The Sword and the Crucible: A History of the Metallurgy of European Swords up to the 16th Century 9004227830, 9789004227835, 9789004229334

Table of contents :
Preface vii
PART ONE. THE FIRST METALS
1. The Extraction of the First Metals 3
2. The Smelting of Iron and the Production of Steel 12
3. Different Ways of Making Steel—Eastern and Western Steelmaking 24
PART TWO. THE FIRST EUROPEAN SWORDS
4. Celtic and Roman Swords 49
5. Pattern-Welding 62
PART THREE. THE "DARK" AGES IN EUROPE
6. The Revival of Science in Europe 85
7. The Survival of Technology From the Ancient World 96
8. Viking-Age Swords and Their Inscriptions 116
PART FOUR. STEEL ARMOUR AND SWORDS
9. The Invention of the Blast Furnace and Finery 187
10. Bloomery Steel and the Development of All-Steel Swords after 1400 202
11. The Mass-Production of Steel for Swords and Armour 210
12. The Decoration of Swords by Etching and Gilding 223
13. Medieval European swords after 1000 230
Further reading  287
Index  291

Citation preview

The Sword and the Crucible

History of Warfare Editors

Kelly DeVries Loyola University Maryland

John France University of Wales, Swansea

Michael S. Neiberg United States Army War College, Pennsylvania

Frederick Schneid High Point University, North Carolina

VOLUME 77

The titles published in this series are listed at www.brill.nl/hw

The Sword and the Crucible A History of the Metallurgy of European Swords up to the 16th Century By

Alan Williams

LEIDEN • BOSTON 2012

Cover illustration: “La bottega della spadaio” (The swordsmith’s shop), Matricola dei Fabbri, Bologna, 1366. With kind permission of Biblioteca del Senato della Repubblica "G. Spadolini" (Rome). Library of Congress Cataloging-in-Publication Data Williams, Alan (Alan R.)  The sword and the crucible : a history of the metallurgy of European swords up to the 16th century / by Alan Williams.   p. cm. — (History of warfare ; v. 77)  Includes bibliographical references and index.  ISBN 978-90-04-22783-5 (hardback : alk. paper)  1. Swords—Europe—History—To 1500. 2. Metallurgy—Europe—History—To 1500. I. Title.  U854.W55 2012  623.4’41—dc23

2012007522

This publication has been typeset in the multilingual “Brill” typeface. With over 5,100 characters covering Latin, IPA, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities. For more information, please see www.brill.nl/brill-typeface. ISSN 1385-7827 ISBN 978 90 04 22783 5 (hardback) ISBN 978 90 04 22933 4 (e-book) Brill has made all reasonable efforts to trace all rights holders to any copyrighted material used in this work. In cases where  these efforts have not been successful the publisher welcomes communications from copyrights holders, so that the appropriate acknowledgements can be made in future editions, and to settle other permission matters. Copyright 2012 by Koninklijke Brill NV, Leiden, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Global Oriental, Hotei Publishing, IDC Publishers and Martinus Nijhoff Publishers. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA. Fees are subject to change. This book is printed on acid-free paper.

contents

v

Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii PART ONE

THE FIRST METALS 1. 2. 3.

The Extraction of the First Metals. . . . . . . . . . . . . . . . . . . . . . . . . . . . .   3 The Smelting of Iron and the Production of Steel. . . . . . . . . . . . . .  12 Different Ways of Making Steel—Eastern and Western Steelmaking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  24 PART two

THE FIRST EUROPEAN SWORDS 4. Celtic and Roman Swords. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  49 5. Pattern-Welding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  62 PART three

THE “DARK” AGES IN EUROPE 6. The Revival of Science in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  85 7. The Survival of Technology From the Ancient World. . . . . . . . . . .  96 8. Viking-Age Swords and Their Inscriptions. . . . . . . . . . . . . . . . . . . . .  116 PART four

STEEL ARMOUR AND SWORDS 9. 10. 11. 12. 13.

The Invention of the Blast Furnace and Finery. . . . . . . . . . . . . . . . .  187 Bloomery Steel and the Development of All-Steel Swords after 1400. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  202 The Mass-Production of Steel for Swords and Armour . . . . . . . . .  210 The Decoration of Swords by Etching and Gilding. . . . . . . . . . . . .  223 Medieval European swords after 1000. . . . . . . . . . . . . . . . . . . . . . . .  230

Further reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  287 Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  291

vi

contents

Contents Contents v Preface vii PART ONE 1 THE FIRST METALS 1 Chapter one 3 The Extraction of the First Metals 3 Chapter two 12 The Smelting of Iron and the Production of Steel Chapter three 24 Different Ways of Making Steel— Eastern and Western Steelmaking PART two 47 THE FIRST EUROPEAN SWORDS

24

Chapter four 49 Celtic and Roman Swords

49

Chapter FIVE 62 Pattern-Welding 62 PART three 83 THE “DARK” AGES IN EUROPE

83

Chapter six 85 The Revival of Science in Europe

85

12

47

Chapter seven 96 The Survival of Technology From the Ancient World 96 Chapter eight 116 Viking-Age Swords and Their Inscriptions PART FOUR 185 STEEL ARMOUR AND SWORDS 185

116

Chapter nine 187 The Invention of the Blast Furnace and Finery

187

Chapter ten 202 Bloomery Steel and the Development of All-Steel Swords after 1400 202 Chapter eleven 210 The Mass-Production of Steel for Swords and Armour 210 Chapter twelve 223 The Decoration of Swords by Etching and Gilding

223

Chapter thirteen 230 Medieval European swords after 1000

230

Further reading 287 Index 291

preface

vii

Preface This is not a history of swords, but a history of the metallurgy of swords made in Europe. Only swords made of iron (and steel) are discussed here, over the time range between their first use in Celtic times and the Early Modern Period, by which time the methods of their manufacture were so well established that little further change would have been possible with the material available. In Europe this would have been a medium-carbon steel derived from a charcoal-smelted furnace, and therefore very low in sulphur, probably low in phosphorus, but never entirely free from slag. There would have been no accurate method for measuring time, and only empirical methods for measuring temperature, so that the heat-treatment of blades would have been difficult. Modern steels are less pure, insofar as they contain alloying elements to make heat-treatment easier, but they are almost slag-free. I have also endeavoured to give some account of the chemical knowledge available at the time because this would have influenced the metallurgy practised, albeit indirectly. It is sometimes held that all science in Medieval Europe was totally divided between scholastic theorists on the one hand and empirical practitioners on the other. But the writings of the academic philosophers often show a surprising knowledge of chemical and metallurgical practice, and while the craftsmen may have been illiterate they were by no means ignorant. The extraction of metals, and the control of their properties were steadily improved, so the links between the theoretical and the practical spheres of activity may have been rather indirect, but they were certainly present. I have confined myself to Western Europe, so Russia is not included, despite the fact that much archaeometallurgical activity is now happening there. Similarly, Muslim Spain is a terra incognita waiting to be investigated. I would like to thank those curators and archaeologists who have helped me over the years. They include (in alphabetical order): Dr. Silke Ackermann (British Museum), Dr. Christian Beaufort-Spontin and Dr. Matthias Pfaffenbichler (Hofjagd- und Rüstkammer, Vienna), Dr. Paul Craddock (British Museum), Dr. Barbara Grotkamp (Klingenmuseum,

viii

preface

Solingen), Prof. Volker Himmelein (Stuttgart), Dr. Robin Hanley and Dr. Robert Bell (Fenland Museum, Wisbech), Dr. Alex Hildred (Portsmouth), Dr. Jiři Hošek (Prague, Archaeological Institute) who has generously shared the results of his research with me, Dr. Birgid Kaland (Bergen Historical Museum), Dr. Janet Lang (British Museum), Jaak Mäll (Estonian History Museum, Tallinn), Dr Arthur MacGregor (Ashmolean Museum, Oxford) and Dr. Michael O’Hanlon (Pitt-Rivers Museum, Oxford), Dr. Rutt Ojasalu (Rakvere Museum), Professor Radomir Pleiner (Prague), Dr. Gerhard Quaas (Berlin) Prof. Heid Gjøstein Resi (Museum of Cultural Heritage, Oslo), Dr. Külli Rikas (Kuressaare Museum), the late Russell Robinson, Dr. Tobias Springer (Germanic National Museum, Nürnberg), Professor Erik Szameit (University of Vienna), Dr. Ulle Tamla (Institute of History, Tallinn University), Dr. Leena Tomantera (Kansali Museo, Helsinki), and Dr. Ralf Wiechmann (Museum for Hamburg History). The owners of those swords in private collections examined here cannot be publicly named, although they can be thanked, as they have also played an important part in this study. Certain museums have not been able, for one reason or another, to supply photographs, and so my own snapshots and drawings have had to be used instead. Unless stated otherwise, all the photomicrographs are the work of the author. A number of these photomicrographs were published earlier in the journal Gladius and are reproduced here, with my thanks. I would also like to thank Dr. Dipak Gohil of the National Physical Laboratory, Teddington, and Dr. David Nicol of Cambridge University, for their help in performing electron microanalysis; also those conservators Lasse Mattila, Simon Metcalf, and David Edge whose advice has been essential. Over recent years, the British Academy, the British Council and the Society of Antiquaries of London have all generously helped me with travel grants, without which the data could not have been collected. There is no cumulative bibliography because all my written sources are named in detailed references and further reading is suggested for those readers who wish to pursue individual topics further. All publications are in London unless stated otherwise.

The Extraction of the First Metals

PART ONE THE FIRST METALS

1

2

chapter one

The Extraction of the First Metals

3

Chapter one

The Extraction of the First Metals Before recipes were ever written down, chemistry was practised. But the substances used and their manipulations can only be deduced from the archaeological evidence. At some time around the tenth millenium bce hunter-gatherers in the alluvial valleys of the Nile and the Tigris-Euphrates discovered that edible plants could be harvested (and at a later date, deliberately sown) to provide a regular food supply. With the cultivation of crops, and the herding, rather than the hunting, of animals, settlement in a single place became possible and the first villages developed. The possession of a food surplus made specialisation possible, and expertise developed in many crafts. One such craft was pottery, from which developed metallurgy. Much of the earliest evidence comes from Egypt, although the uniquely favourable climatic conditions have preserved relatively more artefacts than other comparable Neolithic civilisations, so that Egyptian priority has sometimes been assumed in all technologies. The achievements of India and China, which developed in parallel with the West, are not going to be dealt with in this book, except insofar as their metallurgy sometimes illustrates a very different approach to making swords Silica (Silicon Dioxide) SiO2 The majority of minerals in the earth’s crust consist of silica and the silicates. The entire modern glass, ceramic and cement industries are based upon silicate chemistry. The chemistry of silcate minerals is extremely complex, but may be regarded in simple terms as follows. Silicates may consist of sheets or networks of SiO4 tetrahedra in various arrangements, which have an overall negative charge, such as (Si2O5)n–2n or (Si4O11)n–6n and which are held together by layers of positively charged metal ions such as Mg2+ Ca2+ or Fe2+. The replacement of silicon by aluminium (Al) is extremely com-

4

chapter one

mon, and this forms the aluminosilicates, which also contain positively charged metal ions. Asbestos, talc, mica and the clay minerals have layered structures of this sort, while feldspars have three-dimensional structures. Clays are significant because layers of water molecules are trapped between the sheets of aluminosilicates. Firing clay causes this water to be lost, and a rigid solid to be formed; other reactions also take place. In particular, “vitrification”, or the formation of a liquid silicate which does not recrystallise on cooling, but remains an amorphous solid or “glass”. Pottery that is fired at temperatures around 800oC does not vitrify in the kiln but remains slightly porous; this is known as “earthenware” or “terra cotta”. The earliest fired pottery found (from Çatal Hüyük in Anatolia) is thought to date from around the 7th–6th millennia bce. Pottery is useful because as well as holding liquids to drink or to cook food with, it also makes it easier to store food away from competitors such as rats. It is of great interest to chemists because it marks the start of pyrochemistry—the use of high temperatures to transform natural materials. Moreover, further improvements can be made. The sprinkling of salt, natron or woodash, upon the clay before or during firing leads to the formation of a sodium or potassium silicate, which is of much lower meltingpoint, and so forms a liquid coating on the surface of the pot, in other words a “glaze”. Natron is sodium sesquicarbonate (Na2CO3 .NaHCO3 which is found in Egypt occurring naturally in the Wadi al-Natrun) which will decompose on heating to sodium oxide. The main constituent of woodash (apart from silica) is potassium carbonate (K2CO3), which will decompose on heating to potassium oxide, which reacts in the same way as sodium oxide. Salt (sodium chloride, NaCl) dissociates into sodium oxide and hydrochloric acid. The sodium oxide then reacts with the silica to form sodium silicate. This may be simplified as Na2O + SiO2 = Na2SiO3 What might have begun as decoration, ends in making the pot waterproof and hence more versatile. Experiments have shown that firing between 870oC and 920oC will form a vitreous flux, which flows over the surface of the pot. The glaze does not crystallise on cooling, but remains amorphous, that is to say, it is a glass. By contrast, the much later ceramic, porcelain (which needs no external glaze) needs to be fired in the region of at least 1300o–1350oC.

The Extraction of the First Metals

5

The next stage in decoration was perhaps to try and colour the pots with brightly coloured minerals. Minerals which catch the eye by their conspicuous colours include some copper compounds (malachite, Cu(OH)2.CuCO3 is bright green, while azurite, another form of basic ­copper carbonate, 2CuCO3.Cu(OH)2 is bright blue) which opportunely also colour pottery blue or blue-green—indeed because the same metal ions Cu2+ are present in both the mineral and the glaze. Potters experimenting with coloured earths and glazes to decorate pots during firing may well have been the first experimental chemists, because such experiments could lead to the reduction of copper ores. It then remained for some proto-chemist to identify the tiny globules of metal thus produced with the native copper known independently. Once this identity had been realised, then the distinctively coloured ores of copper could be collected and smelted in a new and separate activity. Heating these ores with charcoal in a hearth—or better still, inside a clay pot (in other words, a crucible) would produce metallic copper. Eventually, glazes themselves would be further developed into another new material, glass. Copper A few metals occur “native” as the pure element. The commonest, by far, is copper; silver, gold, the platinum group metals, and iron are also found. There are also meteorites, of which some 5% are metallic, and they consist of iron which can contain up to 8% nickel. Copper can occur naturally as “native copper” and its ductility means that it can be hammered into simple objects such as beads for a necklace. Of course, cold-working hardens all metals (except gold) and makes them brittle. That this adverse property could be removed by heating (“annealing”) may have been a chance discovery but led onto the further discovery that the metal could be melted, and eventually shaped by casting. Pure copper melts at 1083oC and this temperature is attainable in a charcoal hearth. Exactly how or where the reduction of copper ores to the pure metal (“smelting”) first took place, the novel appearance of the metal, conspicuously different from any other minerals, doubtless made its extraction a welcome outcome. Copper was being smelted from its ore in Mesopotamia as early as the 4th millennium, and perhaps earlier still in Anatolia.

6

chapter one

The first copper ores to be smelted (malachite) were quickly followed by other ores of copper (sulphidic ores) less easy to smelt, and the ores of other metals, with varying degrees of success. Malachite [CuCO3.Cu(OH)2 ] may be regarded as copper carbonate, which on gentle heating forms copper oxide. CuCO3 = CuO + CO2 In the hearth, or kiln, charcoal burns to form first carbon dioxide: C + O2 = CO2 (carbon + oxygen = carbon dioxide) then at higher temperatures (perhaps 1000oC), the carbon dioxide reacts with more carbon to form carbon monoxide: CO2 + C = 2CO (carbon dioxide + carbon = carbon monoxide) The carbon monoxide gas reduces the copper ore to copper: CuO + CO = Cu + CO2 (copper oxide + carbon monoxide = copper + carbon dioxide) The sulphidic ores include chalcocite Cu2S chalcopyrite CuFeS2 and iron pyrites FeS2 associated with tetrahedrite Cu3SbS3 and tennantite Cu3AsS3 so there was sometimes the possibility of extracting other metals, even when only copper was being sought. The smelting of sulphidic ores in very small quantities will yield small beads of copper but any attempt at a continuous operation runs into problems with the accumulation of iron oxide; adding sand or crushed quartz (silica) removes the iron oxide as a slag, such as fayalite (iron silicate), a glass-like material with a free-running temperature (not a sharp melting-point) of around 1200oC: 2FeO + SiO2 = Fe2SiO4 Excavations in Sinai have shown that the Egyptians mined copper ores there, and used a sophisticated smelting technology.1 By about 1200 bce they were reducing the ores in bowl-shaped hearths with charcoal, 1 Tylecote, R.F. Lupu, A. Rothenberg, B. “Early copper-smelting sites in Israel”, Journal of the Institute of Metals (1967) 95, 235.

The Extraction of the First Metals

7

assisted by the blast of bellows. Iron oxide, manganese oxide, or limestone (from shells) were added as fluxes, and the liquid slag formed was “tapped off” to separate it from the copper. These furnaces resembled those subsequently to be used for smelting iron, indeed both types had to produce mostly iron silicate slags with free-running temperatures around 1200oC.

Fig. 1. A reconstruction of a bowl hearth about 80cm high operated by contemporary archaeometallurgists (a Historical Metallurgy Society meeting in the UK, 1995).

A mixture of metal and slag (from the non-metallic impurities) was formed in the furnace, and this could be subsequently broken up and the copper remelted in crucibles to purify it. The presence of a slag layer as well as removing unwanted minerals, also allows the beads of copper to join together to form a larger mass, eventually making an ingot. So the development of smelting goes hand-in-hand with that of casting. This assumes that containers suitable for melting metals in, such as crucibles, have been developed by potters. It should also be remembered that alloys generally melt more readily than pure metals so that bronzes often do not have a sharp melting-point but liquefy over a range between 800° and 1050°C. This process will lead to traces of iron passing into the copper, and programs of analysis performed by archaeometallurgists have tended to show higher levels of iron in copper objects from the Middle East. So it is thought that a slag-forming process was operating there by the early 3rd

8

chapter one

millennium bce, but did not spread to the western Mediterranean until much later.2 Other Metals Other metals were soon to follow. Once it had been realised that heating a certain mineral with charcoal, perhaps in a pottery kiln, or even in a campfire, might yield a metal, then every mineral was undoubtedly tried out in the same way. The most conspicuously coloured ones were probably first, but then every sort would be tested. Soon every metal that could be extracted this way, was being extracted; others which might pass into the copper to form an alloy without being separated themselves were also employed. Some brightly coloured pigments included ores of mercury, arsenic, and antimony, which were smelted without the metal necessarily being separated. Others were the less conspicuous but very widespread ores of iron. The metallurgy of the Mediterranean and Near East before 3000 bce was essentially a copper metallurgy. During the years of the 3rd millennium, all the metals then known—gold, silver and lead—are to be found at many sites.3 The great advance was the widespread use of copper alloys, with both arsenic and tin. At the start of the so-called “Bronze” Age, arsenical ores were perhaps smelted along with copper ores, and a marked improvement in the metallic product noticed. Copper-arsenic alloys or “arsenical bronzes” have lower meltingpoints, and greater hardness than pure copper. Such objects have been found in Egypt and Anatolia from around the middle of the millennium. The conspicuous colour of arsenic-containing minerals no doubt drew attention to them, and their addition to copper ores in smelting does not entail the isolation of arsenic as an element. But their period of use seems to have been relatively limited and they were generally supplanted by tin bronzes which seem to have been developed shortly afterwards, presumably because experiments had being carried out with all minerals, even the less colourful ones, that might be useful.

2 Craddock, P.T., “Early metal mining and production” (Edinburgh, 1995) ch.4. 3 Charles, J.A “The coming of copper and copper-based alloys” 151–182 in Wertime, T. & Muhly, J.D., “The coming of the age of iron” (New Haven, 1980).

The Extraction of the First Metals

9

Tin Tin ores are relatively scarce, only a few European deposits of tin ore (cassiterite, SnO2) are known, in Cornwall, Bohemia, and Spain. There are deposits in Egypt, but the Egyptians only seem to have used arsenical bronzes before 2000 bce, so these deposits were not exploited until after then. The earliest find of pure tin is a bangle from Level IV at Thermi on Lesbos (2300 bce ). The Bronze Age, properly called, starts in the 2nd quarter of the 2nd millennium, the period of Troy II, when tin bronzes appear and start to oust arsenical bronzes. Many earlier objects once thought to be tin bronzes were found, after analyses, not to be, and glazes once thought to be tin oxide were not tin oxide. It has been argued that their use in Europe would have been restricted until trade routes enabled tin or tin ores to be shipped from Cornwall or Spain, or from Bohemia down the Danube, to the Eastern Mediterranean, but the rapid spread of tin-bronzes has still to be satisfactorily explained.4 The properties of the alloy sought could be more easily controlled by the addition of scrap bronze to the copper melt—indeed excavated hoards tend to contain both sorts of scrap, and not tin metal, although its reduction would have been easy. Such control, of course, argues for a systematic use of trial and error, indeed, a “scientific” approach to their metallurgy, to obtain the desired properties in their alloys. Bronzes The addition of tin to copper can be brought about without the separation of tin, so the “tin trade” could equally well have been conducted in the ore as in the metal, although tin ingots are known from before 1000 bce.5 In the Heroic Age of Greece, Homer describes his heroes wearing armour of bronze, although iron seems to have been known, and a large lump of iron (probably a bloom, but perhaps a meteorite ?) was offered as a prize in the funeral games of Patroclus (Iliad, 23). Indeed the rather scanty evidence suggests that both Greek and Roman bronze armour was in no way inferior to iron armour of the quality then available.

4 Dayton J.E. “The problem of tin” World Archaeology 3 (1977) 47–70. 5 Wertime & Muhly, op. cit. 48.

10

chapter one

Perhaps it should be pointed out that the earliest items of bronze armour in Europe were not necessarily Greek, but produced in Central Europe, especially in the Carpatho-Danubian region.6 Most analytical studies to date, however, seem to have been carried out on Greek armour. Chemical analyses (but no metallography) of 13 pieces of Greek armour in the British Museum, ranging from the 7th to 4th century bce, were published by Craddock (1977). They were all apparently raised from sheet bronze, of which the tin content varied between 7.1 % and 11.4 %, but with only traces of lead, which would have made casting easier but cold-working more difficult. On the other hand, it should be noticed that the ­arrowheads had generally had higher tin (or arsenic) contents, making them harder. This is supported by the metallographic work of Smith on bronze armour.7 He studied some fragments of Greek bronze armour from Crete. The plates consisted of bronzes containing 9% to 11% tin, & very little lead; they had undergone moderate working and then annealing, and the average hardness of the flat parts was 155 VPH, comparable to that of a low-carbon steel. Some Chinese bronze swords are reported8 as having core tin contents only 8–10% while the edges (cast separately) were in excess of 20%, which might give an edge hardness in excess of 200 VPH. South Indian gongs and bowls made of a similar composition (25% Sn) have reported hardnesses of over 300 VPH, depending upon their heat-treatment.9 There seems to be relatively little information published on the metallurgy of bronze swords in Europe, although a replica of a Mycenean bronze sword was made by the Danish archaeologist Jacobsen who found it effective as a slashing weapon, severing a pig’s leg with it,10 as well as 6 Snodgrass, A.M. “The first European body-armour” 33–50 and pl.1–5 in “The European Community in Later Prehistory” ed. Boardman, J (1971). 7 Smith, C.S. “Metallographic examination of some fragments of Cretan bronze armor from Afrati” Appendix III in “Early Cretan Armorers” ed. H.Hoffmann, (Fogg Art Museum, Cambridge, Mass.1972) 54. 8 Lian, H. Tan, D. “A study on bimetallic swords in Ancient China” BUMA—V (Proceedings of the 5th international conference on the Beginnings of the Use of Metals & Alloys, Gyeongju, 2002) 227–233. 9 Pillai, R.M.et al. “Shaping of bronze in Ancient India” Trans.Indian Institute of Metals, 59 (Calcutta, 2006) 847–864. Srinivasan, S. Glover,I. “Wrought and quenched, and cast high-tin bronzes in Kerala State, India” Historical Metallurgy 29(2), 1995, 69–88. Several specimens are described with 300–350 VPH. 10 Jacobsen, H. “Er Broncesvaerdene Hug- ella Støodvaaben ?” Vaabenhistoriske Aarbøger IVc (Copenhagen, 1945) 264–271.

