) 129

Fairbank, A. J. 8–9, 14–15, 126 Fauconberge, E 13, 24 Finberg, H. P. R. 51 Finn, R. Welldon 7, 43, 49–52, 61, 76–8, 95, 101, 129 Firth-Clarke, K. 11 Galbraith, V. H. 1–10, 13, 25–6, 33, 38, 40, 43, 60–2, 65, 74, 76–7, 82, 91–2, 104, 117, 121, 134–5, 139–40, 143 Gale, R. 81 Gale, T. 81–2, 85 geld, dates for payment of 115 geld account for Northamptonshire 62, 64–5 geld accounts 7, 40, 50–2, 55, 60–70, 118, 120 geld fallacy 92 Gifford, D. H. 15

Darlington, R. R. 7, 52, 61, 69–70

155

The survey of the whole of England Girard, chancellor, bishop of Hereford, archbishop of York 135–6 Gislebert, abbot of Westminster 122 Glastonbury, abbey of 42, 50, 72, 104–6, 149 , abbot of, see Turstin Gloucester, meeting at 108–10, 115 , abbot of, see Serlo Goisfrid, bishop of Coutances 5, 38, 53, 77, 95, 116, 121, 131 Goisfrid Talebot 144 Goscelm 43–4, 139 Goscelm of Exeter 44 Great Domesday, see manuscripts, Public Record Office, E 31/2 Gullick, M. 10, 15–18, 24, 30–2

Libellus operum Aedeluuoldi episcopi 82, 85, 88, 96 Liber Eliensis, see manuscripts, Ely Liber Exoniensis, see manuscripts, Exeter Lincoln, bishop of of, see Remigius Little Domesday, see manuscripts, Public Record Office, E 31/1 London, bishop of of, see Ricard son of Nigel Longstow hundred (Cambridgeshire) 146 Love, R. C. 85, 87 Lyttelton, C., dean of Exeter 39–40, 53 Mabbs, A. W. 13 Mabbs’s diagram 13–15 Maitland, F. W. 2, 4–5, 8, 31, 82, 117, 141, 145 Malmaynes in Stoke (Kent) 11 manuscripts: British Library, Cotton Tib. A. vi 1, 87–91, 107 British Library, Cotton Tib. A. xiii 116 Cambridge, Corpus Christi College 173 111 Cambridge, Trinity College O. 2. 1 85–7 Cambridge, Trinity College O. 2. 41 2, 82–5, 94–107, 149–50 Canterbury Cathedral Library Lit. E 28 34, 111, 144 Canterbury, Dean and Chapter, Reg. K 111 Ely, Dean and Chapter, ‘Liber Eliensis’ 85 Exeter Cathedral Library 3500 1, 4, 10, 38–45, 49–55, 60–9, 71–80, 104–6, 126–32, 149 Oxford, Bodleian Library, Laud 636 109 Public Record Office, E 31/1 1, 8, 11, 25–30, 132–6 Public Record Office, E 31/2, fos. 0–372 1, 8–9, 11, 13–24, 137–45 Public Record Office, E 31/2, fos. 373–82 13–14, 25, 123, 140 Public Record Office, E 36/284 136 Public Record Office, E 164/27 2, 34, 123–4 Strood, Medway Archives and Local Studies Centre, DRc/R1 111 Marianus, monk of Mainz 108 Mason, J. F. A. 7 Mathildis, wife of Henric I 34 meetings of the king’s court, see Gloucester, Westminster, Winchester, see also Salisbury Miller, E. 121, 123 Milton Regis (Kent) 11 Mollington (Oxfordshire) 130 Morris, J. 10 Mortain, count of, see Rotbert

Haddenham (Buckinghamshire) 114 Hamilton, N. E. S. A. 2, 81–2, 85–6, 88, 95–6, 148 Hamo de Sancto Claro 95 Hardy, T. D. 82, 88 Hardwin de Escalers 104, 149–50 Hart, C. 99, 146 Harvey, S. P. J. 9 Haselora hundred (Dorset) 63 Hemming, monk of Worcester 109, 116 Henric, bishop of Winchester 36 Henric, treasurer 50, 127 Henric de Ferieres 116–17 Hereford, bishops of, see Girard, Rotbert Herstanahaia (Dorset?) 44 Herveus, bishop of Ely 82 Historia elyensis insule 82, 85–7, 96–7 Holt, J. C. 10, 80 Hugo, abbot of Saint Augustine’s 124 Hugo de Grentemaisnil 130 Hugo de Port 65 indexes 128, 133–4, 137–40 Inquisitio Comitatus Cantabrigiensis, see Hamilton Inquisitio Eliensis, see Ellis, Hamilton Ivo Tailgebosc 122 James, M. R. 82–3, 85 Jenkinson, H. 3, 8, 13–18, 24, 126, 133 Joab son of Zeruiah 110 Johnson, C. 3, 98, 100 Jones, J. P. 39 Judgment Day, see Domesday justiciar 35–7

Netherclift, F. 81–2, 85–6, 88 Nigel, treasurer, bishop of Ely 34, 87, 107 Nympsfield (Gloucestershire) 122

Kennett (Cambridgeshire) 90 Ker, N. R. 7, 38–40, 42–3, 49, 55, 61, 66–7, 71–2, 75, 116 Kingston (Kent) 11