The Extraction of the First Metals

11

functioning in its supposed stabbing role. He had a bronze (85Cu 10Sn 5Zn) Sword cast of 62cm length and 4.3cm width, and used it in an attempt to cut through a pigʼs leg. The first blow, having struck the joint, bent the blade—thus showing the advantage of a midrib (which this sword did not have) in slashing. After straightening the blade over his knee, the second blow severed the shank apart from the shin-bone and no dent was made in the blade. Hardness of Metals The hardness of metals can be measured on the Vickers Pyramid Hardness (VPH) scale whose units are kg.mm–2. [in SI units, 1 kg.mm–2 = 9.81MPa , but most archaeological reports still employ VPH numbers] The hardness of pure copper after annealing is around 50 VPH, but after hammering cold it can be raised to 115 VPH. (Pure iron is only around 90 VPH). As little as 1% arsenic will raise the maximum hardness of hammered copper from 124 to 177 VPH. Alloying copper with 10% tin to form “tin-bronze” can raise this to 230 VPH. Copper with from 3% to 7% antimony has a hardness of from 75 to 125 VPH.11 Toughness The usefulness of a sword, or indeed any weapon, depends upon a number of factors. As well as the length and weight of the blade, which determines the energy of impact, the sharpness of its edges, which in turn depends upon the hardness of the metal used, determines how much energy can be concentrated in a narrow band or point on the target. Equally important is the fracture toughness of the metal used, that is to say, its resistance to cracking. A blade which is brittle, and snaps upon impact, is worse than useless. On the other hand, one which is ductile, and deforms upon impact, may still remain of some use in combat. The ideal blade should be both hard, resisting plastic deformation, and tough, resisting brittle failure. This combination of properties depends upon the smith being able to control the properties of the metals used.

11 Buchwald, V. “Metallurgical study of 12 prehistoric bronzes from Denmark”, J.Danish Archaeology, (1990) 64–102.

12

chapter two

Chapter two

The Smelting of Iron and the Production of Steel For much of the Ancient World, and indeed the early Middle Ages in Europe, the only ferrous material available was “bloomery ironˮ. This was the product of heating iron ore with charcoal in a small furnace, perhaps 1 or 2m high. To oversimplify a complex series of reactions, Iron oxide + carbon = carbon dioxide + iron FeO + C = Fe + CO2 The iron ore was reduced to iron, but never melted, since the meltingpoint of pure iron is 1550ºC, and therefore never entirely separated from the slag formed by non-metallic impurities. The lump (or “bloom”) of iron formed might be forged out into bars, plates, or rods but they would still be full of slag inclusions. Meteoritic iron might have been forged into very good tools or weapons by a competent bronzesmith because of its high nickel content, but these would have remained isolated and expensive curiosities.1 The “Iron Age” could not develop in Europe until techniques for the successful reduction of iron ores had been devised and disseminated. Sophisticated techniques had been developed for working copper and its alloys by the second millenium bce and could be transferred to ironworking. Unlike those of copper, iron ores are very widespread, but the extraction of iron is not so simple, because its melting-point is much higher (iron 1550oC; copper 1080oC). Any attempt to reduce iron ores in a simple copper-smelting furnace will give an unusable mixture of iron and slag. If the iron ore is of exceptional purity, then isolated fragments of reduced iron which cannot be combined by melting together, will be formed. But if slag forms, then the particles may be sintered together. Even when the ore contains no earthy matter (“gangue”) itself, there is generally sufficient silica (silicon dioxide, SiO2) present in the stones and

1 Panseri, C. “Damascus steel in legend and reality” Gladius, 4 (Madrid, 1965) 33–43.

The Smelting of Iron and the Production of Steel

13

clay which make up the wall of the hearth to react with part of the iron ore and form a slag. The iron ore is treated here as iron oxide only. 2FeO + SiO2 = Fe2SiO4 (iron oxide + silicon oxide = iron silicate) Slags are complex glass-like mixtures of oxides and silicates; the component of lowest free-running temperature that would generally be found in an ironmaking slag would be fayalite (Fe2SiO4 or 2FeO.SiO2) with a freerunning temperature around 1200oC. So, even though the iron ore might have been reduced at 700o–800oC, unless the furnace temperature reached at least 1200oC the slag would not have been liquefied and therefore could not have been separated from the iron. The use of bog-iron ores might result in an iron high in phosphorus (P) which would harden the iron somewhat. Iron ores reduced under such conditions can produce iron free from most of the slag, which when it liquefies, runs away from the still solid iron, which would be left as a lump (or “bloom”), porous in form and containing very little dissolved carbon but much entrapped slag. Such furnaces are therefore known as “bloomery hearths” and their products as “bloomery iron” or “wrought ironˮ. Repeated heating and forging would necessary to expel much of the slag and consolidate the bloom. If it was skilfully forged, the slag can be distributed in long “stringers” shaped like fibres, rather than globules, with a less deleterious effect on its mechanical properties. Wrought iron remained a favoured material of civil engineers until late in the 19th century on account of its “toughness” (defined in this case as resistance to sudden shocks) and resistance to corrosion. Until 1971 the Aston-Byers Company of the USA marketed a “puddled wrought iron” made by mixing molten pure (Bessemer) iron with molten slag. This may seem to have been a retrograde step, but in some applications (e.g railway couplings) the earlier warning of impending failure that wrought iron gave was appreciated.2 These qualities, however, were not of immediate advantage to the Ancient World. Iron smelting seems to have been first developed somewhere between the Caucasus and the Fertile Crescent early in the second millenium bce.3 From about 1900 to 1400 bce the use of iron ornaments and ceremonial weapons slowly spread; for example, the boy-king of 2 Ward, H.D. “Best Yorkshire” Journal of the Iron & Steel Institute (1972) 396 3 Wertime & Muhly, 1980, passim

14

chapter two

Egypt, Tutankhamun held an iron dagger within his third, innermost, mummiform coffin of solid gold.4 The destruction of the Hittite Empire spread knowledge of ironmaking fairly quickly around the Near East and it was exploited on a considerable scale by the Assyrians. Theirs was the first empire to make use of iron on a large scale outside China. After about 900 bce iron was commonplace, being used for swords and daggers, scales of armour, and fetters for captives, amongst other things. A hoard of some 150 tons has been excavated from the palace of Sargon (710 bce) at Khorsabad. Some of this was found to be steel but there is no direct evidence that quenching was regularly practised.5 And Smith (1968) concluded that quenching was generally avoided by the smiths of neighbouring Luristan (c800 bce) as too difficult a process to control.6 In the second half of the 1st millennium bce, the westward movement of Celtic-speaking peoples spread the knowledge of iron weapons and tools over most of Europe north of the Alps.7 The Greeks used iron extensively, although they continued to employ bronze for body armour in the form of breast- and backplates and one-piece helmets as late as the Persian wars of the 5th century bce8 and even the Romans continued to wear some bronze armour as late as the 3rd century ce. The very low-carbon iron produced in the bloomery hearth is inferior to copper in hardness as well as corrosion resistance. It is greatly increased in hardness by carburisation to steel, although even this is not necessarily harder than work-hardened bronze.

4 Forbes “Studies in Ancient Technology” (Leiden, 1964) vol. IX, 1–174.and 234–268 5 Pleiner, R. Bjorkman, J.K. “The Assyrian Iron Age” Proceedings of the American Philosophical Society, 118 (Philadelphia, 1974) 283–313. And—Maddin, R. Curtis, J.E. Wheeler, T.s. & Muhly, J.D. “Neo-Assyrian ironworking technology” Proceedings of the American Philosophical Society, 118 (Philadelphia, 1979) 369–390. 6 Smith, C.S. “The techniques of the Luristan smith” in R.H.Brill, ed. “Science and Archaeology” (Atlantic City, 1968). 7 Pleiner, R. “Iron in Archaeology” (Prague, 2000) passim. 8 Snodgrass, A.M. “Arms and armour of the Greeks” (1967) 84.

The Smelting of Iron and the Production of Steel

15

Fig. 1. Hardness of metals: the hardness of iron increases as it absorbs more carbon, and becomes steel. But steel does not become superior to bronze until it is quenched, when it can become very much harder.

Tylecote9 reported that during his experiments with a simulated Roman “high-bloomery” of 1.75 m height he obtained unforged blooms of variable carbon content and from 3.8 to 6.5 kg in weight, and even one which melted (to yield cast iron). The higher the fuel/ore ratio, the more reducing the conditions, and hence the higher the carbon content. More recent experiments by Crew10 suggest that working a bloom into an artefact entailed a wastage of three-fourths, so blooms of 3.8 or 6.5 kg might produce objects of 1 to 2kg. For many purposes, this would have been sufficient. The Roman army needed swords, nails, pilum heads, and the elements of flexible body armours, and helmets. Structures like bridges, ships, etc, were made of wooden beams fastened together by iron nails, 9 Tylecote, R.F. Austin, J.N.& Wraith, A.E. “The mechanism of the Bloomery Process in Shaft Furnaces” Journal of the Iron & Steel Institute (May 1971), 342–363. 10 Crew, P. “The experimental production of prehistoric bar iron” Historical Metallurgy, 25 (1991) 21–36.

16

chapter two

and relatively few iron objects substantially bigger than 2kg were ever needed. Some very large objects, such as anchors, or the heating blocks for public baths were occasionally made by forging numerous blooms together. Nevertheless each forge-welding to build up a larger billet will result in some loss of material by oxidation. When the Roman army abandoned Scotland in the 2nd century ce a large deposit of nails was buried at Inchtuthill, to prevent the Celtic tribesmen using them to make weapons, and analyses of some of those recovered, as well as of higher value artefacts like swords, shows a fairly mediocre quality of their metallurgy.11 Roman iron production was marked by its extensive scale rather than its sophisticated products. Their neighbours, the Teutonic tribes of Germany and Scandinavia, at first their opponents, later their soldiers, and eventually their rulers, all practised bloomery ironmaking, but on a small scale compared with the that of the Roman Empire. Thompson relates how a 5th century general, one Aristus, set off to fight the Goths with 15 000 men and no less than 520 wagonloads of weapons.12 Steel in the Occident The product of the bloomery would be a heterogeneous lump, parts of which would be of higher carbon content than others. Early smiths would have found that some samples of “iron” were harder than others, but whether they could be deliberately produced was another matter. The simplest way of obtaining steel is simply to make a large bloom, drop it into water, and then pick out the hardest fragments. These fragments would then have to be forged back together to make anything but the smallest artefact, however, so this method was an extremely inefficient one. A similar technique was used for centuries to select their steel by Japanese swordsmiths, for whom the cost of labour was not a major consideration.13 A more efficient way of proceeding could be to make an artefact of iron, and then convert part of it to steel. This might be done by forgewelding a steel edge, or other crucial part, to an iron back, or by “casecarburising” the edge; that is, heating the iron in contact with carbon for 11 Angus, N.S. Brown, G.T.& Cleere, H. “The iron nails from the Roman legionary fortress at Inchtuthill, Perthshire” Journal of the Iron & Steel Industry, 200 (1962) 952–958. 12 Thompson, E.A. “Early Germanic warfare” Past and Present, 14 (Oxford, 1966) 2–29. 13 Kapp, L. Kapp. H. & Yoshihara, Y. “The craft of the Japanese sword” (Tokyo, 1987) 65.

The Smelting of Iron and the Production of Steel

17

several hours. Adding a steel part to an iron part, however, still does not require the smith to know how to make steel. Its production may be a matter entirely of chance, as long as its presence can be identified. The “iron” bars used to hold the stone blocks of the Parthenon together were made of a banded steel, in which the layers of higher carbon content are quite randomly distributed. Varoufakis suggested that strips of iron and steel were welded back together to make clamps of banded microstructures, but the simpler explanation of a heterogeneous starting bloom seems entirely plausible.14 Other medieval artefacts (such as many examples of armour) show a banded microstructure, suggesting that they have been forged from a very heterogeneous starting material. Deliberate case-carburising depends upon the realisation that iron can be changed to steel; a much more sophisticated notion of the nature of metals. The deliberate steeling of an edge (as opposed to forge-welding a steel edge onto an iron body) argues for such an understanding. It is uncertain when this understanding developed. It may have been developed as early as the 10th century bce; it was certainly developed by the 4th century bce. It was practised regularly throughout the Middle Ages, and was described by Theophilus, around 1100 ce as an appropriate techniques for small tools, such as files. It was also suitable for the cutting edges of swords and knives, but less suitable for armour, and is seldom found therein.15 The absorption of carbon in the solid state is very slow, and hence a concentration gradient would be established, and in all but the smallest articles, heating for sufficient time to carburise the centre moderately would carburise the edges excessively. An alternative method was to carburise small pieces of iron and then forge-weld them together (“piling”). Some Celtic smiths had attempted to overcome the difficulty of making a steel blade by forge-welding several small pieces together to make a piled blade, with bands of higher- and lower-carbon content. The laminated structure is still visible on the surface, especially after corrosion. Several such weapons have been found, dating back to the 6th century

14 Varoufakis, G. “The iron clamps and dowels from the Parthenon and Erechthion” Historical Metallurgy, 26 (1992) 1–18. 15 Williams, A.R. “The Knight and the blast furnace” (Leiden, 2003) especially the cross-sections shown in chapter 4.3.

18

chapter two

bce and techniques like piling remained in use for many centuries.16 By achieving a more uniform distribution of carbon a steel of moderate hardness was attainable without heat-treatment, which was not generally mastered for a long time. Indeed piling was beyond the capabilities of many Celtic smiths who simply made swords out of wrought iron but it was a feature of blacksmithsʼ work throughout the Migration Period and Early Middle Ages in Europe. For example, in 225 bce a Roman army fought at the battle of Telamon an army of Celtic Gauls who slashed at the Romans with their long iron swords, which periodically bent and allegedly had to be placed on the ground and straightened by the foot. Pleiner’s experiments (see Chapter 4) were subsequently to show that this was something of an exaggeration by the historian Polybius. An iron sword might bend a few centimetres, but no more, and it remained entirely usable. On the other hand, the snapping of an incorrectly hardened sword would have been very undesirable from the warrior’s point of view. Certainly, it would be a very long time before the production of steel could be anything other than adventitious. But the abundance of iron ores, however, meant that iron tools and weapons could be made much more cheaply than those of bronze, and would therefore be available to many more people, once the techniques of smelting and forging were generally known. So for many users, stone tools and weapons were succeeded, not by bronze but by iron ones, even though those iron tools and weapons were little better, if at all, than those of bronze. Iron weapons and tools did not become superior to bronze until the discovery was made that quenching after carburisation resulted in a dramatic increase in hardness. The process is a difficult one to manipulate, however, as the hardness is due to the formation of martensite, an excess of which leads to embrittlement. Quenching is mentioned by Homer in perhaps the 10th or 9th century bce17 and quenched edges have been detected on excavated specimens from the 10th century bce onwards18 16 Tylecote, R.F & Gilmour, B.J. “The metallography of early ferrous edged tools and weapons” (Oxford, 1986) British Archaeological Reports, 155. 17 Odyssey, IX, 459. 18 Carpenter, H. Robinson,J.M. “The metallography of some Ancient Egyptian implements” Journal of the Iron & Steel Institute (1930) 417. They examined a selection of iron objects, dating from around 1200 bce to 200 ce. All the specimens consisted of wrought iron carburised to varying extents. The earliest specimen to show quenching dated from about 900 bce. Also see Williams, A.R. Maxwell-Hyslop,K.R. “Ancient steel from Egypt” Journal of Archaeological Science, 3 (1976) 283. Four out of a group of seven tools that might have been Assyrian or Roman (but unfortunately can only be dated between 7th and 3rd century bce) showed definite evidence of carburising and quenching

The Smelting of Iron and the Production of Steel

19

but the difficulty of controlling the carbon content of steel meant that quenching was to remain a hit-and-miss process, and therefore avoided by many smiths, for a long time to come (without the presence of some carbon as an alloying element, iron does not harden on quenching). What can be seen microscopically: Ferrite crystals are pure iron, and appear as irregular white areas. Only the grain boundaries are actually visible. Hardness = 80–120 VPH.

Fig. 2. This sample of medieval iron (magnified approximately 200 times) shows ferrite grains and elongated inclusions of slag

Pearlite is a mixture of iron and iron carbide containing up to 0.8%C. It has a grey, lamellar appearance, and a hardness up to 250–300 VPH. It may be formed when red-hot steels are cooled in air. The form of solid iron stable at high temperatures (austenite) can dissolve a good deal of carbon. On equilibrium cooling, carbon is rejected as iron carbide and the low temperature form of iron (ferrite) is formed. A steel containing 0.8%C (eutectoid) transforms completely to pearlite at a constant temperature. Most steels of below 0.8%C (hypoeutectoid) contain mixtures of ferrite and pearlite.

20

chapter two

A hypereutectoid steel (above 0.8%C) usually contains a mixture of pearlite and extra cementite, sometimes appearing as a network around the pearlitic areas or sometimes as needles within them.

Fig. 3. Pearlite × 800. The alternate layers are made up of ferrite and cementite.

Fig. 4. A typical medieval steel, containing ferrite and pearlite in varying proportions; the carbon content varies from around 0.2% to 0.6%. A forging line can be seen close to the lower surface of the section (magnified × 25).

The Smelting of Iron and the Production of Steel

21

Martensite is formed in quenched steels. The rate of cooling generally has to be rapid to avoid the formation of pearlite. It appears like laths, in steels of up to 0.5%C, with some triangular symmetry. In steels of medium-carbon content (0.5%–1.0%C) it can display an acicular appearance.19 Its hardness depends upon the C% and can vary from 200 to over 800 VPH. This is associated with extreme brittleness (a razor blade, for example). Gentle reheating (in the region between 200ºC and 600ºC) reduces the hardness somewhat but increases the toughness considerably. The preferred modern technique of hardening steels is full-quenching followed by tempering, but this depends upon knowledge of the carbon content and the accurate control of temperature and time. All of these requirements were difficult to meet in Medieval Europe, and so other techniques of heat-treatment were often employed.

Fig. 5. Graphs of tempering modern steels. HB is Brinell hardness. HVN is Vickers Hardness Number. ○ and ∆ refer to a steel of 0.52%C and 0.93% Mn. ● and ▲ refer to a steel of 0.48%C and 0.59% Mn. (adapted from the ASM Metals Handbook, vol.1)

19 Honeycombe, R.W.K. “Steels” (1982) passim

22

chapter two “Slack-quenching”

This was the name given to a method of heat-treatment obsolete by the 20th century, but still in use by many craftsmen. In “full quenching”, the red-hot artefact is plunged into water, or some other liquid, which cools it so quickly that only martensite is formed. In “slack quenching” the artefact is cooled less rapidly, so that pearlite, and perhaps bainite, forms as well as martensite. This mixture is less hard, but also less brittle, and obviates the need for a tempering process. A delayed quench, or an interrupted quench, will result in a slower overall rate of cooling, and so they are means of slack-quenching. If the red-hot artefact is plunged into the cooling liquid, and then withdrawn, then two things will happen; martensite will start to form, but not all the steel will form martensite, and the remainder will form pearlite. Then the heat stored in the interior will travel out into the martensite, and temper it.20 This is “time-quenching” and might be regarded as a variant of “slack-quenching.” Many of the “hardening recipes” collected so assiduously in the Middle Ages are in fact recipes for slack-quenching. Quenching and tempering separately do not seem to have become common until the 15th century.

Fig. 6. Martensite (the dark-etching areas on the left) in a 15th century sword. Since the low-carbon areas have not hardened on quenching, they remain as ferrite (white). (scale bar = 50 microns) 20 Burns, J.L & Brown, V, “Time quenching” Trans.American Society for Metals, 28 (1940) 209–229 and see p. xxxx

The Smelting of Iron and the Production of Steel

23

Bainite may also appear in heat-treated steels. Non-equilibrium cooling, but at a rate not fast enough to form martensite, may result in bainite being formed. It has an appearance somewhat like needles. Its hardness is usually between 200–400 VPH.

Fig. 7. Bainite

Cementite, iron carbide, Fe3C, may be found as part of the mixture in pearlite, or as separate areas, sometimes appearing as needles in hypereutectoid (above 0.8%C) steels,21 and also in cast irons (over 2%C). Annealing may cause the needles to spheroidise, and the cementite then appears as globules. The hardness of cast iron might be 400 VPH, although pure cementite is much harder. Cast iron does not harden on quenching.

21 Examples can be found in Chapter 13; particularly sword I.2.

24

chapter three

Chapter three

Different Ways of Making Steel— Eastern and Western Steelmaking Steel in the Orient This book is principally about European swords, but a note on how swords were made elsewhere may be relevant. There are basically three ways of converting iron to steel, and their practice in different parts of the world led to the distinctive characteristics of the swords made there. First, the blooms of iron can simply be left in the bloomery until more carbon has been absorbed. Then the steely parts of the bloom can be separated somehow. Pieces of iron & steel can be welded together to form a sword, which is then hardened by some form of quenching. This was the procedure in both Europe and Japan. In Europe, it led to the development of the piled and then the pattern-welded sword. The development of larger bloomeries enabled steel to be made in larger quantities, so that the later European Middle Ages (14th century) saw the development of suits of steel plate armour as well as all-steel swords. In Japan it was taken to its highest level, where it formed the basis of swordmaking almost until modern times. Second, small pieces of iron can be separated from the bloom and then heated in closed crucibles, with materials which contain carbon, until a partial or total liquefaction takes place. Rapid absorption of carbon can lead to the formation of a cast steel (“crucible steel”), with a very high carbon content, which needs little further hardening. Some steel of this type seems to have found its way into Europe during the early Middle Ages. Controlled cooling and forging can then develop a pattern resembling watered silk on the surface of the blade (“wootz”, misnamed “Damascus steel”). This was the procedure in Iran, Central Asia and India, where it remained in operation until the 19th century, with products which were high in quality but small in scale.

Different Ways of Making Steel

25

Third, the bloom of iron can be left in the furnace until it melts and forms “cast iron”. This became the usual starting point for ferrous metallurgy in China. Crucible Steel The second of these approaches, employed in India, Persia, and Central Asia, was to make crucible steel. This may have been started by the practice of heating pieces of bloom (to around 1300–1400ºC) in covered crucibles to “purify” them with medicinal herbs. In the presence of carbonaceous plant material, the iron might absorb enough carbon to wholly or partially melt into an ingot of steel, which would have been left after the crucible was allowed to cool down, and broken open. These ingots were exported to centres of arms manufacture where they were carefully forged, at a low temperature,1 into sword blades. Crucible steel was a high-carbon (over 1.0%C, and sometimes as high as 1.8%C), or hypereutectoid steel, which already attained 300–350 VPH in hardness, nor did any amount of sharpening ever remove its edge. It is important to remember that the melting range of steels falls with increasing carbon content. Some of the specimens of high-carbon blades (described in Chapter 8) are far from homogeneous, and may have resulted from the imperfectly melted contents of a crucible. Indeed, it may well have been several centuries before a temperature high enough and a time long enough were employed to form an entirely homogenous cast steel; for example, a steel of 1.8%C will not have a sharp melting-point but a “slushy” range of approximately 1200º-1400ºC. Solidifying such a melt may well explain the heterogeneous nature of some of these specimens.2 Nevertheless, even an imperfectly homogenised crucible steel would have been considerably better than the fragments of bloomery steel which was all that was then otherwise available.

1 Because of its lower melting temperature range, a high-carbon steel, despite being harder, can only be forged safely at a lower temperature than usual for iron. This paradox doubtless baffled many swordsmiths. 2 Paufler, P. et al. “Microstructure of a genuine Damascus sabre” Cryst.Res.Technol. 40, 9 (2005) 905–916. He reports on the different carbon contents reported in samples from the same sword by Zshokke 1.73%C, Verhoeven 1.79%, and Kochmann 2.24%.

26

chapter three Recipes for Producing Crucible Steel

Crucible steel could have been made in one of two ways: (i) by heating wrought iron with organic matter, as described by al-Tarsussi in the 12th century or (more easily) (ii) by heating wrought iron with cast iron, as described by al-Biruni in the 11th century. Pliny describes “Seric iron” as the best, as follows “Among varieties of iron that of the Seres (possibly the Chinese?) carries off the palm and is exported by them together with garments and pelts, while the second best comes from Parthia. No other kind is made of sheer steel for the rest contain an admixture of iron of a softer character. Within our empire excellence depends upon a first class ore...as in Noricum ..or the process of manufacture as at Sulmo.” “Seric iron” could have been crucible steel, or even Chinese white cast iron3 but we cannot identify it with either for certain. Somewhat later, a treatise once ascribed to Zosimos, the 3rd century alchemist, describes the manufacture of crucible steel. “The tempering [sic] of Indian iron. Take 4 lb of soft iron, and the skins of myrobalans, called elileg, 15 parts, belileg, 4 parts, and 2 parts of glassmakers’ magnesia. Grind all together and mix with the 4 lb of iron. Then place in a crucible and make it level … .put on the charcoal and blow the fire until the iron becomes molten and the ingredients become united with it. Note that the 4 lb of iron will need 100 lb of charcoal … such is the premier and royal operation, which is practised today and by means of which they make marvellous swords. It was discovered by the Indians, and exploited by the Persians, and it is from them that they are coming.”