Odo, bishop of Bayeux 64, 66, 77, 95, 116, 120, 144 Ordnance Survey 10, 81 Osmund, bishop of Salisbury 49, 127 Oswaldslow triple hundred (Worcestershire) 116 Over (Cambridgeshire) 98, 104

Lanfranc, archbishop of Canterbury 34, 110–14, 121–3, 144 , see also Acta Lanfranci, pseudo-Lanfranc Lennard, R. 7, 111 Lewknor (Oxfordshire) 34

Palaeographical Society 4, 74

156

Index Papworth hundred (Cambridgeshire) 146 Penenden Heath (Kent) 117 Phillimore & Co. 10–11, 15, 24 Picot, sheriff of Cambridgeshire 104, 149–50 pseudo-Lanfranc 7 Pyrford (Surrey) 122

Stevenson, W. H. 4, 108–9 Tait, J. 5 tallies 35, 64–5, 127 Taunton (Somerset) 72–3 Textus Roffensis, see manuscripts, Strood Teynham (Kent) 32, 112 Thorn, C., and Thorn, F. 10, 15, 23, 31–2, 38–9, 72, 74–5, 126, 144 Thornegrava hundred (Wiltshire) 67 treasurer 36, 62 , see also Henric, Nigel, Ricard son of Nigel treasurer’s clerk 35–6, 62, 127 treasurer’s scribe 36, 127 treasury 35–7, 38, 60–9, 106–7, 118, 120, 126–7, 132 Turstin, abbot of Glastonbury 106

Ramsey, abbot of 90–1 Rannulf Flambard, chaplain, bishop of Durham 73 receipt, see treasury Record Commission 10, 39, 81 Remigius, bishop of Lincoln 5, 116–17, 121 Ricard son of Nigel, treasurer, bishop of London 34–5 , see also Dialogus de scaccario Robert Malet’s mother 120 Robert son of Gerold 77 Robinson, P. R. 82–3, 85 Roffe, D. R. 9–10, 16–18, 104 Roger de Belmont’s men 115 Roger de Corcelles 77 roll of Winchester 37 Rotbert, bishop of Hereford 4–5, 66–7, 108–10, 115, 117, 120 Rotbert, count of Mortain 40, 43 Rotbert de Bruis 33 Rotbert de Oilgi 65, 115 Rouen, cathedral church of Notre-Dame 137 , church of Notre-Dame-du-Pré 137 Round, J. H. 2–5, 7–8, 60, 82, 90–1, 96–8, 100, 107, 121, 123, 144, 146 Rumble, A. R. 88, 102, 133–4, 137, 146

Urry, W. 111–12 Walkelin, bishop of Winchester 5, 71–3, 79–80, 116, 121, 125, 135 Walter, abbot of Evesham 115–16 Walter de Clavile 43–4, 139 Walter (called Walscin) de Dowai 110 Walter Gifard 110, 114, 116–17 Walter Tirel 144 Webber, T. 38, 49, 51–2, 66, 77, 127 Werrington (Devon) 51, 129 Westminster, meeting at 79–80, 120 , abbey of, 114 , abbot of, see Gislebert Whale, T. W. 4, 43–4, 52–3 Whittlesford hundred (Cambridgeshire) 94, 99, 146 Wido de Rainbuedcurt 104, 149–50 Willelm I, king of the English, duke of the Normans 30–1, 72–3, 79–80, 91, 108–11, 115, 120–2 Willelm II, king of the English 30–1, 73, 111, 144 Willelm, bishop of Durham 43, 67–8, 71–3, 79–80, 116, 122 Willelm, bishop of Elmham 67 Willelm de Albigni 144 Willelm de Moion 41 Willelm de Warelwast, bishop of Exeter 55 Willelm Pevrel of Dover 144 Williams, A. 62–4 Willingham (Cambridgeshire) 97 Willoughby, E., dean of Exeter 39 Winchester, meeting at 79, 115 , bishops of, see Henric, Walkelin Wisbech (Cambridgeshire) 96 women 119, 137 Worcester, bishops of, see Sanson, Wulstan , monks of, their chronicle 108, 122 Worda hundred (Wiltshire) 69–70 Wulstan, bishop of Worcester 115–16

S (recipient of letter from Lanfranc) 110, 114 Saint Alban’s, abbey of 114 Saint Augustine’s, abbey of 2, 34, 111–12, 123–4 , abbot of, see Hugo , White Book of, see manuscripts, Public Record Office, E 164/27 Saint Petroc’s, abbey of 104, 106 Salisbury, meeting at 72, 79–80, 110, 120–1 , bishop of, see Osmund Sandwich (Kent) 112 Sanson, chaplain, bishop of Worcester 64, 66, 77–8 satellites 2 Sawyer, P. H. 5, 8, 99-101, 103, 137 Selchelai hundred (Wiltshire) 67–8 seriation of D booklets 135–6 seriation of DB booklets 10, 15–24, 26, 118, 141–2 Serlo, abbot of Gloucester 116, 122 Simeon, abbot of Ely 6, 81, 98, 107, 121–3, 135 slaves in Huntingdonshire 95 Slingsby, M. 11 Soham (Cambridgeshire) 91 Southern, R. W. 2 spelling of English words by DB scribe 137 spelling of French words by DB scribe 14, 141 Staploe hundred (Cambridgeshire) 90–1 Stenton, F. M. 5, 109, 145 Stephenson, C. 117

York, abbot of 140 , archbishop of, see Girard

157

The survey of the whole of England

158