It is clear from the reference to glassmaking that “magnesia” here means manganese dioxide (“glassmakers’ soap” used to decolourise glass by neutralising the green or brown colour due to iron compounds). The plant material is simply there to produce carbon. “Tempering” is used in a different sense to the modern meaning of the world. It is now used to refer to the reheating of a steel after quenching to reduce its hardness. In the Medieval period it is frequently used to mean the initial quenching, or 3 Bailey, K.C. “The Elder Pliny’s chapters on Chemical subjects” 2 vols (1929–32) II, 145. and also see: Bronson, B. “The making and selling of wootz, a crucible steel of India” Archaeomaterials, 1 (Philadelphia, 1986) 13–51.

Different Ways of Making Steel

27

plunging red-hot into cold water. In this case it refers to the making of the steel in the first place. However, while this treatise was ascribed to Zosimos by his editor Berthelot4 in the 19th century, a more recent critical edition by Mertens5 now regards it as the work of Zosimos by association only, since it is part of a larger codex (Marciana 299), containing works ascribed to many other authors, and dating from the 9th–10th centuries. But then it is at least certain that the making of crucible steel was known to some alchemical authors by the Early Middle Ages. In the 9th century, the Arabic writer al-Kindi (801–870) gives a detailed description of different types of iron and steel and their use in making swords, in his letter to the Caliph of Baghdad (probably Al-Mutasim, 833– 842). I have used Wiedemann’s translation here,6 although a new edition of al-Kindi’s treatise has recently appeared, which supplies further details.7 “The iron (hadid), from which the swords are forged, falls into two main classes, the natural ones (madini), which are found in the land [i.e. iron won out of the iron ores directly], and the artificial ones that are not found in the land. The natural one is again separated into two types, firstly the saburqani, that is the male one, (mudakkar), hard and which can be hardened according to its nature (saqa), and secondly the barmahani, this is the female one, that cannot become hardened itself according to its nature.”8

Swords may be forged from each of these iron types [bloomery iron and bloomery steel] or a mixture of them. But then he goes on “The iron, that is not natural, is steel (fulad). It is understood to have been purified. It is produced out of natural iron into which one throws something, that cleans it and hardens it, so that it becomes strong and flexible, and the “firind” (damask) appears on it.”

This is crucible steel. The second part of the Book of al-Kindi deals with the different types of swords. About the different types of steel and iron, we are told the following: 4 Berthelot, M. “Collection des Anciens Alchimistes Grecs” (Paris, 1883) vol. 3, p.328 5 Mertens, M. “Les alchimistes grecs; IV, 1 Zosime”. (Paris 1995) 6 Wiedemann, E. “Über Stahl und Eisen bei den muslimischen Völkern.” Beiträge zur Geschichte der Naturwissenschaften. xxv. Published as–Sitzungsberichte der Physikalisch-medizinischen Sozietät in Erlangen, 43, (1911) 114–131.and see vols.34–37 (1902–06). 7 Hoyland, R.G & Gilmour, B. “Medieval Islamic swords and swordmaking” (Oxford, 2006). 8 Hoyland & Gilmour render these as shaburqan and narmahan respectively.

28

chapter three “An especially splendid one was the Indian steel; that of al-Hind (India) is used only for steel ( i.e. arms) and mirrors are produced from this.”9

al-Kindi goes on to say that “swords may be made out of shaburqan by Rus, Slavs & Byzantines”. Medieval European swords were often made of steel, or at least their edges were. And that “swords may be made out of narmahan by Byzantines & other foreigners”. Swords made only out of iron would be poor swords indeed, but many were.10 Two centuries later, al-Biruni (973–1048) quoting earlier writers, describes how crucibles were made, filled with narmahan together with marcasite, antimony, and magnesia, then covered and heated in the furnace until the contents melted and mixed. Much of this may be disinformation, but he also mentions that the area of Herat is especially noted for making crucible steel by melting together a mixture of narmahan and dus—the liquid “which comes out of iron when it is melted”. We must assume that this is cast iron, because adding slag to iron would not form anything useful. (NB. mixing 3 parts of iron of 0%C and 1 part of cast iron of 4%C will give 4 parts of a steel of 1%C.) The products are called “eggs” on account of their shape. “Either the mixture melts completely in its crucible, and unites; this product is good for making tools such as files; or it does not mix completely. This produces a surface pattern or damask—firind.11 This is used for making highly sought-after sword blades.” It seems the former steel was also used for making swords, but they cannot be as readily identified. But it seems possible that the pattern might have been the result of imperfect melting rather than that of wootz formed by very slow cooling after complete melting. Wootz is crucible steel with a pattern like “watered silk” on the ­surface. According to al-Biruni, the product was made around Herat and exported via North India to Persia & other Muslim lands. The Persians traded in crucible steel, and a Persian blade of the 7th century from Dailaman, in the British Museum, has recently been analysed, and found to be made of crucible steel.12 France-Lanord published an analysis of a

9 Wiedemann, op.cit 120 10 See Chapters 8 & 13 for microstructures. 11 Allan, J. W. & Gilmour, B. “Persian Steel; the Tanavoli collection” (Oxford, 2000) 62. 12 Lang, J. et al. “New evidence for early Crucible steel” Historical Metallurgy 32, (1998) 7–14.

Different Ways of Making Steel

29

Luristan blade of uncertain date, but possibly much earlier, and which was also made of a high-carbon steel.13 Murda ibn Ali al-Tarsusi, dedicated his treatise on Arms & Warfare, written around 1190 ce, to Saladin. He uses numerous sources, some of which he acknowledges. A recipe for steel is said to have been taken from Jabir’s “Book of the Seventy”. Cahen gives the Arabic text in full, and a French translation of the extracts. He describes in some detail the manufacture of crucible steel for swords.14 Both the methods of al-Kindi and al-Biruni are given. “The sabre and its manufacture: different mines and centres of production provide metals of various sorts. Those found in India and China are of a quality unequalled, those of the Maghreb, Spain and Africa are of an inferior sort; in Egypt there are mines whose products are coveted by none. When it is made with these minerals, the fulad (steel) is the most murderous, the most powerful, the noblest, and the highest of fabrications that can be made. The composition has various sorts, and the work of alteration depends on the composition, and the virtues of their rival characters are determined by the virtues of the mixtures formed at the moment of arrangement. … What I relate has come from the sages of old to my knowledge, and they describe the specialist artisans, who are among the most ingenious of the moderns.  (A description follows of the genuine) fulad, and how it is treated at the moment of its combination, with drugs which dry the humidity and make moist the dryness for natural equlibrium, driving out the earthy materials detri­mental to its purity, which are mixed with it from the mine, and purifying it until it catches the light, and reveals its internal structure [my italics].  First one takes soft iron (narmahan), and if it comes from the heads of old nails, it will be of the best sort, Then one throws upon it 17 dirhems of myrobolan of Kabul and as much belleric. Then the iron is put in a vessel, and is well cleaned with water and salt, then it is mixed with the aforesaid drugs, which are placed together in a crucible, and 1½ dirhems of powdered magnesia are sprinkled on, then the fire is blown on it at the foundry, where it may be melted & formed into an egg. This takes several days. Then it is filed and shaped into a sabre. This is deadly dangerous.  Another method: Take 3 ratl of iron (narmahan) and 1 ratl of hard iron (shaburqan); collect all together in a crucible and throw in 5 dirhem of magnesia and a handful of the rind of the sour pomegranate; blow the fire in the foundry to bring it to fusion, and form it into an egg; then take it out and make a sabre.” 13 France-Lanord, A. “Le fer en Iran au premier millénaire avant J-C” in Revue d’histoire des mines et la métallurgie, 1,1, (1969) 75–126. 14 Cahen, C. “Un traité d’armurerie composé pour Saladin” Bulletin d’Études Orientales, XII (Beyrouth, 1947–48) 103–163.

30

chapter three

The 14th century writer al-Jildaki quotes from a book (Kitab al-Hadid) “On Iron”, which was ascribed to Jabir, and was included in the English edition of the latter’s works printed in 1678, which describes the making of both white cast iron and crucible steel.15 The historical crucible steel industry has been described in recent years by various archaeometallurgists, such as Bronson,16 Craddock,17 and Feuerbach.18 The earliest production sites so far excavated are at Tamil Nadu in India (uncertain, but possibly 3rd century) Sri Lanka (6th–12th century),19 but now also the Iranian area has shown sites (9th-12th century)—Merv, now in Uzbekistan, and Ferghana, now in Turkmenistan.20 The Mongol invasions of the 13th century may have put an end to this central Asiatic production. Many Indian sites were documented in the 19th century while they were still in production, especially around Hyderabad and Mysore. A detailed description of the process at Ghattihosahalli in Karnataka, which remained in use until the late 19th century, and where recent excavations have unearthed crucibles, etc is given by Anantharamu et al. (1999). Iron was mixed with wood chips and fired in crucibles of clay tempered with rice charcoal.21 Crucible steel production at some wind-powered sites that have been excavated recently in Sri Lanka yielded ingots of only around 0.5 kg.22 This would have been even less sufficient to make a sword, but a skilful smith might weld several ingots together to make a blade. It might seem to us to have been easier to make the ingots larger in the first place, but then the difficulty of raising a larger crucible to the neces15 “The Works of Geber, Englysh’d by Richard Russell” (1678, reprinted 1928) 231.also see Hoyland, op. cit. 144. 16 Bronson, B. “The making and selling of wootz. A crucible steel of India.” Archeomaterials 1, (Philadelphia, 1986) 13–51. 17 Craddock, P “Early Metal Mining and Production” . (Edinburgh, 1995) chapter 7, and on crucible steel 275–283. 18 Feuerbach, A. “Early Islamic crucible steel production at Merv, Turkmenistan” in Craddock & Lang (2003) 258–266. 19 Wayman, M.L et al. “Crucible steelmaking in Sri Lanka” Historical Metallurgy 33 (1999) 26–42. 20 Craddock, P. & Lang, J. (eds) “Mining & Metal production through the ages” (2003) 231–257. 21 Anantharamu et al. “Crucible steel of Southern India” Historical Metallurgy 33 (1999) 13–25. 22 Juleff, G. “Crucible steel production at Hattota Amune, Sri Lanka” in Conference Proceedings (BUMA—VII) Bengaluru, 2009 (Beginning of the Use of Metals & Alloys).

Different Ways of Making Steel

31

sary temperature could have posed a problem. With the passage of time techniques were doubtless developed for using forced draught and more refractory crucibles to attain higher temperatures. It should be noted that, according to several Indian investigators, iron and steel in India prior to the 6th century were bloomery products only,23 and not crucible steels. Prakash and others examined numerous artefacts from various dates up to the 6th century ce. Iron, low- carbon-steels, some medium carbon steels especially from later contexts, but apparently no hypereutectoid (crucible) steels were found. There were many piled structures and a few heat-treatments. Similarly, Mahmud described (with diagrams) a smelting furnace and a “refinery” in which the metal “is never completely melted” so this would seem to imply the removal of slag from a bloom by reheating. In the 19th century, before its final extinction in the face of competition from the far cheaper British ingot steel, Egerton24 described the manufacture of steel at Hyderabad—which was said to export the best steel to Persia. It was converted in crucibles which were cooled slowly, to form crystals (jauhar) in a 1 ½ lb (680g) cake; then covered with clay & annealed for 12 to 16hr. and repeated until soft enough to forge. Each sword was made of 2 small cakes welded together. Hyderabad during the 17th and 18th centuries was reputedly the source of the best swords in India, and a recent study has shown that 6 out of 10 swords from the Nizam’s Armoury were made of crucible steel25 but not all were hypereutectoid. Some of these swords were then being heattreated to hardnesses of over 500 VPH. If swords much harder than the (very respectable) 350 VPH were demanded, then the heat-treatment of a somewhat lower-carbon steel might have been the best way to proceed. But such a steel would have required a higher temperature for its preparation. If the cast steel required was of only 0.9%C (and therefore much easier to heat-treat) then the operating temperature needed would have had to be raised to nearly 1500ºC.

23 Agrawal, O.P. Narain, H. Prakash, J. “Development of iron metallurgy in Ancient India” Archeometallurgia ricerche e prospettive, ed.Sanpaolo, E.A. (Bologna, 1992). See also—Mahmud, S.J. “Metal technology in Medieval India” (New Delhi, 1988). 24 Egerton, W. “An illustrated handbook of Indian arms” (1880) 56. 25 Williams, A. & Edge, D. “The metallurgy of some Indian swords from the Arsenal of Hyderabad and elsewhere” Gladius, 27, (Madrid, 2007) 149–176.

32

chapter three

Similarly, to avoid forming large crystals of cementite, the historical use of crucible steel in drawing wire for making vina or sitar strings26 might have necessitated a near-eutectoid steel. The swords whose microstructures are illustrated here seem to have been such steels. Examples of crucible steel in Indian swords: A Shamshir blade with narrow fullers. A “firanghi” style blade—probably of the 18th century. From the Arsenal of Hyderabad. The microstructure contains very uniform fine pearlite with occasional isolated grains of grain-boundary cementite and no visible slag inclusions. This is apparently a crucible steel which has been given a fast air-cool after forging. Microhardness (Vickers, 100g) range 345–403: average = 371 VPH.

Fig. 1. The sword blade

Fig. 2. Microstructure (scale bar = 50 microns)

26 Buchanan, F.  A Journey from Madras through the Countries of Mysore, Canara and Malabar (London, 1807) 361–2. I am grateful to Paul Craddock for drawing my attention to this reference.

Different Ways of Making Steel

33

A Shamshir (sword) blade with three fullers. A blade of the late 17th or early 18th century in the “firanghi” style. From the Arsenal of Hyderabad. The microstructure contains very uniform tempered martensite, with very few slag inclusions. This is apparently a crucible steel which has undergone some form of heat-treatment to further harden it. This blade must represent one of the best swords ever made anywhere. Microhardness (Vickers, 100g) range 457–599: average = 550 VPH.

Fig. 3. The sword blade

Fig. 4. Microstructure (scale bar = 50 microns)

Mongol Armour Some time ago, a Mongol scale armour was excavated from a 13th century site in Tuva, and studied by A. Matveev.

34

chapter three

Fig. 5. The scale armour

Professor A. Lemeshko27 was instrumental in enabling a scale from the armour (measuring approximately 75 × 92 mm) to be analysed in the Conservation Department of the Wallace Collection, London. The analysis was carried out non-destructively by microscopic examination of a side edge of the plate. The section shows a homogeneous pearlitic microstructure (corresponding to a steel of around 0.6%C) with almost no slag inclusions.

Fig. 6. Microstructure of the armour scale. Fine pearlite and ferrite, with almost no slag (scale bar = 50 microns)

Since the Mongols had conquered much of the crucible steel-producing regions of Central Asia, it is tempting to ascribe a crucible steel origin to this armour. 27 Matveev, A. “КОПЛЕКС МОНГОЛЬСКИХ ДОСПЕХОВ XIII В. ИЗ ТУВЬІ (общаяха­ ратерисика)” (Mongol armour of the 13th century from Tuva) 232–236 and plate 1; published in “Drevnie kultury Centalnoy Azii i Sankt-Peterburg” (“Ancient Cultures of Central Asia and St.Petersburg”) 1998; a celebration of 70-years jubilee (postmortem) of Alexander Danilovich Grach, the archeologist who discovered the armour.

Different Ways of Making Steel

35

Indian Swords in Medieval Europe These sources support the contention that crucible steel was known to be used in making swords in Medieval Europe. The Franks valued Saracen swords, and one of Charlemagne’s vassals captured “spatha india cum techa de argento parata”. Count Eccard of Mâcon left a “spatha indica” in his will, as well as “tabulas saraciniscas”. It has been suggested by the historian who collected these references that this referred to a Saracen sword, but it may just as well have meant what it said, an Indian sword.28 Dinnetz has collected numerous literary references to these blades in medieval Spain.29 We may mention, among others, these interesting quotes: (i) (from the 11th century): Ibn al-Labbana who referred to “Indian swords”. (from the 12th century): (ii) Abu Bakr al-Sayrafi who suggested that Indian (hinduwani) swords should be used because they were sharper than other swords and better able to pierce the heavy armour worn by the Christian soldiers. (iii) Al-Zuhri who said that Seville produces “Indian steel”. (iv) Ibn-Abdun who said that makers of scissors, knives, scythes, etc.in Seville may only use [materials translated as] steely iron mudakkar or cast steel amal al-tara’ih. (from the 14th century): (v) Ibn Hudhayl who described Frankish swords as mudakkar with “steel edges on an iron body, unlike those of India”. Unfortunately, no surviving artefacts like these have yet been identified, but who knows what archaeology might yet find in Spain? Crucible steel could have been imported from India to Spain via Egypt and the Maghreb—a route not readily available to merchants from Western Europe, after the alternative route via the Volga was closed.

28 Coupland, S. “Carolingian Arms & Armour in the 9th century” Viator, 21 (Berkeley, 1990) 29–50. 29 Dinnetz, M.K. “Literary evidence for crucible steel in Medieval Spain” Historical Metallurgy, 35 (2001) 74–80.

36

chapter three Western Swords Traded Eastwards

Al-Biruni adds that the “Rus” had invented a process of weaving together strips of soft iron and steel to produce a serviceable blade, less likely to break than an oriental one, especially in the very cold northern winters. Pattern-welded blades may have been inspired by damask blades, but their composition is quite different, and the patterning itself seems to have been abandoned by the 11th century. The carbon content of Western blades is much lower, but their hardness can be increased by quenching (an easier process when only thin bands of steel along the edges are involved). Despite the evident superiority of crucible steels, Western blades offered a useful combination of properties, at presumably a much lower price, than Oriental ones, and there are references to their being exported to Muslim lands;30 for example, Saracen pirates demanded 150 Carolingian swords as part of the ransom for Archbishop Rotland of Arles in 869. This was to be paralleled much later when European merchants exported many blades with German & Italian marks to Mughal India.31 Damascus Steel An important subdivision of crucible steel was wootz, sometimes called “Damascus steel” which was used to make swords of unique quality and correspondingly high price, recognisable by a characteristic pattern (caused by large crystals of cementite, Fe3C) on their surfaces, reminiscent of “watered silk”. These patterns needed etching with weak acids to make them appear. al-Biruni describes how they may be worked, and gives a recipe of one Mayzad ibn Ali, a smith from Damascus, for producing imitation blades from soft iron by etching the patterns with an acidic solution of zaj, or ferric sulphate. Wootz was made, as other crucible steel was, by melting iron with carbonaceous material in a sealed crucible over several days until it wholly or partially melted into a cake of steel, but then allowed to cool extremely slowly (over days, rather than hours). These cakes were exported to centres of arms manufacture, such as Damascus, where they were carefully 30 Zeki Validi, A. “Die schwerter der Germanen nach arabischen Berichten des 9–11 Jahrhunderts” Zeitschrift für Deutsche Morgenland gesellschaft, 90 (1936) 19. 31 See for example North, A. et al. “Swords and hilt weapons” (1989) 191.

Different Ways of Making Steel

37

forged, with considerable difficulty, into sword blades. Since the meltingpoint of steel falls with increasing carbon content, a lower temperature than usual has to be employed to forge a blade of higher carbon content than usual, notwithstanding its hardness. This forging broke up the cementite (iron carbide) network left over from the casting, reducing brittleness, and producing the characteristic pattern (“watered silk”) visible after etching on the surface of the blade. The blade so formed needed no further heat treatment to harden it, although later attempts may have been made. Articles by Belaiew32 and Panseri33 were among the first to show the microstructure of Damascus steel, and they were followed by FranceLanord34 and Piaskowski.35 Extensive efforts by Verhoeven and Pendray36 to recreate this process have led to a series of very detailed papers on its possible metallurgy. However it has become clear, especially after the archaeological work in the Middle East cited above, that wootz was only a small part, albeit a special part, of a crucible steel industry, the extent of which has been discovered in recent years. The fame of the easily identifiable Damascus steel has tended to overshadow the use of crucible steel in general.37 It may be that most of the crucible steel made did not possess any very characteristic pattern visible on the surface of the blade, and therefore has not been recognised; so the quantity of crucible steel employed may well have been considerably underestimated in the past. Not only swords, but helmets and even shields38 were being made of wootz in Central and South Asia by the 18th century. From a purely mechanical point of view this was a retrograde step, as the large crystals of cementite which generated the pattern would have tended to increase the brittleness of the blades.

32 Belaiew, N.T. “Damascene steel” Journal of the Iron & Steel Institute, (1918) 417–439 and (1921) 104, 181–184. 33 Panseri, C. 1965 “Damascus steel in legend and reality” Gladius, 4, Caceres, 5–66. 34 See above (ref 13) and also see Zschokke, B. “Du damasse et des lames de Damas” Revue de Métallurgie 21 (Paris, 1924), 635–669. 35 Piaskowski, J. “Metallographic examination of two Damascene steel blades” Journal for the History of Arabic Science, 2 (Aleppo, 1978) 3–30. 36 Verhoeven, J.D. Pendray, A.H. & Peterson, D. 1992 “What is a Damascus Steel?” Materials Characterisation (New York, 1992) 29, 335–341 and ibid. (1993) 30, 175 and 187. 37 al-Hassan, A.Y. & Hill, D.R. “Islamic technology” (Cambridge and Paris, 1986). 38 Shield OA2271 in the Wallace Collection, London, contains a central disc of crucible steel weighing approx.0.85 kg.

38

chapter three

Fig. 7. A “Damascus” blade from the Wallace Collection, London. OA 1404. Copyright Trustees of the Wallace Collection.

China In China, although bloomery iron was known, a liquid product of the furnace was more usual. Such a furnace was called a “blast furnace” although any furnace with bellows to supply air might be so called. It differed from the bloomery only in a somewhat larger size and a different fuel:ore ratio. Since the melting-point of steel falls with increasing carbon content, if it absorbs over 2%C, it will then melt at 1150ºC, forming a liquid called “cast iron” or “pig iron”. This may be cast into shape, but cannot be forged, nor hardened by quenching, and so would seem useless for making weapons, although Rostoker has pointed out that while it was very brittle, it was by no means useless, especially for stabbing weapons, being both hard and cheap.39 So much cast iron was converted to wrought iron by fining (blowing air over the liquid). The finery was in use by 1st century bce, and with its spread, the bloomery gradually became uncompetitive. The Indirect Process (Blast Furnace + Finery) offered considerable economies of scale, and thus favoured large-scale production. This may well have been a factor in the rise of the state of Qin in the 3rd century bce, and the Han state established a monopoly on iron in 117 bce.40 An alternative to fining was make “malleable cast iron” by annealing to break down the very brittle cementite. White cast iron has most of its

39 Rostoker, W. “White cast iron as a weapon and tool material” Archeomaterials, 1 (Philadelphia, 1967) 145–8. 40 Wagner, D., “Iron & Steel in China” (Cambridge, 2008) 177.

Different Ways of Making Steel

39

carbon present as cementite, Fe3C. In grey cast iron this has mostly been broken down to iron and carbon (graphite): Fe3C → 3 Fe + C This was used from 4th century bce for tools and even for swords, and this continued into the Tang & later periods.41 Cast iron might be converted to steel if the process of decarburisation could be halted at a carbon content of around 0.5%–0.8% rather than continuing to complete removal. This difficult requirement might have been brought about either by very skilful fining or perhaps by co-fusion (heating cast iron with wrought iron). Surprisingly, crucible steel seems not to feature greatly although such a technique might have been expected to be familiar to a culture that produced porcelain—which needs to be fired in the region of at least 1300o–1350oC. But, according to Wagner there is no direct evidence that cast steel was made in China. Central Asian products from centres like Merv (10th century) and Ferghana (9th–13th centuries) would probably have been known. Exports of bin iron from Persia and Jaguda (Ghazni) to China in 6th–7th centuries are recorded.42 This was an imported steel of high quality. Curiously, bin iron disappears from Chinese sources after the 7th century, then reappears from 10th–17th centuries. This might have been a consequence of the Islamic conquest of Persia, followed by the rise of trade routes to China used by Arabs. An account of an embassy sent by the Yuan to Hulagu Khan in 1259 mention that bin iron was made in India. In 1368 Cao Zhao wrote in a handbook for connoisseurs: “Bin iron is produced by the Western Barbarians. Some have a spiral pattern..others a sesame-seed or snowflake pattern … When a knife or sword is wiped clean and treated with gold thread alum,43 the pattern appears … Its value is greater than silver. Forgeries have a black pattern.”

This may be the earliest description in any language of watered steel.44 In 1637 Song Yingxing compiled a technical compendium Tian gong kai wu, which describes cast and wrought iron: “when they are refined together the product is steel (gang)”. He went on to say 41 Wagner, (2008) 164. 42 Wagner, (2008) 268. 43 This may be identical with zaj, ferric sulphate. In fact any alum will be sufficiently acidic in soluion to act as an etchant. The “black pattern” might be the results of quenching a low-carbon alternative. Tempered martensite will etch dark. 44 Wagner, (2008) 271.

40

chapter three “It is said that among the knives and swords of the Wo barbarians (the Japanese) there are some which are “purified by 100 refinings” and which when placed in the sunlight under the eaves, fill the hall with light. They do not use the method of mixing and refining cast and wrought iron. This is also called steel, but it is an inferior product.”45

In his magisterial books upon the development of ferrous technology in China up to the Han dynasty, Wagner first discusses bronze swords.46 The bronze sword appears to have become a common weapon in the Eastern Zhou period, and in battle scenes of the 5th century bce, most of the warriors wear swords at their belts. These early bronze swords are seldom over 50cm in length, but longer swords up to 100cm in length, appear fairly suddenly in the mid-3rd century bce. One was found in the First Emperor’s mausoleum. Most iron swords were also long, and it has been suggested that the development of the long bronze sword was related to the development of the long iron sword. He goes on to suggest that long bronze swords were difficult to cast, but did not become obsolete until the mastery of quenching steel swords gave the latter a superior performance. It is not impossible that the Qin conquest of all China47 was made possible by superiority in iron weapons production, although relatively few Chinese swords seem to have been analysed. Wagner lists some 35 iron & steel swords from the pre-Han period,48 of which 15 came from a single (early 3rd century bce) mass grave of soldiers. Some metallographic examinations of swords and tools, carried out in China, were also given in translation.49 Out of 18 examples, 3 were made of bloomery iron, 2 were iron apparently obtained via cast iron, 5 were low- or medium-carbon steels, and 8 were steels that had been hardened in some way. Another sword fragment50 from Hebei rather resembles a contemporary (3rd century bce) Celtic sword; the section was piled and the carbon content varies from 0.15% to 0.6%C, the edges show martensite; but no hardness value is given.

45 Wagner, (2008) 342. 46 Wagner, D. Iron and Steel in Ancient China (Leiden, 1993) 191-et seq. see also idem. “Iron & Steel in China” (Cambridge, 2008). 47 Roughly contemporary with the Roman conquest of the Western Mediterranean world. 48 Wagner, (1993)446–7. 49 Wagner, (1993) 461–473. The metallography itself is illustrated on pp 296–334. 50 Wagner, (2008) 132.

Different Ways of Making Steel

41

A recent paper describes the analysis of six Chinese swords, mostly dating from the Han dynasty, from the British Museum.51 Five were from the Han Dynasty and one was described as Medieval (13th–15th century). All were long, and straight-edged, except for one (1050) of the Han swords. Four of the Han swords had similar compositions. They are low in phosphorus and sulphur, and between 0.6 %–0.8% C. They show varying amounts of banding. All have largely ferrite-pearlite microstructures, except for one which shows the presence of a little bainite, suggesting an accelerated cooling. No hardness values were reported. All four contained numerous inclusions, both elongated calcium-aluminium silicates (slag inclusions?) and some spherical silica/iron silicate ones (forge-welding inclusions?) sometimes with phosphorus. It is suggested that bone-ash might have been used as a carburisation accelerator. The fifth Han sword had ferritic edges, and a pearlitic centre, with very few inclusions, but nodules of graphite. This was evidently cast as a sword, and then decarburised. The medieval sword was very different, having ferrite and pearlite (around 0.15%C) in proportions varying much over distance. The higher C% areas had higher P% and it is suggested that this was a pattern-welded sword, despite its late date. Japan Steelmaking resembled European methods rather than Chinese ones. The tatara furnace was a very large bloomery, producing a variety of products, including the high-carbon (up to 1%C) steel known as tamahagane used for making sword blades. Tanimura has discussed the tatara process in some detail.52 Reddish iron sand (akome, mostly ferric oxide, some silica) is charged first. This is reduced easily, absorbs carbon quickly, and tends to form pig iron, which is eventually tapped at the bottom of the furnace, along with the slag. When the furnace reaches a high temperature, black iron sand (masa kogane, magnetite, very low in P) is charged. This is reduced more slowly, and absorbs up to 1 %–1.5%C in the reducing zone, at about 1200ºC. In this zone, some of this steel may absorb more carbon and become pig iron, 51 “Early Chinese ferrous swords from the British Museum collections” Wayman, M.L. & Michaelson, C. 226–232, in “Metals and Mines” ed. S.LaNiece, D.Hook, P.Craddock. (BMArchetype, 2007). 52 Tanimura, H, ‘Development of the Japanese sword’. Journal of Metals (1980) 63–73.

42

chapter three

which would flow out. The steel from the black iron sand (which does not liquefy) accumulates at the bottom of the furnace in lumps (kera). After 3 days, the furnace is broken open, and the kera pulled out. The main part is the hypereutectoid steel, tama hagane, above which are pieces of lowercarbon steel. He also said that wrought iron (hocho tetsu) made from tatara steel, for the core of the sword, was very low in slag inclusions. The specimen shown was made by decarburising tatara, (this might have the advantage of a lower slag content than directly made bloomery iron) but whether this was generally done historically is not clear. The photomicrographs published by various authors suggest that this was not universal. Bain53 pointed out that a good deal of the working was evidently to homogenise the metal used. As well as the use of a high-carbon steel, the skill of some craftsmen included selective quenching so that only the cutting edge became martensitic and around 700 VPH while the core remained pearlitic.54 The yakiba is the hardened zone ; the hamon is the pattern along the boundary of this. Misty patterns called nie (may be seen by the naked eye) and nioi (which needs a glass) are caused by the same thing—a mixture of martensite and fine or even irresolvable pearlite, in varying proportions. It is interesting, but perhaps not surprising, that in time Japan was exporting swords to China.55 After China was reunified by the Song dynasty (960–1279) trade with eastern Asia was actively fostered by the dynasty … (disguised as “tribute” to China and “gifts” in return). ..but judging from the presents sent to China in this era...exports included seaweed, mercury, swords, bows and arrows, lacquerware, fans, screens, gold and silver.”56 Although a great deal has been written on the making of Japanese swords,57 precise metallurgical information remains less than plentiful. 53 Bain, E C, ‘Nippon-too, an introduction to old swords of Japan’. Journal of the Iron and Steel institute (London, 1962), 200, 4, 265–282. 54 Smith, C.S, ‘A metallographic examination of some Japanese sword blades’, in La tecnica di fabbricazione delle lame di acciaio presso gli antichi. “Documenti e Contributi per la storia della Metallurgia—Quaderno I” del Centro per la Storia della Metallurgia A.I.M (Milano 1957), 42–68. 55 “Swords of the Samurai” Harris, V  & Ogasawara, N  (1990) 10. 56 Kinosita, Y. “The Past and Present of Japanese Commerce” (NewYork 1902, reprinted 1968) 36. 57 See for example Kapp, L. Kapp. H. & Yoshihara, Y. “The craft of the Japanese sword” (Tokyo, 1987).

Different Ways of Making Steel

43

A report on the metallurgy of a very early sword has been published by Kitada.58 The straight, single-edged sword was excavated from a Kofun mound (200–600 ce). A piled microstructure of ferrite and pearlite (up to 0.45%C) was found, with no martensite. The maximum hardness was 345 VPH. He also performed tensile tests with these results: “The tensile strength and the elongation are from 450 to 700 MPa (44–69 kg.mm–2) and 15–30%, respectively. The tensile strength and elongation are equivalent to those of modern steel, containing 0.2–0.5 % carbon. The difference of obtained values is due to a variation in microstructures.”

Chikashige59 described the metallography of 11 swords, showing, by means of diagrams, the joining of steel edges to iron cores; steel edges and sides with iron cores, and even all-steel piled blades. Unfortunately, he did not include photomicrographs, nor microhardness results. There are several papers on isolated examples by Bain, O’Neill, Smith and others. O’Neill60 showed the section of a sword that he had had analysed but gave no data for the origin of the sword illustrated. The section shown had steel edge and sides, and had been hardened to 724 VPH (edge), 154–191 VPH (core), and 309 VPH (side). Piaskowski61 described the sectioning of one sword; of unspecified date, but probably later rather than earlier. Metallographic examination revealed relatively high carbon steel of 0.6 to 0.8% carbon on the edge and sides, with a soft low carbon steel core of 0.2% C. There were fewer slag inclusions in the core steel than in the edge steel. The steel layer had been wrapped round an iron, or low-carbon core, and forge-welded to it. Then it had been quenched to give a bainitic microstructure and an edge hardness of 283 VPH (sic). The great pioneer of archaeometallurgy, Cyril Stanley Smith, published metallurgical data on 4 swords,62 and then referred the reader to Tanimura and Tawara. 58 Kitada, M. “Fine structures: mechanical properties and origin of iron of an ancient steel sword excavated from an old mound in Japan” 129–133 in J. Mei and Th. Rehren (eds), Metallurgy and Civilisation: Eurasia and Beyond (London 2009). 59 Chikashige, M, “Alchemy and other chemical achievements of the Ancient Orient” (Tokyo 1936). 60 O’Neill, H. “Metallurgical features in welded steels” Transactions of the Institute of Welding, 9 (1946) 3–9. 61 Piaskowski, J. “Metallographic examination of a Japanese sword” Historical Metallurgy, 27 (1993) 110–117. 62 Smith, C.S, ‘A metallographic examination of some Japanese sword blades’, in La tecnica di fabbricazione delle lame di acciaio presso gli antichi. “Documenti e Contributi per

44

chapter three

Table 1. Data for four Japanese swords examined by Smith (1957).



(century of production)  ca.1940  18th

19th

16–17th

Chemical composition (%) C (edge) C (body) Mn Si P Cu

1.02 1.02 0.37 0.18 0.015 0.21

0.69 0.43 0.005 0.02 0.075 0.01

0.50 0.50 0.005 0.04 0.034 0.01

Microhardness (kg/mm2)

edge body

0.62 0.10 0.01 0.07 0.046 0.01

-----

-----



~ 900*

~ 800

#

~ 800*

~400

~200



~200- 400



* about 5mm from the cutting edge # about 4mm from the cutting edge

Tanimura paid tribute to his teacher, Prof.Tawara, and expressed his own interest in this subject. In 1936 a swordmaking laboratory was set up on the campus of Kyūshu University. Between 1936 and 1938 various research projects were carried out, with the participation of 10 swordsmiths, descendants of swordsmiths, representing different schools, and using metal made in a Tatara. But then he went on to say “… I have not had the heart to cut an old, precious sword ….” So two swords made in this laboratory (Wakizashi A 387mm, and Tanto B 250mm) were tested to destruction. They seem to have been made from fairly homogenous steels; chemical analysis of samples from A showed C% varying from 0.79% to 0.83%C, while B varied from 0.43 % to 0.47%C. The microhardness of A varied from around 630 to 840 within the hardened edge, then it fell abruptly in 12mm to 350–400 VPH in the body of the sword. The microhardness of B varied from around 600 to 720 within the hardened edge, then it fell abruptly in 8mm, to 200–230 VPH in the body of the knife. The most extensive data that has been published on the metallurgy of Japanese swords is that by Tawara.63 He described the smelting, forging, la storia della Metallurgia—Quaderno I” del Centro per la Storia della Metallurgia A.I.M., Milano 1957, 42–68. 63 Tawara, T. nippon too no kagateki kenkyu (“Scientific study of Japanese swords” Tokyo, 1953)

Different Ways of Making Steel

45

hardening and polishing processes in some detail. Most interestingly, he had three swords and other blades sectioned and described their performance in destructive tests. In 1541 three Portuguese traders were driven ashore by a storm at Kagoshima; two years later Fernao Mendez Pinto was also driven ashore by a storm at Tanegashima & introduced firearms to Japan (although the Japanese had presumably been aware of gunpowder since the Mongol invasions of the 13th century; the Marine Archaeology Institute of Nagasaki has recovered grenades from a Mongol shipwreck of 1281 near Hakata.).64 More traders followed soon & they introduced sheep, goats, potatoes, and tobacco to the country.65 Missionaries also arrived—until 1638 when Japan was closed to foreigners & Christianity forbidden. Japanese armour underwent considerable changes after the introduction of firearms, since it had been designed to protect the wearer against arrows, but that will have to be the subject of another book. The manufacture of swords continued to follow traditional methods until the 19th century.

64 Seiho, A. “The origin of firearms and their early transmission” (Tokyo, 1962) and see D ­ elgado, J.P. “Relics of the Kamikaze” Archaeology, 56 (NewYork 2003). 65 Kinoshita, Y. “The Past and Present of Japanese Commerce” (New York 1902, repr 1968) 80.

46

chapter three

Celtic and Roman Swords

PART two THE FIRST EUROPEAN SWORDS

47

48

chapter four

Different Ways of Making Steel

49

Chapter four

Celtic and Roman Swords The first iron swords in Europe were being made by the Celts about the 7th century bce. The Celtic sword has been the subject of a magisterial study by Radomir Pleiner, and this section draws heavily upon his work.1 Iron working was introduced to Europe by Celtic-speaking peoples, and the first iron long (> 70cm) swords were copies of bronze originals. In his book, Pleiner collected reports on 119 swords and summarised the results of the metallographic analyses (and some chemical analyses) on 93 swords, excavated from all over Europe, i.e.the modern countries of Britain, France, Germany, Switzerland, Czechia, Slovakia, Poland, Austria and Hungary. It was possible to examine 59 of these swords in full cross-section. Such an examination, of course, supplies the most information, but it is not always possible with museum exhibits. This group was further divided on the basis of their carbon contents. Steels with less than 0.3%C were classed with iron, and indeed although such alloys are described as “mild steel” today, they should more correctly be called “iron”, rather than given “steel” as a courtesy title, because they behave like iron rather than like steel, and cannot be appreciably hardened by quenching. In this book also, low-carbon steels will be grouped with iron, and only hardenable steels (with 0.4%C or more) will be called steel. Out of these 59 swords, 21 were made of iron. The majority were made of several pieces of metal forge-welded together. In some cases the pieces are merely piled that is welded parallel to the longest axis of the blade. In other case, the pieces are twisted as well as piled, that is, the welds run transversely, inclined to the longest axis. Such swords may well have been the ancestors of the so-called “pattern-welded” blades. The remaining 38 of the group had one (12) or two (26) hardenable steel edges. A few (8) seem to have had all-steel blades, in one case made out of a single piece of steel, while the remainder were made of several pieces of steel piled together. The other 30 had a steel edge, or edges, welded to a piled core, or a layer of steel welded between two layers of 1 Pleiner, R. “The Celtic Sword” (Oxford, 1993) passim. There were of course, earlier iron swords in Greece and the Aegean area.

50

chapter four

iron in a sandwich structure, or even a thin steel layer wrapped around an iron core. Six of the single-edged swords had carburised edges, while the others were (incompetently) piled. Evidently only small pieces of steel were available, and their accurate location in the most useful part of the sword was still uncertain when the smiths were forging piled bundles. For this reason, the examination of samples which do not show a complete cross-section provides us with limited, but still, perhaps, useful, information about the construction of swords. Pleiner summarised the results of 23 such partial examinations. Only 5 of those had steel edges. Many of the others did in fact have steel layers, but not where they were needed most, on the cutting edge. A further 11 were examined only in their cores; 3 of these had steely layers there. Now, although approximately half (46 out of 93, including those with some steel in their cores) of these swords had steel edges, and so were hardenable, only a very small number seem to have been hardened by quenching, namely four to form martensite, and four more to form very fine pearlite. So it is clear that the Celtic smiths did not generally practise quenchhardening. They had attempted to overcome the difficulty of carburising iron uniformly by using only very small pieces of steel, of which several could then be piled together, and forged into a swordblade of fairly heterogeneous composition.2 By achieving a fairly uniform distribution of carbon a steel of moderate hardness was attainable without heat-treatment, the techniques of which were not generally to be mastered for many centuries. The technique of piling numerous small pieces of iron or steel was widely employed. For example, Panseri described a piled Etruscan spearhead, dating back to the 6th century bce which he examined in section;3 the laminated structure is still visible on the surface, even more conspicuous after corrosion. A bundle of thin iron rods from a 4th century bce site in Greece, which was probably the precursor of a piled sword or spearhead is illustrated in Pleiner (1969).4 Piling remained in use for many centuries as a feature of blacksmithsʼ work during the Roman Empire, the 2 Reggieri, A. Garino, C. “Esame tecnologico di un gruppo di spade galliche della lombardia nord-occidentale” Sibrium, 2 (Varese, 1955) 44–55. They discuss some piled Gallic swords from Lombardy. 3 Panseri, C. Garino, C. Leoni, M. in an anthology entitled “Documenti e Contributi per la storia della Metallurgia” (Milan, 1957). 4 Pleiner, R. “Iron Working in Ancient Greece” (Prague, National Technical Museum, 1969) Fig. 6.

Celtic and Roman Swords

51

Migration Period and throughout the Middle Ages in Europe. The later technique of “pattern-welding” (sometimes misleadingly known as “Damascene” work) was to grow out of piling. Roman Swords Despite the extent and duration of the Roman Empire, our knowledge of Roman arms and armour is still surprisingly limited, perhaps because relatively little has survived. The short sword (gladius) may have been better suited to disciplined infantry, or its size may have been governed by metallurgical limitations. It seems to have been largely supplanted by the long sword (spatha) perhaps because of the greater use of cavalry, around the 3rd century ce. A sword from the Roman Republican period (3rd–2nd century bce) in Slovenia was found to have an iron edge and a steel (0.4%C) body, like

Fig. 1. Half-sections (approximately from the centre to the cutting edge) from three Roman swords—adapted from Lang (1988).

52

chapter four

the much later spatha discussed below; a particularly unfortunate combination.5 A recent and extensive survey is that of Lang, who analysed six Roman swords (gladii) from the British Museum and the Chichester Museum.6 Four of them showed signs of piling; one, although steel, showed no signs of piling, and one was carburised. The three earlier ones had been hardened by some form of quenching; their edges were martensitic, although their cores contained fine pearlite. The three slightly later ones showed more obvious piling, and no trace of hardening. The same author has also published analyses of eight Roman (1st century ce) daggers.7 The technology of these differed somewhat from that

Fig. 2. Half-sections from three more Roman swords—adapted from Lang (1988). 5 Kmetič, D. Horvat, J. Vodopivec, F. “Metallographic examinations of the Roman Republican weapons from the hoard from Grad near Šmihel” Arheološki Vestnik, 55 (­Ljubljana, 2004) 291–312. 6 Lang, J. “Study of the metallography of some Roman swords” Britannia 19 (1988) 199–216, & plates V–X. 7 Lang, J. “A metallographic examination of eight Roman daggers from Britain” Sights and Sites of the Iron Age, Raftery, B.ed. (Oxford, 1995) 120–132.

Celtic and Roman Swords

53

Fig. 3. Half-sections from four Roman daggers—adapted from Lang (1995).

of the swords. All seemed to have been made of at least two layers; three were made of low-carbon steels, although one of these had been heattreated in attempt to harden it. The other five had steel edges and had been hardened by heat-treatment. The martensite in the microstructures of each of the hardened daggers seems to have been mixed with fine pearlite, suggesting that some form of quenching less drastic than a full quench had been employed. The author points out that there seems to have been a much higher proportion of daggers than swords made of hardened steel, but the requirements may have not been quite the same. Daggers are not only smaller, and therefore easier to quench, but the failure in battle of a brittle dagger might not have had such immediately drastic consequences as the failure of a sword. It should be borne in mind that if the core of a blade was iron, it would not harden on quenching, but an all-steel sword might become brittle on quenching.

54

chapter four

Fig. 4. Half-sections from four Roman daggers—adapted from Lang (1995).

This author has analysed a sword (a 1st—2nd century ce gladius) in the Rhineland Museum, Bonn, which was a fairly uniform medium-carbon (0.3%–0.7%C) steel, and which had not been quenched.8 Although corrosion had left visible corrugations on the surface, suggesting a piled blade, no visible traces of layers were to be seen in the microstructure. Evidently any forge-welding had been carried out with some degree of skill. A sword from Whittlesea examined by Tylecote9 was of comparable hardness. Two Roman swords in the Warsaw Archaeological Museum also showed piled steel structures but they were not quenched.10 The same 8 Williams, A.R. “Roman arms and armour: a technical note” Journal of Archaeological Science, 4 (1977) 77–87. 9 Tylecote, R.F., Gilmour, B. “The metallurgy of early ferrous edged tools and weapons” British Archaeological Reports, 155 (Oxford, 1986) 164–7. 10 Biborski, M. Kaczanowski, P. Kedzierski, Z. Stepinski, J. “Metallographic analysis of two Roman swords from the State Archaeological Museum, Warsaw” Wiadomosci Archeo-

Celtic and Roman Swords

55

team of researchers has analysed three Roman swords from Vindonissa (Vindisch) and Augst (near Liestal) on the Rhine limes now in Switzerland.11 One spatha from Augst (1st–2nd century ce,) was made from a layer of 0.6%C steel fixed between two layers of iron by forge-welding to give steel edges, and then hardened to 330 VPH.

   Fig. 5. The mode of assembly   of the parts of the blade.

Fig. 6. Section of the edge of a sword from Augst, inv.no.1961.4401. F = ferrite P = pearlite M = martensite nP = nodular pearlite (almost irresolvable); adapted from Biborski, Stepinski et al.

Taking the opposite route, a gladius from Vindonissa (2nd—3rd century ce) was made from a layer of 0.6%C steel wrapped round an iron core by forge-welding to give a steel case of 216 VPH.

logiczne (1982) 47, 15–24. In the same issue of this journal (pp 3–13) there is a note by J.Kolendo on the inscriptions found stamped on these blades. One was… VM.. RO and the other was … ACT. 11 idem. “Ergebnisse der metallographischen Untersuchungen von römischen Schwerten aus dem Vindonissa-Museum Brugg und dem Römermuseum Augst” Jahrbuch der Gesellschaft pro Vindonissa, (1985) 45–80. The drawings of the swords have been transposed by the printer (pers.comm. J. Stepinski); the spatha is from Augst, not Vindonissa.

56

chapter four

  Fig. 7. The mode of assembly  Fig. 8. Section of the edge of a sword from Vindonissa, of the parts of the blade.  inv.nr.1938.  G = granular carbides (presumably reheated pearlite or  overtempered martensite); adapted from Biborski,  Stepinski et al.

Fig. 9. A photomacrograph of the edge.

Celtic and Roman Swords

57

Good steel was available (at a price, no doubt) from Noricum, the province approximately corresponding to modern Styria. The Erzberg is a mountain of siderite iron ore, worked for thousands of years, which was to supply the steel for the Innsbruck armoury in the 15th century. A typical example of a Roman knife was published by Pressingler.12 It had a cutting edge of high-carbon steel, corresponding to perhaps a quarter of the total mass of the blade, welded to an iron back. The whole knife was then given some sort of heat-treatment to produce an acicular microstructure in the edge, which was presumably hardened, although no hardness values were quoted. Elemental analyses were, however, given. Part of knife

C

Mn

Si

P

S

As %

Edge Back

1.30 0.02

0.13 0.08

0.05 0.10

0.013 0.013

0.004 0.008

0.001 0.031

There is a Spatha probably of the 5th century ce found at Steinbrunn in Burgenland (Austria) in 1965. (Hofjagd- und Rustkammer, Vienna A.2348a). It was not possible to section this sword, but samples were taken from the edge and from the centre. Surprisingly, the carbon content was higher in the centre, suggesting that this was made by forging several pieces of metal together, with the intention of attaching steel edges to an iron core, but the smith became confused and got the better steel in the centre rather than at the edges. Afterwards it was hardened by some sort of heat-treatment, so that the lower-carbon edge has formed a mixture of ferrite and martensite, of hardness up to 230 VPH; but the higher-carbon centre has been cooled somewhat more slowly and transformed to very fine (largely irresolvable) pearlite, of hardness up to 330 VPH. So the edge of this spatha is less hard than the centre, which is not desirable ! and it would have been a much less effective sword than the spatha from Augst of three centuries previous.

12 Presslinger, H, Maier, C. & Lorenz, T. “Metallographische untersuchungen an einen rmerzeitlichen Messer aus Norischem Stahl” BHM 5 (1991) 184–188.

58

chapter four

Fig. 10. Sample from an edge (scale bar = 50 microns).

Fig. 11. Sample from an edge at higher magnification; the microstructure consists of equiaxed ferrite grains and (dark) areas of martensite. (scale bar = 10 microns) Microhardness = 199–232 VPH.

Summary Out of these Roman swords so far analysed, although numerically not a very large data base, it is significant that around half were not hardened, although made of hardenable steels. It seems likely that the uncertainties

Celtic and Roman Swords

59

entailed in quenching an unknown metal led to a preference for a reliable sword of moderate sharpness over an unreliable sword of greater sharpness. These do show some technological progress over the Celtic swords, of which around half had steel edges (46 out of 93) but less than 10% were hardened (8 out of 93). A Victorian Piled “Sword”

Fig. 12. This is the polished surface of a 19th century knife from the author’s kitchen. It was made from “shear steel” that is a bar of iron which was surrounded by carbon-containing material and heated for several days. The heterogenous steel bar formed was then folded and forged out several times in a attempt to equalise the carbon distribution. The result, however, was a layered steel not very much different from the best Celtic swords. The heat-treatment was slightly more successful, however, and its hardness is around 275- 510 VPH.

Mail Armour Not only did the Celts make their swords by piling together numerous small fragments of bloomery iron, but their distinctive form of body armour was made flexible by linking together innumerable tiny iron rings. Varro describes mail as a Celtic invention,13 and the oldest piece of interlinked mail yet found was excavated from a 3rd century bce Celtic grave in Romania. Rusu illustrated what appears to be both riveted and 13 Varro, “De lingua latina”, V, 116. (Loeb ed. R.G.Kent, 1938). 1st century bce, in the course of an explanation of the meanings of words: Lorica quod e loris de corio crudo pectoralia faciebat; postea subcidit gallica e ferro sub id vocabulum, ex anulis ferrea tunica. The “lorica” [is so called because] they used to make chest armour from thongs of rawhide, and later the Gallic one of iron, an iron tunic of little rings, was included under this name.

60

chapter four

welded links of circular cross-section. Most of the links were of wire between 0.8 and 1.8mm thick and were between 8.5 and 9.2mm in diameter. There also were rows of butted links, which might have been a repair. There was also excavated an iron helmet surmounted by a bronze bird, whose wings would have flapped as the owner charged into battle.14 Most Roman legionaries were protected by mail shirts by the 1st century bce, using a pair of heavy javelins and a short sword.15 From Central Europe Celtic-speaking and iron-using groups travelled westwards into Germany, thence into Gaul and Italy, one group sacking Rome in 390 bce. Many of the Romans’ military successes during the 2nd and 1st centuries bce were gained in fighting more of these Celtic warriors in northern Italy, Gaul and Spain, and finally Britain in the 1st century ce. In his account of the battle of Telamon near the coast of Tuscany in 225 bce, Polybius describes how the Celtic swords bent in combat, and how the Celtic warriors had to straighten them underfoot before continuing to fight, only to be eventually defeated by the Romans.16 The Romans prevailed through better discipline and defence17 although the analyses given above suggest that their swords were little better than those of the Celts. Throughout the Migration Period and into the early Middle Ages, the mail-shirt (byrnie, hauberk) remained the principal body defence for those warriors fortunate enough to be able to afford it. The Ripuarian laws of the Franks fixed the price of a mail brunia at 12 solidi, i.e.6 oxen or 12 cows.18 A sword and sheath were to be priced at 7 solidi, a helmet at 6 solidi, and a spear and shield at 2 solidi. Its adaptability, however, meant that it was frequently repaired and reused, so that very little has survived intact from this period, if indeed, much was ever made ab initio in this period. The Vikings continued to employ mail, both alone and with modification. A series of boat-burials at Vendel in Sweden, Valsgarde, and else 14 Rusu, M. “Das keltische furstengrab von Ciumesti in Rumanien” Germania, 50 (­Berlin, 1969) 267–278 and plates 140–149. 15 Robinson, H.R. “The Armour of Imperial Rome” (1980) passim. 16 Pleiner, op. cit. 163–4. tested replica iron swords and found that they did bend, but by less than 1 cm. Polybius had evidently exaggerated. 17 Camillus (dictator 363 bce) had “helmets constructed entirely of iron, (as protection against the Celtic swords) and shields sheathed with copper … for the iron of the Celts was badly wrought, and their swords buckled.” Pleiner, op. cit. 26. 18 Beck, L. “Die geschichte des eisens in technischer und kulturgeschichtlicher be­­ ziehung”, (Braunschweig: 1884–1903) I, 725.

Celtic and Roman Swords

61

Fig. 13. The statue of a Celtic warrior from Glauberg, Hesse. Possibly 5th century bce. He wears an articulated armour of some sort (a mail shirt?), carries a long shield and sword and his headdress is apparently of a ceremonial form.19

where from the 7th to the 9th century have yielded samples of mail with rectangular reinforcing pieces of plate.20 This innovation might be ascribed to an eastern influence. Certainly, Vikings employed the rivers of Russia to trade to the Black and Caspian Seas as well as setting up the Kingdom of Kiev (by tradition founded in 864 ce), from which nucleus Medieval Russia was to grow. These eastern influences may well have been a factor in Viking swordmaking, as a later chapter will try to illustrate.

19 Frey, O.H. Herrmann, F.R. “Ein frühkeltischer Fürstengrabhügel am Glauberg im Wetteraukreis, Hessen. Bericht über die Forschungen 1994–1996.” Germania 75 (Berlin, 1997), 459–550 20 Arwidsson, G. “Valsgarde 8: Die grabefunde von Valsgarde” (Uppsala, 1954)

62

chapter five

Chapter FIVE

Pattern-Welding The earliest examples of pattern-welding appear in swords found in ritual deposits (presumably as booty) during the 3rd century in the area between Germany and Scandinavia, and some of which bear Latin inscriptions. Pattern-welded swords have been found in many localities outside the frontiers of the Roman Empire.1 Whichever side of the frontier they were made, their techniques (single-chevron and double-chevron) may be seen as an elaboration of Celtic piling. It may be that those people who wanted to make long swords, but whose furnaces could produce only small pieces of iron, were forced to devise techniques for forging numerous small pieces of iron into one large billet. This was probably the origin of “piling”, employed by the Celts, and its decorative offspring, “pattern-welding” which formed a desirable serpentine pattern on the surface of the polished and etched blade. By the Vendel period (550–800), pattern-welded swords were common in Scandinavia, but by the Viking age, they had given way to plainer swords made of only a few pieces (or even one piece) of steel. Although very few Celtic swords were hardened by heat-treatment, and none from Merovingian France, the proportion so improved starts to increase in later centuries.2 Although many swords without pattern-welding were being made in Scandinavia and the Baltic region, it seems to have continued in use as a decorative element, used in side-panels welded on to the blade,3 or even as the letters forming the supposed maker’s name inlaid into an all-steel blade.4 So pattern-welding continued to be employed, especially in Central and Eastern Europe, even as late as the 13th century.5 The earliest finds of such “pattern-welded” swords have been from Nydam. 1 Biborski, M. Kaczanowski, P. Kedzierski, Z. & Stepinski, J. “Badania nad technologia mieczy z mlodszego okresu przedrzymskiego z obszaru kultury przeworskiej” Varia Barbarica, (2002) 81–104. includes a bibliography of other studies on swords from this culture. 2 See table on p 233. 3 Anteins, A.K. “Structure and manufacturing techniques of pattern-welded objects found in the Baltic States” Journal of the Iron & Steel Institute, (June 1968) 563–571. 4 See Chapter 8—on VLFBERHT swords. 5 Gurin, M. “Kuznechnoi Remeslo Polotskoy Zemly 9–13c.” (Blacksmiths’ crafts in the Polotsk lands) (Minsk, 1987) includes 32 pp of plates.

Pattern-Welding

63

Nydam (Period of the Late Roman Empire) During 1859–63, a series of excavations in the south of Jutland (Slesvig, now Schleswig) unearthed a large deposit of Roman-era swords and armour. An account in English of the finds from the peat bogs (or mosses) of Thorsbjerg and Nydam was published by the Danish archaeologist responsible.6 Nydam yielded three boats, more than 100 swords, of which 90 were “damascened”, as the author called them. He called them “Damascene blades” although they have nothing whatever in common with Damascus steel. “Damascening” is the name given to the decorative technique of inlaying one metal with another. The later name “pattern-welded” is more appropriate. They were made up of numerous strips of metal forge-welded together in such a way as to form a visible pattern on their surfaces. They were all two-edged swords, 75–100cm in length, and had been unusually well preserved by the tannic acid in the moss water. Names stamped in Roman letters on their tangs include RICVS, RICCIM, COCILLVS, and VMORCD. This might have reflected a fashion for Romanising Teutonic or Celtic names, rather than necessitating Roman manufacture of the blades. A considerable quantity of Roman coins, mostly denarii from Vitellius to Commodus and Macrinus was located in the boats. The latest coin was from the year 217 ce, although dendrochronology has now ascribed a date of 320 ce to the timbers of the boat.7 No human remains were found in Thorsbjerg, but the remains of mail shirts instead. Five or six very corroded hauberks (mail shirts) were found rolled up, and in two cases they had been placed inside clay vessels. Two of the larger fragments were made of round section wire, and rows of riveted & welded links alternate, one of which was made of 10mm, the other of 9mm diameter rings. In one instance, the rivets were of bronze. In addition, strips of mail made up of bronze rings were also found. These were presumably borders from the hems of mail shirts. At Vimose, a number of spear and axe-heads were found together with some 17 single-edged (seax) and 18 double-edged swords (spatha). Many of these were pattern-welded like those from Nydam, and some had surviving hilts. Also some of them had inlaid names like TASVIT and ICAXI.8 The name RICCVS or SICCVS turned up later on another pattern-welded 6 Engelhardt, C. “Denmark in the early Iron Age” (London, 1866) passim. 7 Guratzsch, H.ed. “Landesmuseen Schloss Gottorf” (Munich, 2002) 143. 8 Engelhardt, C. “Fynske Mosefund, II, Vimose Fundet” (Copenhagen. 1869) passim.

64

chapter five

blade from a 3rd century grave at Pasewalk in eastern Germany. The overall carbon content was found to be < 0.4%C, but metallography was not possible because of its corroded state.9 There was one fragment of mail which was made of links apparently flat in section and alternately riveted and welded. Another, very damaged, was round in section, and consisted of two parts attached to both sides of a bronze hinge (75mm high), presumably part of a body armour. These had been deposits into a “sacred lake” which gradually filled up with vegetation, and hence a layer of peat formed, and the lake became a moss. In 1864, the Prussian army invaded Denmark and annexed SchleswigHolstein. The excavations were suspended for many years, and the finds were transferred to Prussian stewardship. The whole Nydam treasure with many other iron arms and appliances was transferred to a museum in Kiel. It is now in Schloss Gottorf (Schleswig), while the Vimose finds are in the National Museum, Copenhagen. Excavations have recommenced in recent years. The Examination of “Nydam Swords” A colonel of artillery in the Danish army, O. Blom, found by some rather drastic experiments that some of these swords had been made of steel. In 1868 he published the results of his examinations of fragments of weapons from the Danish Iron Age.10 He determined their carbon content by heating them in turn to redness, and then quenching them in water. The results of five such experiments were given. A fragment of a sword blade from the peat-bog of Vimose “became hard to the file, and showed a close-grained and uniform fracture”. Blom concluded that this was made of an “excellent steel” and went on to say that “he had had the metal forged into a little knife for himself, which performed very well” [sic]. Another fragment of a sword blade from Dallerup (Jutland) was also of a similar steel. A spear-head also from Vimose became harder after quenching, but showed a granular fracture; Blom concluded that this was a “fairly good steel”. A javelin-head from the peat-bog of Kragehul (Fyen) 9 Schultze, E. “Kaiser- und völkerwanderungszeitliche Baggerfund aus der Ücke bei Pasewalk” Bodendenkmalpflege in Mecklenburg-Vorpommern (Lübstorf, 1993) 191–212 + Beilage. 10 Blom, O. “Om Materialt I den Ældre Jernalders Vaaben” Aarbøger for Nordisk Oldkyndighed og Historie, (Copenhagen, 1868) 1–13.

Pattern-Welding

65

was a “fairly soft iron”, and a fragment from a “damascene” (he meant “pattern-welded”) sword blade from the peat-bog of Nydam was only a “soft iron”. Beck11 observed that the Nydam blades were produced through joining together alternate layers of steel and iron. He realized that the appearance of the “pattern-welded” blades was caused by their method of manufacture. If pieces of steel and iron were to be welded together in layers one on top of the other, then a parallel-striped pattern would be obtained. If such a bar were to be drawn out along its long axis, twisted, and forged out and then welded together with a similar (but oppositely twisted) bar, then one would obtain a “chevron damask” with a central line of symmetry. Repeated manipulations might produce a “flowery damask”, and the most elaborate patterns might be produced according to the variety of the forgings. He tried to model the production of these patterns in wax, and observed that ” the flowery damask is absolutely not formed as Mr. Engelhardt suggests, by [chiseling] the soft iron with a gouge, but the interlacing of steel and iron is extended throughout the whole mass. I cut fragments from these swords with a chisel and I was convinced that this interlacing of iron goes all the way through the whole mass. I had the steel examined chemically and determined that it contains 0.6% of carbon and was to be regarded as a good “tool steel”; I had the steel parts quenched and found that it could be hardened and its steely nature shows still today as it did 1600 years ago.” Although the Nydam finds bore Latin inscriptions, and Beck described them as “Roman damask” their origin may have been a Teutonic development of Celtic techniques. In many cases, the pattern-welding seems to have been decorative rather than functional, as it did not extend through the section of the blade. Beck believed, and some subsequent writers have accepted this argument, that this combination of iron and steel produced a material suitable for a good sword, because it would have combined hardness with toughness. This may be true for modern metals, however while bloomery iron may be soft, this does not mean that it is tough. Its high content of slag inclusions makes it a brittle material rather than a tough one.12 So the incorporation of iron might have been for reasons of economy, or even for decorative effect, but it was not to improve mechanical properties. 11 Beck, L. “Die geschichte des eisens in technischer und kulturgeschichtlicher beziehung”, (Braunschweig: 1884–1903) 5 vols, especially I, 556–563. 12 For fracture testing see Williams (2003) 932.

66

chapter five

Metallographic studies were undertaken in the 20th century on some of these swords by Neumann, and later, by Schürmann, Høeg and Thomsen. In 1927, Neumann13 examined some small pieces of swords from the Nydam finds metallographically and described the type and position of the damask structure. Unfortunately the sword pieces examined by him were very largely overlaid with rust so that only a restricted metallographic examination was possible. Neumann could however distinguish, by observation of the different swords’ surfaces, three types of what he called “welded-damask”. These were streifendamast that is, a piled structure; winkeldamast that is chevron-pattern-welding, and doppelwinkeldamast, that is, a doubled-chevron pattern-welding.

Fig. 1. A piled swordblade; a single-chevron-pattern blade: a fragment from a doublechevron-pattern blade. (after Neumann).

He succeeded in showing that on one of the sword fragments examined, damask strips were welded onto both sides of a harder steel billet and likewise that the edges were made out of a harder steel, and had been welded onto the two narrower sides of the steel billet. The presence of a “martensitic-troostitic” microstructure led him to pronounce that the quench-hardening of steel was known by the manufacturers of the swords. In fact this seems likely to have been some form of “slack-quenching.” Microstructures of two areas were shown. The “damask” pattern appeared to be made of higher C% and lower C% steels welded together, and contained varying amounts of ferrite and pearlite. The edges contained what might be pearlite and an acicular material which he reported as martensite, but no hardness measurements were given. The dark areas were described as “troostite” mixed with “martensite needles”. He used the then current term troostite for irresolvable pearlite, which is fre-

13 Neumann, B. “Römischer damaststahl” Archiv für das Eisenhüttenwesen, 1 (1927) 241–244, pl. 14–5.

Pattern-Welding

67

quently the product of some heat-treatment less drastic than a full quench (a “slack-quench”), but the nature of which cannot now be determined. The term “irresolvable pearlite”, “nodular pearlite”, or even “an ­irresolvable ferrite-carbide aggregate” would now be employed instead of “troostite”. While Beck had earlier found one of the Nydam-swords to contain 0.6% C Neumann was able to publish a fuller analysis of one:  C 0.62%, Si 0.15% Mn 0.363% P 0.054% S 0.073% However, the carbon content was quite unevenly distributed over the cross-section of the blade. In 1959 Schürmann14 was able to examine three further specimens, in better condition, so that the construction of the swords, as well as their hardness, could be more precisely determined. These showed examples of the three sword “types”, that Neumann had identified from their surface appearances. Analysis of parts of three Nydam swords Sword

C%

Si%

Mn%

P%

S%

N%

Sword A Middle portion hard edge soft edge

0,52 0,43 0,29

0,061 n. d. n. d.

0,03 n. d. n. d.

0,141 0,117 0,161

0,013 n. d. n. d.

0,018 n. d. n. d.

Sword B

0,423

0,003

0,016

0,211

0,020

n. d.

Sword C

0,52

0,107

0,016

0,146

0,011

0,007

Sword A: This was welded together from eleven layers. There was a middle strip of steel with a panel of banded metal attached to one side of it. There may have been a second panel on the opposite side, which has since been lost. There was an edge of an apparently single piece (but perhaps forge-welded from 2 or 3 pieces) of steel, and the opposite edge appeared to be made of two pieces of metal, one iron and one steel. The banded panel was made up of high-P% and low-P% irons, to form the pattern. The higher C% areas had formed a mixture of martensite and “troostite” (nodular pearlite) on quenching, with martensite predominating near the edges. 14 Schürmann, E. “Untersuchungen an Nydam-Schwerten” Archiv für das Eisenhüttenwesen, 30 (1959) 121–126 and also see 127–130.

68

chapter five

The hardness of the banded panel varied between 170–205 VPH; the hardness of one edge was 191–536 (reached at the edge) VPH, and the other was 250–650 (edge)VPH. Sword B: This was rather fragmentary, but in the middle there was a banded structure of high-P% and low-P% with ferrite and pearlite in varying proportions, reaching 189 VPH at one side and 235 VPH at the other. There might have been steel edges attached to this banded centre. But however, in contrast to the other swords A and C, it did not show any signs of hardening despite the high carbon content. Sword C: No banded panels were present; this is made up of two parallel layers, one of ferrite, and one of martensite and pearlite. It was probably made from one high C% layer which would form a hardenable middle band and edges, with two low-C% layers (which when quenched, would have remained soft) welded upon either side of it, but one of them has been lost. It has been quench-hardened, and the higherC% layer was 383–575 VPH (at edge). Schürmann also discussed the possible hardening of these blades in a separate article in the same journal.15 After heat-treatment and hardness tests on a blade fragment of the Nydam sword B, he pointed out that the presence of phosphorus tends to discourage the absorption of carbon, and therefore affects the hardenability of steel. For those interested, there is a more recent, comprehensive, paper on the effects of phosphorus in steel.16 Five swords from Vimose were analysed by E. Høeg for a paper by Klindt-Jensen.17 Two out of these five were hardened, to form “troostite” amongst other microconstituents. (1) spatha 22920. The centres of the edges were made of a 0.6%C steel with a fine pearlite/ferrite microstructure. The whole section of the body was made of alternating layers of 0.2%–0.5%C steels. (2) spatha C3799. A sword described as “facetted”, i.e. with a midrib, rather than a fuller, and described as all-steel, although the section published

15 Schürmann, E. & Schroer, H. “Härte- und Glühversuche an dem Klingenbruchstück eines Nydam-Schwertes” Archiv für das Eisenhüttenwesen, 30 (1959) 127–130. 16 Stewart, J.W, Charles, J.A, & Wallach, E.R. “Iron-phosphorus-carbon system” Materials Science and Technology, 16 (2000) 275–303. 17 Klindt-Jensen, O. “Keltisk tradition I Romersk Jernalder” Aarboger for Nordisk Oldkyndighed og Historie, (Copenhagen, 1952) 195–228.

Pattern-Welding

69

suggests it may have been made up of three pieces of steel, of around 0.7%C. The body has a fine pearlite/ferrite microstructure. The edges showed some martensite, “troostite” and pearlite, so, some attempt had been made to harden it by quenching, although hardness values were not quoted. (3) C 1554. A low-carbon (0.2% C) steel with a ferrite/ pearlite microstructure. (4) A broken blade with no inventory number. The centre is a piece of low-carbon (0.1%C) steel with some phosphorus banding. On either side of this body two panels made up of layers of 0.2%–0.4%C steel were welded. The edge was around 0.8% C with a microstructure of martensite, “troostite” and pearlite, so an attempt had been made to harden it by quenching, although hardness values were again not quoted. (5) A broken blade with no inventory number. The body was made of bands of carbon-free phosphoric iron, and the core of 0.9%C steel with a pearlitic microstructure, with sides of 0.2%C steel.

Fig. 2. Section of spatha 22920; section of spatha 3799; section of broken blade (4) adapted from Klindt-Jensen. The darker bands are steel.

Subsequently, Robert Thomsen investigated six swords from Illerup,

70

chapter five

together with a further ten sword fragments from Nydam, and presented some of the results in 198718 and then with a much fuller account in 1992. The swords from Illerup were investigated by metallography only, although three spearheads of modest quality (iron with up to 0.6% P) were also included. The fragments from Nydam were investigated by metallography as well as some combustion analyses of 1g samples to determine average carbon contents. He divided the 16 blades into three classes based on their method of manufacture: Group 1: those made up of parallel layers (four), without pattern-welding on their surfaces. Group 2: those swords with pattern-welded (two), or at least piled (five), mid-sections, and with edges welded on. Group 3: those with pattern-welded panels (five) only welded onto the sides of the mid-section. (Examples of all three classes were found in both locations.) Out of these 16 sword-fragments, 12 had steel edges, although only one seemed to have been hardened. No hardness values were reported however, so this remained uncertain. Subsequently, 8 of these swords from Nydam were re-examined by Buchwald19 and he found that two of them had been edge-hardened to martensite. How this had been done is a matter for debate. Buchwald speculated that the hot blade had been drawn through wet clay. Another possibility might be that a bundle of blades was wired together (face-to-face) to prevent warping and quenched together; then only the edges would have been exposed to the cold water. At present, this must remain an open question.

18 Thomsen, R. “Pattern-welded swords from Illerup and Nydam” 371–377 and pl.xixxxii.in “Archaeometallurgy of iron “ed. Pleiner, R. (Prague, 1989). Idem. “Metallografiske undersoelgelser af sværd og spydspidser fra mosefundene I Illerup og Nydam” Aarboger for Nordisk Oldkyndighed og Historie (Copenhagen, 1992) 281–310. 19 Buchwald, V.F. “Iron and steel in ancient times” (Copenhagen, 2005) especially Chapter 11.

Pattern-Welding

1 cm Fig. 3. Sections of some blades from Nydam. Scale bar = 1cm.

1 cm Fig. 4. Sections of some blades from Nydam. White areas = iron. Lines = steel.  Dots = iron-phosphorus alloys. Adapted from Buchwald (2005).

71

72

chapter five

The use of pattern-welded panels welded to the side (in Group 3) confirms that, even at this early date, they had frequently come to have a decorative, rather than a functional, role. Such a decorative role continued to be found for the pattern-welding inlay which made up the supposed maker’s name in some “Ulfberht” swords six or seven centuries later, and even an entire panel on an Ulfberht” sword from Olomouc (see Chapter 8). Another example, contemporary with the Nydam finds is that during the 2nd century ce people of the Przeworsk cultural group migrated southward through the Carpathians (and towards the Danube frontier of the Roman Empire) into what is now eastern Slovakia. Numerous patternwelded swords have been excavated from their cemeteries, and one example studied recently had a pattern-welded core (ferrite/pearlite microstructure and high-phosphorus iron) to which steel edges had been welded. There was some surface decarburisation reported, apparently the results of burning during cremation.20 The edges do not seem to have ever been quenched as they had a pearlitic microstructure (200 VPH), and not the globular one which might have resulted from an overtempered marten­site. Migration Period Pattern-welding A Migration-Period sword excavated from Bešeňov (SW Slovakia) which had its hilt decorated with silver and niello, is an example of the decorative employment of pattern-welding.21 One surviving panel of patternwelded material was welded to the side of an all-steel blade, which was itself made from at least three pieces of steel. The core had a ferrite/pearlite microstructure (200–300 VPH), the decorative pattern was made up of a ferrite/pearlite band and a ferrite (high-phosphorus iron) one. Many of these features are also to be found in the swords from Nydam. One edge had a ferrite/pearlite microstructure (0.1%C), while the other edge, which had more carbon, had been quenched to give a mixture of “troostite” (very fine pearlite of 400–500 VPH) and martensite (800 VPH). It is an open question whether the maker originally hoped to have two sharp edges, and was let down by the steel he used. 20 Mihok, L. “Metallographic examination of pattern-welded swords from the Early Roman period in Eastern Slovakia” Archaeomaterials, 7 (Philadelphia, 1993) 41–51. 21 Pleiner, R. “Metallographische Untersuchung des Schwertes von Bešeňov” Studijne Zvesti Archeologickeho Ustavu Sav, 35 (2002) 77–82.

Pattern-Welding

73

Two swords from a 6th century Lombard cemetery in Hungary have been published by LaSalvia.22 One had a pattern-welded (high- and lowP% bands) core and a low-carbon steel edge, with a ferrite/pearlite microstructure. The other had what seems to have been a piled core, and a steel edge with a fine-pearlite microstructure, which might have been the results of a heat-treatment, but no hardness values were quoted. The opportunity was taken to compare these with the results obtained earlier from some swords excavated from a later, although undated, Lombard cemetery in Benevento (South Italy) upon which some metallography had been carried out by Rotili.23 One sword had a piled structure, with a largely ferrite/pearlite microstructure. The carbon content (corrected from atomic percent) was 0.23% C. Some martensite was identified, mixed with pearlite near the tip, and the maximum hardness determined, avoiding corrosion products, was 410 VPH. The other sword was piled from several pieces of metal, whose carbon contents varied from 0.29% to 0.40% C. The core and the edges have a microstructure containing martensite as well as fine pearlite, and ferrite, in varying proportions. The hardness of the edges reached 516 VPH. As the wide extent of this technique was gradually appreciated, more papers on pattern-welded swords appeared, such as those by FranceLanord. More pattern-welded swords have been identified as such, especially by the employment of X-radiography on severely corroded blades.24 France-Lanord pointed out that many swords excavated in western Europe were made by this method.25 They are found in some numbers in France (from the 6th century) and also in England, but most plentifully in Alemannic cemeteries. They are also common in Scandinavia from the 8th century. From about the 11th century, however, they seem to be replaced by “all-steel” swords. France-Lanord had a number of swords in the museums of Nancy, Virton, Arlon, and elsewhere, cleaned of corrosion, and polished and etched to reveal pattern-welded structures hitherto unseen.

22 LaSalvia, V. “Archaeometallurgy of Lombard swords” (Florence, 1998); 23 Rotili, M. “La necropolis longobarda di Benevento” (Naples, 1977) 24 Lang, J. & Ager, B. “Swords of the Anglo-Saxon and Viking periods in the British Museum; a radiographic study” 85–122 (more than half were pattern-welded) chapter 7 in “Weapons and warfare in Anglo-Saxon England” ed. Sonia Chadwick Hawkes, Oxford University Committee for Archaeology Monograph No. 21, 1989. 25 France-Lanord, A. “La fabrication des épées damassées aux époques mérovingienne et carolingienne” Pays Gaumais , 10 (Nancy, 1949) 19–45.

74

chapter five

Fig. 5. Surfaces after polishing and re-etching.

Their surfaces were examined by both photomacrography (at low-magnification), and photomicrography (at high-magnification). He also had two swords from the Lorraine Historical Museum sectioned both longitudinally and transversely and showed that the pattern itself was made up of low- and high-carbon bands, reaching up to around 0.4%C. So he refuted the possibility that the pattern was the product of wires attached to the surface. He examined some forty Merovingian period (5th–8th century) swords, of which only seven were not pattern-welded. None of these swords seem to have been hardened by quenching. He suggested that there was in improvement in the mechanical properties of such blades, and reported that there was an increase in bend test strength of 2.5 to 3 times over “ordinary swords”, but did not define what he meant by these. It is quite possible that any improvement in toughness may simply have been the result of intensive forging reducing the average size (rather than the total volume) of slag inclusions.26 The possible range of techniques used by medieval smiths is described at length by Pleiner.27 Later Pattern-welding A large number of pattern-welded swords from Sweden have been studied over the years by Lena Thålin-Bergman and the results were assembled for publication by Arrhenius.28 26 Op. cit. 37; and see France-Lanord, A. (1943) 27 Pleiner, R. “Early European Blacksmiths” (Prague, 2006) passim. 28 Bergman L. T & Arrhenius B. “Excavations at Helgö XV (weapon investigations— Helgö & the Swedish hinterland)” Stockholm 2005.

Pattern-Welding

75

Over five hundred weapons were X-radiographed; pattern-welding was commoner than previously thought—but corrosion made it difficult to detect without repolishing the blades. It was found on some two-thirds of the swords examined, although only five were then examined by metallography.. Around 10% of Gotland swords had pattern-welding of a type found in Roman times and 50% had pattern-welding of the more elaborate types commonly found from around 400–900 ce. In the boat grave cemetery from Vendel the swords in early graves (550–650 ce) had pattern-welding of similar types to that commonly found in Anglo-Saxon & Frankish areas. From the 9th century onwards spear heads with pattern-welding were found, but largely in the form of decorative strips. About half of those examined had been hardened. Seaxes with pattern-welding had been known in early Westphalia and possibly influenced Scandinavia, which after 650 followed its own Nordic typology -in Sweden & Norway they showed pattern-welding but not in Denmark.29 Some representative examples may be quoted. Merovingian period sword SHM 7480.40 It had pattern-welded strips welded onto a narrow core; one strip has 0.5%C and another has 0.9%C. The core has 0.1% C and edges have 0.5%C. The edge had not been hardened, and measured only 185 VPH. Sword SHM 31202.5  6th–8th century. This had the edges welded to a core, with two strips of pattern-welding also welded onto the core. The edges were steel of 0.5%C ; it had not been hardened, and measured only 162 VPH. The pattern-welding contained strips with 0.5 % P, while the core and edges contained only 0.1 % P. Sword fragment from Helgö. H 2814; 8th-10th century. The core was ferritic with pattern-welded strips [of /\/ form] welded on; both core and edges were low in carbon (0.1%C) but contained different levels of phosphorus P% (up to 0.4% in one layer) which provided the contrast in appearance after polishing and etching. Later knives from Helgö show piled structures without pattern-welding.30 29 Bergman, op. cit. 10. 30 Tomtlund, J.E. “Metallographic examination of 13 knives from Helg” Early Medieval Studies, 5 (Lund, 1973) 42–63.

76

chapter five Replication

Anstee and Biek carried out a series of experiments in which they attempted to make a pattern-welded sword, and eventually succeeded in replicating the microstructure of the Palace of Westminster sword.31 This is a 9th century sword found in the Thames, which was made up of layers of carbon-free iron and low-carbon (only 0.2%C) steel. The maximum hardness achieved was only a modest 188 VPH. They took strips of different metals and twisted and forged them together in varied combinations. The pattern did not seem to depend upon large differences in carbon content; other elements, such as phosphorus, could play a role, and even lines of slag at the interfaces between the layers of iron might be enough to form a pattern. A considerable amount of replication was carried out by Denig and included in Robert Thomsen’s paper (op. cit. 1992). Some more metallographic studies: British Museum 81–59–18. A sword excavated at Canwick Common, Lincoln. Possibly 9th–10th century. This had a piled microstructure, apparently made by forge-welding several pieces of steel together. The blade was not fully homogenised, and layers of lower C% were observed. It was hardened by some form of heattreatment; probably a slack-quench. The lower C% areas formed a mixture of ferrite and martensite, with an irresolvable material. The higher C% areas formed a mixture of martensite with this irresolvable material. This may have been nodular pearlite, which is frequently to be found in slack-quenched blades. The microhardness ranged from 306 to 630 VPH and averaged 505 VPH.32 Hof jagd- und Rüstkammer, Vienna. A.2050. A sword with a multi-lobed pommel, perhaps from the 9th century, found at Atzenbrugg.33 The excavated condition of the blade made investigation of its microstructure easier. 31 Anstee, J.W. & Biek, L. “A study in pattern-welding” Medieval Archaeology, 5 (1961) 71–93 and pl.IV-XVI. 32 Lang, J & Williams, A. “The hardening of iron swords” Journal of Archaeological Science, 2 (1975) 199–207. 33 Thomas, B. & Gamber, O. “Katalog der Leibrüstkammer, I.” (Vienna, 1976) 45.

Pattern-Welding

77

The centre shows a core decorated with pattern-welding. Samples detached from the cutting edge show that the edges were made of steel (about 0.6–0.7%C) welded onto an iron core. The sword was not, however, quenched.34

Fig. 6. The steel cutting edge is welded onto an iron core. Microhardness of the edge (average) = 262 VPH.

Sword (perhaps 9th century) with a pattern-welded blade in a Private Collection.

Fig. 7. Close up of blade with pattern-welded centre.

34 Williams, A.& Edge, D. “Some Early Medieval swords in the Wallace Collection and elsewhere” Gladius, 23 (Madrid, 2003) 191–210.

78

chapter five

The edge shows a microstructure consisting of a mixture of slightly-tempered martensite, with areas of nodular pearlite and very few ferrite grains, and a few small slag inclusions, elongated by forging. Microhardness ranges from 360–434; average = 389 VPH

Fig. 8. The centre shows a microstructure consisting of areas of very fine pearlite and no visible slag inclusions (scale bar = 50 microns).

Fig. 9. The edge shows a microstructure of tempered martensite and nodular pearlite (the microconstituent formerly called “troostite”) (scale bar = 50 microns).

Pattern-Welding

79

Fig. 10. Blade edge: note the feathery material which may be bainite, and two phases in the slag inclusion (scale bar = 10 microns).

Microhardness ranges from 236–496; average = 354 VPH . This is a medium–carbon steel which has undergone a heat-treatment to harden it. The centre has been cooled at a slightly less drastic rate than the edge, which is probably why its hardness is slightly less. This may be the consequence of the edges of a blade, being narrower, cooling faster when immersed in a liquid. A Sword with a pattern-welded blade in a Private Collection. A Scandinavian sword with a pattern welded blade; probably late 8th century or early 9th century.35

Fig. 11. The edge shows a microstructure consisting of a uniform mixture of pearlite with a few isolated small grains of ferrite (perhaps 0.6%C overall) and numerous very small grey slag inclusions.

35 Oakeshott, op. cit. 24.

80

chapter five

Fig. 12. Pearlite and a little ferrite (scale bar 50 microns).

Microhardness ranges from 210–241; average = 224 VPH SEM analysis indicated that the metal matrix is virtually pure iron. There are both some elongated and round inclusions of iron silicate (with Ca, Al, Mg, K) a typical bloomery slag. Appendix—Some Destructive Tests Albert France-Lanord managed to obtain samples from six franciscas of the 5th–6th centuries CE and carried out micromechanical testing with the apparatus of Chevenard. Tensile testing of Merovingian steels Limit Elastique E (kg/mm2)

Resistance R (kg/mm2)

Elong.%

16.8 7.7 14.1 10.5 9.1 7.0

30.5 29.7 38.4 30.4 34.6 24.7

54.7 54.7 49.3 59.1 61.9 53.3

The carbon content of these weapons was not reported. Their high ductility suggests a low C%. These results might be compared with modern materials: (at room temperature)

Pattern-Welding

81

Tensile testing of modern steel Steel Yield Strength (kg/mm2) 0.2 % C steel 0.4 % C steel 0.6 % C steel 0.8 % C steel

29.5 35.3 37.2 37.6

Tensile strength (kg/mm2) 39.4 51.9 61.6 62.6

 Elongation % 36 % 30 % 25 % 23 %

Note that cooling the steel quickly after forging to give fine pearlite may increase the tensile strength by half as much again.36 It is interesting to compare these with the results of the mechanical testing carried out by Zschokke on some Damascus blades.37 He was a friend of the traveller Henri Moser who sacrificed six Indo-Persian blades (from his collection of 2000) in the interests of Science. Two blades from Solingen were also compared. One (# 11) was a pattern-welded blade of 0.61 %C, and the other (# 12) was a cast steel blade of 0.50% C. The Indo-Persian blades all had a surface pattern of some sort. These were due to rows of cementite in the hypereutectoid steels (1.3 to 1.9 %C) except for that in # 8 which was only 0.6% C, in which the pattern was due to rows of ferrite grains in a pearlitic groundmass. He carried out detailed chemical and microscopical analyses as well as (destructive) mechanical tests. Bending was studied by the Heyn test: samples were placed in a vice, and bent through 90º by repeated gentle blows with a hammer, then gripped again in the vice, and bent back through 90º again. The number of such bendings until breaking is taken as an indication of its resistance to folding (“chiffre de resistance au pliage”). This is suitable for ductile metals, but the samples (75 × 6 × 3.5mm) which were cut from the sabres were of such hard steel that they broke without showing any plastic distortion. A second experiment was to try and measure the impact needed to break the samples. A quasi-static test was used (comparable to a modern 3-point bend test), on an Amsler machine. The samples were placed on shelves 50 mm apart, and a load applied with a blade whose edge was a semicircle, of radius 3mm. Each test was carried out three times and the results averaged. 36 Askeland, D. R. “The Science and Engineering of materials” (1996) 333. 37 Zschokke, B. “Du damassé et des lames de Damas” Revue de Métallurgie (Paris, 1924) 21, 635–669.

82

chapter five

His results are tabulated below: the first column is the number of the sword; the second is its carbon content, R is the modular resistance to failure (the units reported were tonnes.cm–2 which are converted to kg. mm–2 by multiplying by 10 in order to compare them with the results of France-Lanord and Salin). #7, #8, #9 and #10 are Indo-Persian sabres. These blades are generally three or four times as resistant to bending as the Merovingian ones. W is the Work done in bending the sample to failure, reported in kg. cm, and multiplied by 0.098 to convert these units to J. A is the bending angle at rupture; this is a measure of elastic deformation—which is very small. The angle of bending of the Indo-Persian wootz steels ranged between 17º and 27º although the Solingen swords managed 69º and 78º. Tensile testing of Indo-Persian (“Damascus steel”) sabres Sword number

Carbon content

R

W

A

7. 8. 9. 10. 11. (Solingen) 12. (Solingen)

1.87% 0.60% 1.34% 1.73% 0.61% 0.50%

134 153 115 145 216 300

 9.4 22.1  5.5  6.3 36.1 62.2

27º 59º 19º 17º 69º 78º

The Revival of Science in Europe

PART three THE “DARK” AGES IN EUROPE

83

84

chapter six

The Revival of Science in Europe

85

Chapter six

The Revival of Science in Europe The works of the great scientists of the Ancient World, Archimedes, Aristotle, Ptolemy, Euclid, Hippocrates, and others, did not survive in Latin—the Romans seem to have been completely incurious about theoretical science. So Greek science was lost to Medieval Europe (Bede was possibly the last man in Britain to read Greek) and its revival had to await the arrival of the Arabs. The eastern half of the Roman Empire with its capital at Byzantium (renamed Constantinople) survived the final collapse of the Western Empire in the 5th century but found itself threatened by a resurgent Persia. King Chosroes invaded in 603, capturing Syria, Palestine, and Egypt, and going on to besiege Constantinople itself, in co-ordination with the Avars, in 625. The Byzantine (as he should now be called, although he insisted on the title of “Roman”) Emperor Heraclius recruited Turkish auxiliaries and invaded Persia in turn. Chosroes was defeated and forced to return his plunder, including the True Cross from Jerusalem. Within a few years, both empires were to be overcome by a new power. In 622 the Arabic merchant Mohammed was exiled from Mecca to Medina. From this date, his followers now based their calendar (AH = After the Hijira). He proclaimed a new monotheistic religion, Islam, which owed a good deal to Judaism and Christianity, and incorporated considerable appeal to the followers of both. On his death in 632, united under a caliph, the Arabs invaded the eastern provinces of the Byzantine Empire, capturing Damascus in 634 and Jerusalem in 637. The Sassanian Empire fared no better, being conquered by the Arabs over the next few years (637–651) while Egypt had surrendered in 640 and Alexandria the following year. The speed of these conquests suggests a considerable dissatisfaction among the inhabitants of the Near East with their Imperial masters, exacerbated by religious divisions ever since the Councils of Nicaea (325) which had tried to impose uniformity of belief among Christians, and of Chalcedon (451) which led to the final separation of the Monophysite churches of the East. Many of the armies of Islam would hereafter be

86

chapter six

Egyptian, or Persian, Turk or Berber, rather than Arabs from Arabia, but they all subscribed to Islam and essayed to read the Koran in Arabic. So the label “Arab” is widely used rather than the more accurate “Muslim”. Whatever their geographical origin, the armies of Islam conquered the Roman provinces of Africa and invaded Spain in 711, continuing on to an unsuccessful attempt to invade France in 732. These successor-states to Rome, Byzantium and Persia soon became centres of a new version of the civilisation of Antiquity. The Arabs made Damascus the capital of their caliphate. During this period their armies had overrun North Africa, Spain, Central Asia as far as Samarkand, and reached the borders of India. In 750 the Abbasids made themselves caliphs, and founded the city of Baghdad not far from the ancient capital of Babylon, nor indeed far from the Greek capital of Seleucia and the Persian capital of Ctesiphon, and their rule survived until the Mongol invasions of 1258. Their far-flung empire did not survive intact, however. Spain became an independent caliphate in 912, and Egypt another in 972. Arabic-speaking and Muslim, but tolerant of Jews and Christians (provided they paid their poll-tax), and employing Roman and Persian technology, the various caliphates became patrons of science. But Baghdad, under the Abbasid caliphs (most famously Harun al-Rashid (763–809), the caliph of the “Arabian Nights”) was the cultural centre of the Muslim world. Many scholars flourished at the court of the Abbasids, and much Greek science and Roman technology was revived through the medium of translations from the Greek and Syriac. Khalid ibn Yazid, who had the State Accounts kept in Arabic instead of Persian, and “being himself a scholar, was greatly interested in the Sciences. He was particularly attracted to the Art (of Kimiya, that is, Chemistry) so he ordered some Greek philosophers who lived in (Cairo) to be summoned and instructed them to translate those Books on the Art from Greek and Coptic into Arabic. This was the first translation from one language into another in Islam.1” As well as those works on alchemy from Egypt, Greek Science also arrived via another, more roundabout, route. Orthodox intolerance had driven the Syrian (Nestorian) Christians from the school of Edessa in 489 to the Persian city of Jund-i-Shapur,2 and they were joined by the 1 Stapleton, H.E. “The antiquity of alchemy” Ambix 5, (1953) 1–43 quoting the “Fihrist” 2. 2 The Sassanian ruler of Persia, Shapur (240–273) had established his capital at Jund-iShapur, (120 miles NNE of Basra) with a medical school and academy, to where scholars from all quarters of the Earth, perhaps also including India, were attracted.

The Revival of Science in Europe

87

Neoplatonists from Athens after 529.3 These cities saw the translation of many of the works of Greek Science into Syriac,4 from which they were readily translated into Arabic. With the break up of the Roman Empire into separate units of various sizes, ruled by those descendants of Germanic armies who had formerly been employees of the Empire, and nominally guided by a Church whose leaders looked to Rome for guidance and who purported to know some Latin, there seems to have been a general stagnation of trade and widespread economic collapse. The cash economy vanished, and was supplanted by a land-based feudal system. This was accelerated by further invasions of non-Christian peoples lured along former trade routes by the former wealth of the Roman Empire. The Magyars followed the Danube valley into Central Europe; the Vikings from Scandinavia raided the coasts and rivers all over Northern and Western Europe, and the Arabs, or rather Muslims, overran Spain and Sicily. Most destructive was the turning of the Mediterranean from a trading highway into a battleground. Eventually the Magyars were to form the Christian Kingdom of Hungary (1001), and the Vikings the Christian Kingdoms of Denmark, Norway and England (under Knut in 1017 and then again in 1066 via the Duchy of Normandy). With a revival in economic life, the Christian states of Western Europe started to recover military ground. Those Normans who had been less successful in Normandy were engaged by the Pope to defend the Church in Italy. They conquered Southern Italy from the Lombards and Greeks (1030–1071) and Sicily from the Arabs (1061–1090), going on to sack Rome itself in 1084 and later lead the First Crusade. The capture of Jerusalem by the 1st Crusade in 1099 might be seen as part of a counterattack against Islam. Sicily The Norman capital of Palermo saw a flourishing cosmopolitan culture, where the written languages were Greek and Arabic. The Emperor 3 Stapleton, H.E “The Sayings of Hermes quoted in the Ma’al Waraqi of ibn Umail” Ambix (1949) 3, 69–90: especially 88. This discusses a treatise “Risalah of Jamas the Sage for Ardashir , the King, on the Hidden secret … a wonderful treatise translated from Persian into Arabic”—probably compiled in Persia c550 ce after the Neoplatonists expelled by Justinian had been settled by Nushirwan at Jundi Shapur near Susa. 4 Sarton, G. “Introduction to the History of Science” (3 vols in 5 parts, Baltimore 1927– 1948; reprinted 1950) I, 381–2, and 417. Marco Polo found Nestorian churches all along the main roads as far as Peking.

88

chapter six

Frederic II (1194–1250) made Palermo his capital, and his court was a centre of intellectual activity. His court astrologer and alchemist was Michael Scot (1175–1232) and he welcomed Muslim and Jewish scholars to his court as well as employing Saracen troops in his battles with the Pope. Frederic knew six languages, collected a menagerie of strange animals, and took an interest in mathematics and philosophy. Stories about his “experiments” were the subject of gossip rather than reliable chronicles, as they reflect the attitudes of scandalised churchmen, but they do indicate a very different view of the world, almost “modern” one might say.5 Infants were ordered to be brought up in total silence, with their wetnurses commanded to bathe and suckle, but not to speak to or fondle them, in order to find out whether their primal language was indeed Hebrew. In the event, the infants all died before articulating a word. He was also said to have sealed up a live man in a cask until he died, the cask they being very slowly opened to see whether anything emerged, in order to prove that the soul totally perished with the body. Nothing was observed. Michael Scot, the first translator of Averroes, wound up in the service of the Emperor, to whom, it is said, he one day explained “infinity” as follows. When Frederic was not looking, he drew a circle in the earth. Then he asked the Emperor where he had started the circle.6 Sicily may be thought to have been an ideal location for the transmission of knowledge, but although some Arabic works on falconry were rendered into Latin for the Emperor, translations do not seem to have been as widespread as elsewhere. Spain In Spain, there had been a Caliphate at Cordoba since 912 where Muslim, Jewish, and Christian scholars and craftsmen flourished side by side. The science of Greece and Rome survived, but it was taught in Arabic. A manuscript of Dioscorides was sent from Byzantium to the Caliph of Cordova in 948 for his library. There were no Greek readers in Spain, so a learned monk, Nicholas, was sent, who with the aid of a Spanish Jew, Hasdai ibn Shaprut, translated the Greek MS into Arabic.7 5 Baird, J.L. ed. “The Chronicle of Salimbene de Adam” (New York, 1986) 352. 6 Burnett, C. “Michael Scot and the transmission of scientific culture from Toledo to Bologna” Micrologus, 2, (Turnhout, 1994) 101–126. 7 Singer, C. “The earliest chemical industry” (1948) 65.

The Revival of Science in Europe

89

The states of the North, Leon, Castile, Navarre, and Aragon slowly began to reconquer Southern Spain. In 1085 the former Roman capital, Toledo, was conquered by King Alfonso VI of Leon & Castile, with the support of the Pope and some help from France. Its libraries seem to have been a revelation to the western churchmen who arrived there. Scholars of any religion who could read Arabic were eagerly sought out by Archbishop Raymond and translations into Latin were carried on at a great rate. Gerard of Cremona (1114–1187) was the most productive of translators. Having become dissatisfied with the knowledge available in Italy, he is said to have gone to Toledo and learned Arabic. He spent the rest of his life in making translations of some 66 treatises, presumably with the help of a group of assistants. His output included Ptolemy’s Almagest, the mathematics of Archimedes and the medical works of Hippocrates. The alchemical works of Jabir as well as the book “On Alums & Salts” attributed to Razi, were translated by various scholars during the 12th century. Euclid’s Geometry was translated by Adelard of Bath, and the Algebra of al-Khwarizmi by Robert of Chester, both around 1140. The development of the pointed “Gothic” arch towards the end of the 12th century and its application to a multitude of cathedrals over all Europe, has been credited to the rediscovery of Euclid, although the general revival of economic life might be an equally important factor. The Liber de compositione alchemiae, was translated by Robert of Chester, who stated that he finished the work in 1182, and said: quid sit alchemia nondum cognovit vestra Latinitas—“the Latin peoples do not yet know what alchemy may be”.8

They were soon to find out. Universities Universities had developed in Medieval Europe as places of advanced study originally centred on certain cathedral schools. The “trivium” (Grammar, Rhetoric, and Logic—or the reading, writing and speaking of Latin) was succeeded by study of the “quadrivium”(Astronomy, Music, Arithmetic and Geometry). Mastery of these Seven Liberal Arts made the

8 Sarton, G. (op. cit. 1927–1950) II, 2, 929.

90

chapter six

student just that, a Master of Arts. Further study might be in Medicine, Law, Philosophy or Theology (“the Queen of the Sciences”). The University of Paris had developed by the 12th century, and Oxford somewhat later, as a centre for the study of Philosophy and Theology. Medicine was the specialism of Salerno and Montpellier; Law that of the oldest university, Bologna. During the 13th century, much of their teaching, especially of Philosophy and Theology, became the province of the friars (Franciscans and Dominicans) and colleges developed for the residence of students, especially the less affluent ones. According to the chronicler Matthew Paris, Paris had 3000 students by the mid-13th century although most of these would not have been taking a degree.9 The recovered science and philosophy of the Greeks found its way into the university curriculum very quickly. Most of Aristotle’s scientific work had been translated by Gerard of Cremona, together with the Com­ mentaries (explanations) of Averroes. The acquisition of the complete works of Aristotle had a markedly unsettling effect upon the orthodoxy of the time, particularly his emphasis on knowledge obtained by empirical methods, which opened up new vistas of thought, including the possibility of applying reason to dogma. In 1209 the Provincial Council of Paris decreed that “neither the books of Aristotle on natural philosophy, nor commentaries on the same, should be read, whether publicly or privately, at Paris” and in 1215 this prohibition was renewed in the statutes of the University of Paris. However, these works were studied, and were swiftly accommodated to orthodoxy, and so by 1254, the University was not merely permitting them, but decreeing how many hours should be devoted to their study. Robert Grosseteste (a Franciscan, later Bishop of Lincoln, c1175–1253) insisted on the importance of direct translation from the Greek and his pupil Roger Bacon (another Franciscan, 1214–1294) stressed the value of experiments. Crombie famously dated the “rise of modern science” from 1277 when Stephen Tempier, Bishop of Paris, condemned various Averroist propositions which included the necessarily circular motion of the heavens and the uniqueness of the world, as being constraints upon the omnipotence of God.10

9 Rashdall, H. “The Universities of Europe in the Middle Ages” (3 vols, Oxford, 1936, reprinted 1987) passim. 10 Crombie, A.C. Chapter 3 in “Perspectives in Medieval History”, Drew, K.F. ed. (Chicago, 1963).47.

The Revival of Science in Europe

91

Transmutation Alchemy was also studied, not merely as a means of copying gold, by making base metals look like gold, but as a means of turning them into gold. Scholars debated whether one form of metal could be transmuted into another. The alchemists regarded all substances as being composed of one primitive matter (prima materia) and owed their specific differences to the presence of different qualities imposed upon it. The prima materia was early identified with “mercury”, not ordinary mercury of course, but the “mercury of the philosophers,” which was the “spirit” of mercury, freed from the four Aristotelian elements—Earth, Water Air, and Fire—11or rather from the qualities which they represented. These elements would not have meant substances like garden soil or tap water, but entities like Solid, Liquid, Gas, and Energy. Thus the operator had to remove from ordinary mercury, Earth or an earthy quality, and Water or a liquid quality, and to “fix” it by taking away Air or a volatile quality. The prima materia thus obtained had to be treated with “sulphur” to confer upon it the desired qualities that were missing. This “sulphur” again was not ordinary sulphur, but some principle derived from it, which constituted the philosopher’s stone or elixir—white for silver and yellow or red for gold. The doctrine that the metals were composed of “mercury” and “sulphur” persisted in one form or another down to the 17th century. Of course there were numerous variations. Thus in the alchemy of Michael Scot, it is said that there are four spirits—mercury, sulphur, arsenic and sal ammoniac—and seven metals—gold, silver, quicksilver, copper, tin, lead and iron.12 For another example, Vincent of Beauvais attributed to Razi the state­ment that copper is potentially silver, and any one who can eliminate the red colour will bring it to the state of silver, for it is copper in outward appearance, but in its inmost nature silver.13 Albertus Magnus Albert of Cologne (c1193–1280, later Bishop of Ratisbon, still later Saint Albert the Great, the patron saint of scientists) did a great deal to assimilate Aristotelean science, arriving from the Arabic, with the current teach11 Toulmin, S. & Goodfield, J. “The architecture of matter” (1962) 52–4. 12 Thorndike, L. “Michael Scot” (1965) passim. 13 Thorndike, L. “A history of Magic and Experimental Science” (New York, 1923–58) I, 669.

92

chapter six

ings of the Church, and wrote voluminously on philosophy and science. He saw no contradiction in writing about scientific topics, and there is no reason to suppose that the practise of alchemy per se was disapproved by the Church. Most of the chemical writings quoted are from churchmen; disapprobation of whom was generally the result of heretical theological views, rather than any mistrust of chemistry itself. Partington pointed out that the study and practice of alchemy were forbidden to the Franciscans and the Domi­nicans, not once, but several times during the 13th century with penalties increasing from imprisonment to excommunication.14 So the practice of science alone would not have led any cleric into trouble, but its application to magical ends. Albert in his book De Mineralibus (“About Minerals” written between 1248 and 1256) stated that alchemy cannot change species but merely imitates them—for instance, it colours a metal white to make it resemble silver or yellow to give it the appearance of gold.15 But he went on to say “those who colour metals white or yellow...they are deceivers, and do not make real gold or real silver …. I have had tests made on some alchemical gold and silver that came into my possession; and it endured six or seven firings...then all at once it was consumed and reduced to dross.”

Several chemical treatises were attributed to Albert including De alchimia: Semita Recta in which he mentions the Liber Fornacum “Book of Furnaces”.16 This book (ascribed to Geber) contains the first mention of nitric acid as well as the recipe for making crucible steel discussed in Chapter 3. Nitric acid was made by distilling alum or vitriol (naturally occurring metal sulphates) with saltpetre.17 The knowledge of saltpetre (potassium nitrate) meant that the discovery of gunpowder was not far away. Incendiary (and eventually explosive) mixtures involving saltpetre were known to Albert as well as to other writers such as Roger Bacon in the 13th century. Elsewhere in the Semita Recta,18 he said “Iron of the alchemist is not magnetic” This was perhaps a reference to the fact that iron becomes nonmagnetic at red heat—when the alchemist was experimenting with it. 14 Partington, J.R. “Albertus Magnus on Alchemy” Ambix, 1 (1937) 3–20. 15 Wyckoff, D. “Book of Minerals” (1967) 233 16 “Libellus de Alchimia” ascribed to Albertus Magnus and also known as “Semita Rectaˮ; trans. Heines, V. (Berkeley 1958) 70. 17 “A history of Greek Fire and Gunpowder” Partington, J.R. (Cambridge, 1960). 18 “Libellus de alchimia” transl. V. Heines (New York, 1958) 19.

The Revival of Science in Europe

93

In De Mineralibus19 Albert observed that “iron cannot be liquefied like other metals, but only softened. But, nevertheless, in a great fire, especially if sprinkled with sand and sulphur, it is distilled [sic] and purified. ” This was a reference to the removal of liquid slag, especially if sprinkled with sand which forms an iron silicate slag by reacting with hammerscale. “Distillation” here seems to mean any process involving a liquid, perhaps even the removal of excess slag as a liquid. 2FeO + SiO2 = Fe2SiO4 In this respect he was somewhat behind the times, as the ability to melt iron was being developed in Northern Europe during the 13th century (see Chapter 9). He went on to say “Steel is not a different specific form of metal from iron; it is merely the more subtle and watery part of iron extracted by distillation [sic]; and it is harder and firmer, because of the force of the fire and the fine division of its parts, which become harder when heated…smiths search out special waters to quench when they harden swords.”20 Of course, it is the heating and sudden cooling (“quenching”) of steel which causes it to become very hard, because of an internal crystalline change. The liquid in which it is quenched is less important than the time and the technique, but quenching fluids were a topic of deliberate mystification among craftsmen until well into the 20th century. Despite debates about theoretical considerations such as transmutation, the scholars of the university and the cloister were not as distant from the activities of craftsmen like metalworkers as might be supposed. A famous medical school had been established at Salerno during the 11th century. The “Questiones Alani” were composed there before 1225; one observed that “steel floats in mercury, but iron sinks”.21 The reason for this statement might possibly have been that a porous bloom of iron was used, full of cavities which might fill with a liquid, and tending to hide it from sight, (while not actually sinking) while an ingot of crucible steel would float.

19 Wyckoff, op. cit. 187. 20 Wyckoff, op. cit. 235. 21 Lawn, B. “The Salernitan questions” (Oxford, 1963) 37.

94

chapter six Knowledge of Liquid Iron in Medieval Accounts

As far as the scholastic compilers of the 13th century were concerned, iron could not be melted (m.pt.1550ºC), so it had to be forged, that is hammered red-hot into shape. But by the 14th century, simultaneous with the developments in iron production described in Chapter 9, scholars had begun to refer to iron being melted. These references are quite separate from those recipes for iron-arsenic alloys which featured so frequently in medieval collections.22 It is evident that achieving higher temperatures and handling liquid metals regularly marked a steady improvement in practical chemical skills throughout the Middle Ages. The 14th century also saw the spread of the cast-iron producing blast furnace into many parts of Europe. So that practical metallurgists (I use this term deliberately, as their progress in manipulating iron and steel must have been based upon systematic trial and error) who already knew that a liquid steel (crucible steel) could be obtained from the East, now knew that a liquid iron could be made in European furnaces. What is very significant is that this knowledge (that iron and steel could be liquefied) also spreads into the written accounts produced by scholars, who have generally been assumed to be merely desk-bound commentators, divorced from the workshop. Words for a special type of oriental iron start to appear. In the second half of the 13th century, Thomas of Cantimpré (1210–1280) mentions a fusible iron in his Encyclopedia “De rerum naturis” Book XV (De metallis): ...est et aliud genus ferri in orientis partibus quod vulgare alidea dicitur. Incisionibus aptum est et sit file sicut cuprum (et) argentum. Sed fusile non est sicut ferrum alairum partium…

which may be rendered: “..and there is another type of iron in oriental lands which is commonly called alidea It is suitable for cutting and may be (drawn) like copper or silver, but in melting it is not like the iron of other lands.”

His work draws on other encyclopaedic works such as “Speculum naturale” of Vincent of Beauvais.23 This metal is called alidena in Vincent and andun, or andana elsewhere.24 So Indian (or at least an Oriental) steel 22 Williams, A. “A note on liquid iron in Medieval Europe” Ambix, 56 (2009) 68–75. 23 Ferckel, C. “Thomas von Chantimpré über die Metalle” pp.75–80 in “Festgabe Lippmann—Studien zur Geschichte der Chemie” ed. Ruska (1927). 24 Lippmann, “Entstehung und Ausbreitung der Alchemie” (Berlin, 1919) 614.

The Revival of Science in Europe

95

seems to have been known as a steel both harder and of lower meltingpoint than European steels. At the end of the 13th century, Pseudo-Geber, or the author of the “Summa Perfectionis” (probably Paul of Taranto)25 had described iron as: ignibile, et non fusibile fusione recta “capable of firing, but not fusible with an [ordinary] furnace”. Certainly, iron in the form of filings can be burned, and sparks would accompany most operations in the smithy. But at the time, in Western Europe, iron was not melted in the usual course of events. During the course of the 14th century, however, this was to change. An anonymous treatise on metals of the 14th century from Southern Italy was first discovered by Zuretti (1924), since when Colinet26 has made a new edition and a French translation, which is the source of this quotation. The treatise was probably written in Calabria (a region of Italy then bilingual) circa 1377. “Iron, of which the Indian one is the best…. Iron can colour silver, cannot be easily separated from gold, and can be “purified” with acids …. Iron cannot be melted, except with great effort.”

William Sedacer (d.1380) was a Catalan compiler of alchemical recipes. His principal work called “Sedacina”, which was compiled when he was active at the court of Aragon, and a member of the household of the Infante John, in 1377. In this collection he wrote about the properties of different metals, and said: … Indorum sapientes vero de India inter cetera corpora ferrum elegerunt quia cito facit suum opus et leviter…

which may be rendered: “the sages of India choose iron from India from among the other types [of iron] because it is the quickest to work”27

Since the recipes following this one are all concerned with the liquefaction of iron, usually by adding arsenical compounds, it may be assumed that “quickest to work” means “the lowest melting-point”, evidently crucible steel.

25 “Summa perfectionis—a critical edition”, ed. W.R. Newman (Leiden, 1991) 11. 26 Colinet, A. “L’anonyme de Zuretti” Les Alchimistes Grecs, X (Paris, 2000) p.30. 27 Barthélemy, P. “La Sedacina ou l’oeuvre au crible” (Paris, 2002) 2 vols.

96

chapter seven

Chapter seven

The Survival of Technology From the Ancient World Alexandria had been the main commercial centre of Egypt, the terminus of the Indian Ocean trade, and a centre of science and technology, especially alchemy, the study of imitating gold. As well as being able to draw upon a wealth of experience in the doubtful arts of colouring the surface of metals to bamboozle the unwary, alchemy (al-kimiya = [the art of] the black [land] i.e.Egypt) also flourished because glassblowing had developed what we must call laboratory glassware; vessels of blown glass which alone enabled the experimenter to observe the success or failure of his manipulations. Arabic Alchemists Alchemy underwent a revival in the Arabic world, especially in the circle known as “The Brethren of Purity”, associated with the name of Jabir ibn Hayyan. It may be that the Islamic world offered a greater variety of material resources than the Roman Empire, for the scope of Arabic alchemy was eventually to become greater than that of Alexandria. Nor should the importance of contacts through trade, as well as invasions, with Indian & Central Asian technology be forgotten. Jabir (Geber) Jabir ibn Hayyan was the most influential chemist of his time and said to have died in 822.1 His works were so influential that they were frequently copied, and added to, over the centuries, so that many works came to be ascribed to Jabir, which may have had little connection with him. The most important chemical treatise of the Middle Ages was the Summa perfectionis which was supposed to have been written (in Latin) by the Arab 1 Holmyard, E.J. “Jabir ibn Hayyan” Proc.Royal Society of Medicine—History of Medicine Section, 16 (1922–23) 46–57.

The Survival of Technology From the Ancient World

97

“Geber”.2 A great deal of ink has been spilt over the authorship of this book, but unfortunately, the mention of any particular chemical, such as saltpetre, by Jabir or “Geber” is no guarantee that it was known to Arabic chemists of the 8th or 9th century. It may well be that it was a later discovery and described in a 13th century treatise which was ascribed to the prestigious author of 500 years earlier. Kraus, however, in his exhaustive book on Jabir suggested that “Jabir” was the nom de plume of a group of Ismaili propagandists writing in the 9th–10th centuries and his Arabic works were in fact forgeries. Likewise, the ascription of Jabir’s authorship to any book is no guarantee that he, or in fact any Arab, is the author, especially if it only exists in a Latin version. The first mention of Jabir is in the Fihrist al-uluum or “Catalogue of the Sciences” of Ibn al-Nadim, a librarian of Baghdad (c.988)—which says that “the Adepts claim he was the Supreme Alchemist; after the fall of the Barmakids [the family of viziers] he lived at Kufa…and he practised alchemy -later some one found a mortar with 200 ratl of gold in a portico—in which was also found an installation built for the alchemical treatments of dissolution and fixation … but other Scholars say he did not exist...Razi refers to him as his master.”3 Jabir was said to be the author of over a thousand books, but many of these have not survived even in Arabic. Some of Jabir’s recipes were translated by Berthelot; others by Holmyard, but many still remain untranslated.4 Holmyard’s project to edit and translate the works of Jabir never progressed beyond the first part of Volume I which contained 11 Arabic texts.5 Many more texts were edited and published by Kraus. The Book of Natural Properties (Kitab al-hawass) contains many of the recipes and techniques of Antiquity. But there is a group of recipes from this Book which are most interesting—6 Chinese grease for skins, leather, arms; Chinese glues; Chinese & Indian inks; making Silk waterproof; the imitation of Tibetan wood etc—because all of these suggest an oriental Source.

2 Ruska J “The history of the Jabir problem” Islamic Culture xi (1937) 303–12. 3 Kraus, P. “Jabir ibn Hayyan” (2 vols), vol. I . “Corpus dʼ écrits Jabiriens” (1943) and vol. II (1942–sic) “Jabir et la science grecque”: Memoires de l' institut d' Égypte, Le Caire, no.45. see I, xlix+ A ratl is approximately a pound (or 450 g). 4 Stapleton, H.E. “Note on the Arabic MSS on Alchemy in the Asafiyah library, Hyde­ rabad” Archeion, XIV, (1932) 57–61. 5 Holmyard, E.J. “The Arabic works of Jabir” (1928, only vol. I ever published). 6 Kraus, op. cit. II 79.

98

chapter seven

And there are also recipes ascribed to Jabir for the transformation of iron (narmahan) to steel (fulad) quoted by al-Jildaki (see Chapter 3). al-Razi Muhammad ibn Zakariyya al-Razi (860–925) who was known to Medieval Europe as “Rhases”, was said to have been the pupil of Jabir and a physician. Works ascribed to him include the books “On Alums and Salts” and “On the Secret of Secrets”. The hermetic, mystical, tradition was preserved in Jabir, and a belief in the possible transmutation of base metals into gold. The more practical tradition of making base metals look like gold, was exemplified by Razi. An anecdote is told about him, that having his eyes affected by his chemical experiments, he visited a physician, who cured his affliction and then charged him 500 dinars. “This” he was supposed to have said, “was the true path to making gold” and foreswore Chemistry for Medicine.7 A conclusion with which many of us might agree. Ruska went on to publish (1937) a translation from the Arabic of alRazi’s book “Secret Of Secrets”, which gives a very clear picture of 10th century chemistry.8 His treatises included detailed instructions for distillation, washing (solution), crystallisation, refining,etc. with very little mysticism.9 The third book of this was devoted to the treatment of different Substances; among the substances to be distilled were: mercury, sulphur, sal ammoniac (all of which would be obtained again in a purer state, although accompanied by some spectacular colour changes in the case of sulphur) and acetic acid, (vinegar, which could be concentrated by boiling, because because pure acetic acid boils at 118°C). The sal ammoniac (ammonium chloride, NH4Cl) could have been obtained in a crude form by the dry distillation of animal products such as, but not limited to, camel dung. Crude petroleum could be purified by distillation—when black it is refined by mixing it with sal ammoniac & distilling it several times to get a colourless distillate. Even the higher fractions were distilled “until the distillate does not catch fire immediately on contact with a flame”.i.e.until the collection of something like diesel fuel. Refining urine meant distilling 7 Lippmann, E.O.von “Abhandlungen und Vorträge …” (Leipzig, 1906 & 1913) 400. 8 Ruska, J. “al-Razi’s buch Geheimnis der Geheimnisse ” Quellen und Studien zur Geschichte der Naturwissenschaften und der medizin, Bd.6 (Berlin, 1937). 9 Partington, J.R. “The chemistry of Razi” Ambix, 1, (1937) 192–196.

The Survival of Technology From the Ancient World

99

it seven times; vitriols were “dissolved by distillingˮ.10 The products were called “spiritsˮ. Evidently the word “distillation” was also used to mean turning the vitriol into a liquid by dissolving it in water.11 al-Razi also mentioned what might be crucible steel. “What concerns the metals … we do not need to describe, apart from the Chinese iron, which resembles a mirror...in the way that it shines, (otherwise) it is unknown.”12 It could be argued that “Chinese iron” was cast iron, but the grey cast iron employed widely by the Chinese would have been wholly unsuitable for making a mirror. Of course, it could have been used to make the crucible steel, or else there has been some confusion between one metal (crucible steel) of Central Asiatic origin, and another metal (cast iron) of Chinese origin. Al-Kindi was also to mention mirrors in his book on swords (see Chapter 3). We know that crucible steels were used for making mirrors because an Islamic mirror in the British Museum (perhaps of the 14th century) has in fact recently been analysed, and found to have been made out of crucible steel.13 In the early 15th century, there is a reference to “the mirror of Floron” —a magic mirror, used for the revelation of past, present, and future— Fac fieri speculum de puro calibe ad mensuram palme unius in rotundo. “having been made of pure steel of size one palm (15cm) around”.

“Pure steel” was surely crucible steel, since bloomery steel, because of its slag content, would not supply a flawless surface, no matter how much it was polished.14 The later chronicler al-Dimashqi (1258–1327) said that “iron belongs to Mars … its hardness and strength varies with different mines. The best is the iron of China.. It cannot be cast (by fire) because of its dryness … its pores are impervious to fire. Acids exert a great influence on it; the acid of pomegranates dissolves it into a black solution, vinegar into a 10 Forbes, R.J. “A history of the art of distillation” (Leiden 1948) 38 11 Garbers, K & Weyer, J. “Chemie und Alchemie der Araber im Mittelalter” (Hamburg, 1980). This includes a book on “Perfumes & distillation” which was edited by Garbers (Leipzig, 1948). 12 Ruska, J. “al-Razi’s buch Geheimnis der Geheimnisse” Quellen und Studien zur Geschichte der Naturwissenschaften und der medizin, Bd.6 (Berlin, 1937) 85. 13 Craddock, P. & Lang, J. “Crucible steel—bright steel” Historical Metallurgy, 38 (2004) 35–46. 14 Kieckhefer, R. “Forbidden Rites—a necromancers manual of the 15 century” (Stroud 1997): CL M 849 is a MS of the first half of the 15th century: the mirror of Floron (fols 37r -38r).

100

chapter seven

red solution, like gold, and salts dissolve it into a yellow solution, like saffron”.15 This seems to be slightly confused. The iron of China (“cast iron”) is certainly the easiest to cast, but the author may have meant to say that “Chinese iron was the hardest, but that iron in general could not be cast”. Iron and concentrated acetic acid will eventually form a dark red solu­tion, which is believed to be a complex ion, hexa acetate tri iron [Fe3(CH3COO)6] 3+ and on boiling the solution, the iron is precipitated as a complex hydrated oxide & acetate.16 Iron will dissolve in hydrochloric acid (or mixtures of acidic salts with chlorides) to give, on standing, a yellow solution of [FeCl4]– ions. Distillation Aristotle17 had observed that sea water when boiled and condensed becomes sweet, but the earliest description of the distillation of pure water from sea water is, however, by Alexander of Aphrodisias (~200 ce) who describes sailors boiling sea water and collecting the condensate with sponges. This method was also extended to tar.18 Pliny19 described how resin, from trees such as the cedar, could be boiled and the vapours condensed by a fleece held over the vessel, and subsequently wrung out, to give a clear liquid (turpentine); the residue is rosin. The knowledge of this technique does not seem to have been universal throughout the Roman world, however. Muslim ex-pirates settling at Fraxinet (Freinet) in Provence, after 889, had to show the locals how to make bottle-stoppers from the bark of cork trees, and how to produce tar (goudron) from pine-resin.20 These simple forms of distillation had been elaborated by the Alexandrian alchemists into three separate vessels for (i) the boiling of the liquid (cucurbit, bikos), (ii) for the condensing of the vapours (ambix, alembic,) and (iii) for collecting the distilled liquid (phial). The apparatus 15 Mehren, A.F.M. “Manuel de la cosmographie du moyen age par al Dimashki” (Kopenhagen 1874) 60. 16 Remy, H. “Treatise on Inorganic Chemistry” (1956) II, 285. 17 Meteorologia, II, 3. “The Works of Aristotle”, ed. W.D. Ross (repr.Chicago, 1952) II, 464. 18 Forbes, R.J. “Bitumen and petroleum in Antiquity” Ambix, 2 (1938) 68–92. 19 Bailey, K.C. (ed) “The Elder Pliny’s chapters on chemical subjects” (2 vols, 1929–32) xv, 7. 20 Reinaud, J.T. “Muslim colonies in France” (repr.Lahore, 1955).

The Survival of Technology From the Ancient World

101

Fig. 1. Distillation apparatus with multiple receivers (Wellcome Library, London).

102

chapter seven

for performing this needed to be made from (blown) glassware, otherwise any control of the distillation is almost impossible. From Alexandria, the techniques of distillation, as well as glass-making, passed to the Arabs, together with much other chemical knowledge. Having developed a technique which produced more desirable products from one material, such as pine-resin, there can be no doubt that early chemists then applied them to every available substance which might be “improved”. Greek Fire One of the most significant applications of distillation was that of petroleum to produce a volatile liquid, naphtha (from the Arabic word, naft) to be used in “Greek Fire”. This was developed in the eastern Roman Empire, while it still had access to the oil-fields of Mesopotamia, and found military application, especially defending the narrow channels of the Bospho­ rus and Dardanelles against seaborne attack. The use of incendiaries in warfare has a long history and by no means the first, but the most famous example of their use was this liquid called “Greek Fire” supposedly invented by Kallinikos of Heliopolis (Syria). The history of this famous weapon has been related by many historians, especially Partington.21 The onrush of Arab conquest was stemmed at Constantinople in “the siege of seven years” (674–680), when their fleets were destroyed by a “liquid fire” sprayed from the Byzantine ships. Further attacks by the Arabs in 716–7 and the Varangians (Vikings) from Kiev in 941 were defeated by similar means. al Dimashqi (see below) gives an early description of crude petroleum, and adds: “many types of naft are water white and so volatile that they can not be stored in open vessels”. This inflammable liquid, naphtha, possibly mixed with resins or other materials, was sprayed at the enemy by means of a brass force-pump. It floated on the surface of the water, and was ignited in the air or on the surface, probably by fire-arrows. It would have been ideally suited to use in the confined spaces of the Dardanelles and Bosphorus, but less useful in the open sea, or at sieges. The name “Greek Fire” however, was transferred to any incendiary weapon. The incendiaries catapulted by the Egyptian Mamelukes at the 21 Partington, J.R. “A History of Greek Fire and Gunpowderˮ (Cambridge 1960). And also Hime, H.W.L. “The Origins of Artillery” (1915).

The Survival of Technology From the Ancient World

103

Crusading army of King Louis IX in 1250 were also called “Greek Fire” by de Joinville22 Steam Distillation Perfumes had been made in Antiquity from herbs and flowers by the method of enfleurage, that is by rubbing the petals with oil or fat to dissolve their essential oils. Simple distillation , if attempted, would be likely to decompose the essential oils. For example, geraniol (C10H18O, the active ingredient in rose-, lemon-, and geranium-oils) will boil at 229°C but decomposes shortly afterwards, with disgusting results; the solution, still employed today, is steam distillation. It consists of distilling the plant products with water (or steam); a mixture of 2 immiscible liquids boils when their combined vapour pressures reaches atmospheric pressure. In the presence of steam a mixture of geraniol and water vapours will boil at around 90°C, and a mixture of the two liquids will condense. On standing, most of the water will separate, leaving the essential oil behind. The manufacture of essential oils and perfumes by steam distillation was an important industry based in Persia & Babylonia; perhaps starting in the 9th century & well-established by the 13th century,23 when the chronicler al-Dimashqi (1258–1327) described, with illustrations, how ten different types of essential oils were made there from violets, lotos, narcissus, lilies, carnations, and others. The centre of production of rose water was Shiraz, where steam distillation had replaced older methods by the 9th or 10th century, which exported its product all over the world from China and India to Spain, and paid an annual tribute to the Caliph of 30,000 phials. Of course, even distilled water had it uses. It was recognized as being purer, and Gilbertus Anglicus (c1220) recommended distilled water for travellers.24 Sulphuric Acid The distillation of many other substances was doubtless attempted, in the hope that interesting products could be obtained from them also, and 22 Joinville, Jean de, “Memoirs of the Crusades” (trans. Marzials, F. 1908) 186, 195. 23 “Manuel de la cosmographie du moyen age par al Dimashki” Mehren, A.F.M. (Kopenhagen 1874). 24 Sarton, op. cit. II, 2, 658.

104

chapter seven

eventually useful products were obtained from salts, such as alums and vitriols. Sufficient useful products must have been obtained to encourage repeated boilings and distillations, but the regular preparation of mineral acids would have needed the availability of glass retorts since the vapours were so corrosive. If iron pyrites (or similar minerals) are allowed to weather, then iron sulphate, or a mixture of iron and copper sulphates, is formed. One name for iron sulphate was copperas which reflects the confusion that its green colour, reminiscent of copper compounds, led to. It was also called green vitriol, misy, or atrament. The process of distilling green vitriol in cucurbit and alembic “..to give a white water distillate” (= sulphuric acid) was described in the Kitab risala jafar al Sadiq (Book on the Philosopher’s stone); perhaps of the 13th century, according to Ruska.25 The green crystals of FeSO4.7H2O will lose water of crystallisation, and then decompose (like other sulphates, such as alums) to give a mixture of the gases, sulphur dioxide and sulphur trioxide; if the latter is collected in the same receiver as the water, sulphuric acid will form. then

2FeSO4 = Fe2O3 + SO2 + SO3 SO3 + H2O = H2SO4

Sulphur dioxide in water will form the less stable sulphurous acid, H2SO3, which will slowly decompose on standing to form more sulphuric acid, H2SO4. SO2 + H2O = H2SO3 3H2SO3 = 2H2SO4 + H2O + S so sulphuric acid is formed. It was known by Razi that boiling vinegar produced a stronger liquid; it would therefore have been logical to try boiling this product also. If the liquid collecting in the receiving flask is boiled (not distilled) again, then the acid will become more concentrated; since the boiling-point of sulphuric acid is about 270°C, with some decomposition into SO3—much higher than water (100°C). Eventually a maximum boiling-point mixture of 98% sulphuric acid

25 Ruska, J. “Arabische Alchemisten, II, Gafar al-Sadiq” (Heidelberg, 1924) 73.

The Survival of Technology From the Ancient World

105

Fig. 2. Distillation without water-cooling is shown on the upper furnace in this 15th century manuscript of alchemical texts attributed to Ramon Lull. Note the presence of a balance as well as the furnaces and alembics. The lower furnace seems to be used simply for digesting solutions (Wellcome Library, London).

106

chapter seven

(317°C) is reached; much higher than water. This is the “concentrated sulphuric acid” of commerce, also known as oil of vitriol. The destructive distillation of other vitriols (blue, copper; white, zinc) would eventually give a similar oil. Saltpetre and Gunpowder The manufacture of gunpowder, and indeed modern explosives also, depends upon the availability of nitrates, because they are the strongest oxidising agents found naturally. The essential component of gunpowder is potassium nitrate (saltpetre, KNO3). Wherever discovered, once fairly pure saltpetre had been isolated, its use in weaponry would have followed rapidly. Incendiaries were probably the first applications of saltpetre, since saltpetre will “improve” any incendiary mixture by speeding up the rate of burning. For example, in the year 1312, during the streetfighting at Rome, the Emperor Henry VII used a mixture of sulphur together with saltpetre to destroy the barricades.26 Charcoal and saltpetre make a very effective incendiary mixture, and then a little sulphur might have been added to make it easier to ignite, since sulphur will form a liquid at the fairly low temperature of 114°C, which will then start the reaction between the other two solids. Explosive saltpetre-containing devices are described in the 13th century by Latin authors, including an anonymous author calling himself “Mark the Greek”, Roger Bacon, and Albert of Cologne (St. Albertus Magnus). Roger Bacon (c1214–1292): this remarkable Franciscan has been hailed as an early English Leonardo da Vinci, and credited with the invention of gunpowder, because in the treatise Epistola de secretis operibus artis et naturae (“Book of the secret works of art & nature”) probably dating from 1257, he attempted to prove that the effects commonly attributed to the work of evil magic are natural and can be imitated by experiments. Among various recipes is the notorious one from BM Sloane 2156, which was used by Hime to “prove” that he invented gunpowder.27 Another work of his, the Opus Tertium which was somewhat later (dated around 1266–7) gives an unambiguous recipe for gunpowder, albeit only used in 26 Lippmann, E.O. von, “Beiträge zur Geschichte der Naturwissenschaft und der Technik” (Weinheim, 1952) II, 83–85. 27 Partington, J.R. “A History of Greek Fire and Gunpowder” (Cambridge 1960) 74.

The Survival of Technology From the Ancient World

107

a firecracker. This was translated by Little,28 and I have followed his version. “For example there is a childʼs toy of sound and fire made in various parts of the world with a powder of saltpetre, sulphur and willow charcoal. For with an instrument of parchment the size of a finger which is filled with this powder, one can make such a noise that it seriously distresses peoplesʼ ears, especially of those taken unawares, and likewise the terrible flash is also very alarming... If one were to make such an instrument with a solid body, then the violence of the explosion would be much greater.”

Roger Baconʼs works are not the only source of gunpowder recipes in 13th century Europe. They are to be found in an anonymous 13th century treatise called Liber Ignium (The Book of Fires) of Mark the Greek, which is a collection of some three dozen recipes, some for incendiaries, including military applications, and some for entertainment. A number of the same recipes turn up in a much larger compilation called De Mirabilibus Mundi (On the Marvels of the World) attributed to Albertus Magnus. Recipes for distilling wine to make “burning water” (i.e. alcohol) and distilling oil to make “Greek fire” are included. His recipe for “gunpowder” is quoted here.29 “Flying fire: take one pound of sulphur, two pounds of willow charcoal six pounds of saltpetre; these three should be ground together very finely on a marble slab; afterwards as much as needed may be placed in a paper packet for making flying fire or thunder. For flying, the packet should be long, slender, and full of the best powder; for making thunder, it should be short, wide, and half full.”

This mixture was evidently employed to make both rockets and exploding fireworks, although the true “gun” as a missile weapon was not to appear until the 14th century. Nitric Acid However, the addition of saltpetre, even if impure, to the vitriol in the cucurbit when distilling it was to give an entirely different product, and one which had astonishing properties as a solvent. It led to the formation of nitric acid. 28 Little, A.G. “Part of the Opus Tertiumˮ, Proceedings of the British Academy (1928) xiv, 290. 29 Partington, op. cit. 70.

108

chapter seven

Fig. 3. “Flying fire ” is essentially a rocket without a stick. This drawing is adjacent to a recipe for gunpowder. Kyeser “Bellifortis” (1405 ) fol.102b.

The Survival of Technology From the Ancient World

109

KNO3 + H2SO4 = HNO3 + KHSO4 Nitric acid is more volatile than water and much more volatile than sulphuric acid, and will boil over at 83°C. Efficient condensation is needed, otherwise the laboratory will rapidly fill with extremely corrosive fumes, red in colour, and excruciatingly painful to inhale. This would speedily supply a strong motive, if any were needed, for improving the efficiency of condensation by water-cooling the alembic. The nitric acid collected (aqua fortis) will rapidly dissolve all metals, except gold, with the spectacular generation of toxic red fumes of nitrogen dioxide, NO2. The addition of saltpetre to the solids being distilled could not have happened before it became generally known, which was some time in the 13th century. Eventually the manufacture of nitric acid became a side-line of the extensive saltpetre (and hence gunpowder) industry; located in Venice from the 15th century and in France and Germany from 1500. The first mention of nitric acid seems to be in the Liber fornacum “Book of Furnaces”30 ascribed to Geber, the Latin writer who was probably not Jabir ibn Hayyan, although Holmyard was convinced otherwise. An English translation of the “Book of Furnaces” was published by Richard Russell in 167831 in “The Works of Geber” which gives this recipe for making nitric acid: “Red mercury is sublimed thus; one pound of it is mixed and perfectly well ground together with one pound of Saltpeter and one pound of Vitriol, and from them it is sublimed red and splendid.” If by red mercury, the oxide (HgO) is meant, then it is being decomposed to mercury, which is boiled away, and the vapour reoxidised. If this was carried out in earthenware containers, then it would seem as though the HgO had never changed, but only sublimed. At the same time, nitric acid would be formed, which would dissolve any mercury it came into contact with. It is possible that some mercury condensed on warmer parts of the apparatus, separately from the nitric acid. In the 14th century Magister Ortholanus (Guillaume dʼ Ortolan, later Bishop of Bazas, d.1417) said that after the heating of a mixture of nitre, alum and blue vitriol, the first drops of distillate should be thrown away

30 Multhauf , R. The Origins of Chemistry (1966) p.207 “Liber fornacum” chapter 15—“a dissolving water from Saltpetre & Vitriolˮ. And see Darmstaedter, op. cit. (1922) 179. 31 “The Works of Geber” was reprinted and edited in 1928 by Holmyard. It also contains the recipe for making crucible steel quoted in Chapter 3.

110

chapter seven

as they contain nothing but water. It was to be tested by putting a few drops on a clean knife and watching the reaction.32 Nitric acid will dissolve iron very rapidly, unless very pure nitric acid is used, in which case a layer of oxide will form so quickly as to render the iron passive. Aqua Regia If salt is also present in the mixture of solids in the cucurbit, then a mixture of hydrochloric and nitric acids will be formed (“aqua regia” or “royal water”) which will dissolve all metals, including gold. This is probably due to the formation of free chlorine, via nitrosyl chloride, NOCl. Gold is insufficiently reactive to dissolve in nitric acid alone, but can react with chlorine to form (red) gold chloride, AuCl3. This forms a stable yellow solution in water, HAuCl4. An early mention of this may be in the Summa Perfectionis ascribed to Geber (which was probably written by the late 13th century writer Paul of Taranto) which describes a red elixir from gold. It is certainly mentioned in the Liber de inventione veritas sive perfectionis “The book of the finding of truth or perfection”,33 also ascribed to Geber, which gives this recipe. Chapter 20 “of red medicines for Venus and Marsˮ: “Take tutia, calcine and dissolve it in the Water of vitriol and peter; then with that water imbibe the calx of sol, That it may drink in double its own weight of the Same Water. Afterward by distillation receive the Water from it and revert it upon the Calx four times.”

Tutia is usually Zinc oxide, but in this recipe it must be a chloride, or nothing will happen. If a chloride is mixed with sulphuric acid and saltpetre, and the result distilled, it will be aqua regia, which will dissolve gold, more readily if it is finely divided. The gold chloride solution will be a “Red Medicine” because if either copper (Venus) or iron (Mars) is dipped into it, a layer of gold will immediately be deposited. To the observer, copper or iron will have been turned into gold, which would have been mightily impressive. The fact that it was only the original gold, being re-precipitated after dissolving, would not necessarily have been immediately apparent. 32 Forbes, R.J. “A short history of the art of distillation” (1970) 86. 33 Geber, ed. Holmyard, 221

The Survival of Technology From the Ancient World

111

A very dilute solution of gold chloride, Aurum potabile (drinkable gold) was regarded as an efficacious medicine in the 14th century. It would certainly have been an expensive one—which has often been a significant factor in patients’ attitudes to medicines. Various Recipe Collections Several Latin collections of chemical recipes, which may be briefly discussed, survived in Western Europe, although no works of theoretical Science in Latin seem to have survived the fall of the Roman Empire. The technologies of the Roman Empire continued to be practiced, and their recipes continued to be collected. Craftsmen were not perhaps a literate class, even if they ever felt inclined to record their trade secrets, which would have been seldom. Not until the spread of monasteries following the Benedictine rule (established at the Abbey of Monte Cassino about 529 ce, and which enjoined manual labour rather than begging as the monks’ principal means of support,) throughout Western Europe were craftsmen to be found in communities of scholars, although the two groups were not necessarily identical. Not all monastic copyists had a complete understanding of the recipes they copied, and it was accepted in the Middle Ages that the abstract labour of the intelligence was in a separate category to the manipulation of material substances. But the main written source for our knowledge of medieval chemistry does come from these monastic compilations. (The analysis of medieval artefacts provides another source of knowledge, and one frequently overlooked.) A late 8th /early 9th century Latin MS from Lucca, known as “Compo­ sitiones variae” contains a series of recipes for making pigments, dyestuffs, and for colouring metals. Many of these recipes clearly derive from Antiquity and are repeated again, with additional material, in the 12th century compilation “Mappae Clavicula”. This has been published several times, most recently by Smith and Hawthorne, who edited and translated it, and also compared its contents with the Lucca MS.34 As the editors observe, it is “a compilation of compilations”; many of the recipes are repeated, and many are downright confusing. There are recipes for making various pigments, writing with gold, gilding,35 “extend34 Smith, C.S. & Hawthorne, J.G. “Mappae Clavicula—a little key to the world of Medieval techniques”. Transactions of the American Philosophical Society, 64, 4 (Philadelphia, 1974). 35 In the case of iron, by coating it with copper first (chapter 12).

112

chapter seven

ing” gold, soldering gold and silver vessels, colouring metals (generally to look like gold), colouring glass, and dyeing leather. There are also incendiary mixtures, based upon naphtha, with other ingredients such as pitch and sulphur, but not with saltpetre. Out of some 382 recipes in the 12th century version of the “Mappae Clavicula”, 107 also appear in the earlier Lucca MS, but the later version contains the earliest known recipes for making soap (by heating woodash and olive oil) and alcohol (by distilling wine) in Europe. Other collections of recipes that might be mentioned include the 10th century “On the colours and arts of the Romans” by Eraclius, the 13th century “Book of Fires” by Marcus Graecus (which includes an early recipe for gunpowder)36 and the 12th century guide to every craft needed for equipping a church De diversis artibus (“On divers Arts”) by Theophilus.37 This book was written in the 12th century, by an anonymous Benedictine of Reichenau, perhaps Roger of Helmarshausen. It is a fund of practical information about silversmithing, glassmaking, organ building and bellcasting. There is also some chemical and metallurgical information to be had. On metalworking, there is a recipe of his, much quoted, for case-hardening small tools:38 Small files are made of iron, cut with a hammer and chisel, and then coated with old pig fat and strips of leather. They are covered with clay, leaving the tangs bare, and heated for some (unspecified) time. “Hastily extract them from the clay, and quench them evenly in water. Then take them out and dry them at the fire.” The organic matter would decompose to carbon, which would diffuse slowly into the iron, and form an outer case of steel,39 which would then harden on quenching. He does mention that “some files are made of steel”, and they can be quenched by plunging red-hot into water. The “drying” afterwards, mentioned in both of these cases, was perhaps a gentle reheating to temper the quenched steel. Engraving tools are to be hardened by heating the tips until they are red-hot, and then quenching. This would probably result in a slack (less than full) quench, which might have been more suitable for such tools. 36 This work is discussed extensively in “A history of Greek Fire and Gunpowder” Partington, J.R. (Cambridge, 1960). 37 Theophilus: “On divers arts”, translated & edited by Hawthorne, J.G. & Smith, C.S. (Chicago, 1963, reprinted 1979) 38 Theophilus, op. cit. 93–95. 39 Around 12 hours at 900°C to give a case depth of 2mm (ASM Handbook, 4, 142.)

The Survival of Technology From the Ancient World

113

The microstructures of those swords discussed later should be studied with these recipes in mind. His account was somewhat confusing about smelting iron.40 The ore is to be “dug out ..and smelted down into lumps. Then it is melted on an iron-workers forge, and hammered so that it becomes suitable for any kind of work.” When this edition was prepared it was generally believed that the “fining” or decarburisation of liquid cast iron was not earlier than the 15th century, and therefore the liquid Theophilus mentions must have been slag. However, the date of the invention of this process has since been advanced, and it is now thought to have been practiced in 12th century Sweden (see Chapter 9). This is still somewhat earlier than Theophilus, but now it seems less unlikely that he might have been reporting upon other early experiments that he had seen, in this case, on fining cast iron. The “Liber Claritatis” is a 13th century collection of recipes—said to be by Geber, and which included some from the “Book of Alums” by al-Razi, as well as many new recipes. A number of these (from 161 in total) were tested by the editor of the manuscript, Ernst Darmstädter,41 who suggested that some of the new products that might have been obtained included iron and silver chlorides but whether they had been recognised as such by the author is doubtful.42 However, the contents do include making various precious stones, “microcosmic salt” (ammonium hydrogen phosphate) from urine, alcohol by distillation, “corrosive sublimate” (mercury chloride HgCl2) and of particular interest to historians of metallurgy, a knowledge of cast iron.43 This recipe may be quoted: Et scias quod fundit ferrum et omnia corpora. “And you know how to cast iron and other bodies.”

Chapter 66: to make Red glass. Glass is melted and sal gemma cast upon it but neither copper nor any reddening agent seems to be mentioned. The melting-point of cast iron (1150ºC) is not far above the free-running temperature of glass at about 1000ºC (not a sharp melting-point as it

40 Theophilus, op. cit. 183. 41 Darmstädter, E. “Liber claritatis totius alkimicae artis” is split into seven parts; the journal (although with the same editor) changes its title: Archivio di Storia della Scienza, 6 (Rome 1925) 319–330, and 7 (1926) 257–266; to Archeion, 8 (1927) 95–103 & 214–226; 9 (1928) 63–80, 91–208 & 462–482. 42 Multhauf, R. “The Origins Of Chemistry” (1966, London, reprinted 1993 ) 170. 43 Ruska, P. “Über die quellen des Liber claritatis” Archeion XVI (1934) 145–67. Ruska has shown that this recipe book was later than Michael Scot (1220) but earlier than Geber.

114

chapter seven

Fig. 4. A drawing from the Codex Oldanis (Adapted from Carbonelli, op. cit. fig. 151).

is not a crystalline solid) so that it is possible that craftsmen familiar with working glass might have also been familiar with cast iron. A 15th century treatise, discussed by Carbonelli44 is the Codex Oldanis, written between 1411 and 1463 by three brothers of the well-known 44 Carbonelli, G. “Sulle fonti storiche della chimica e dellʼalchimia in Italia” (Rome, 1925) chapter VI. (117–141) includes, inter alia, the treatises by Venetus Franciscus, p.178,

The Survival of Technology From the Ancient World

115

Milanese family Oldani, who later fell out of favour with the Sforza. Its title is Ars sive doctrina de transmutatione metallorum. Carbonelli called it “the oldest illustrated alchemical-technological treatise in Italianˮ.45 It includes this picture of an armourersʼ forge. The forge has an anvil with a “bick” (beak) such as an armourer would use for shaping his works. Above the forge there is a small bellows for kindling the fire, two U-shaped bars (formers ?) and a jug. To the left there are the large bellows for raising the temperature, and three small containers (for tempering liquids ?). Above the forge is the label: Ista est fucina cum quo fonduntur planeta et ad multa alia operatur etc. “This is the force whereby the planets (= 7 metals ?) may be melted and many other operations carried out.”

This accompanies a 15th century treatise Tratto della fusione del ferro e dell'accaio—which contains 12 recipes in total. Despite the name, none of these have any relevance to steelmaking, or the production of arms or armour. Many of them describe heating iron with a sulphide or oxide of arsenic; a low-melting-point (840ºC) iron arsenic alloy with a silvery appearance could be produced. However, what is significant is that a book of broadly alchemical recipes included such an illustration. It seems very plausible, especially considering the production of steel for swords and armour, and its successful heat-treatment, that the armourer/weaponsmith of the 14th/15th century was acquainted with experimental chemistry.

Regio de Bartholomeus Nicholaus Frater, p.155, and Penestrina de Julianus, p.140. Fractional distillation is shown on p.115 for making the “quintessenceˮ. 45 The Codex is in the University Library of Pavia (Aldini, No.74, Catalogue, L.de Marchi (1894), I, p.36) and see: Archivio di Storia della Scienza 6 (Rome 1925) 245.

116

chapter eight

Chapter eight

Viking-Age Swords and Their Inscriptions Pattern-welded swords made up of numerous pieces of iron and steel were gradually displaced in Europe by swords made of fewer, larger, pieces, or eventually a single piece, of steel. But many swords found in and around Scandinavia show the use of a different kind of steel, which does not seem to appear there after the 11th century, and in the rest of Europe not at all. Its use is related to the inscriptions which appear on many Viking swords, and which have long been thought to be makers’ marks. The best known of these was “Ulfberht”. Early Analyses of Viking Swords Some very early quantitative analyses of Viking-age swords have been undertaken. Lorange1 took the view that the quality of a sword depended on its carbon content, so he had three pattern-welded blades from Norway analysed (by T.N. Holme at Lillehammer) and their carbon contents were found to be 0.414%C, 0.401%C, and 0.520%C. These did not include any “Ulfberht” swords. Some years later, a number of swords were analysed for Petersen2 by the engineer K. Refsaas, of Trondheim. He set about an extensive series of analyses of 9 swords from the Oslo University collection. Due to the condition of the swords he had to be content with samples of 1g weight, which were analysed by direct combustion in a stream of oxygen. He was aware of the possible difference in composition between edge and blade, and attempted to carry out 2 analyses, wherever possible. The highest carbon content (0.75%C) of those tested was 4690, a sword from Aker, Hedemarken, with an VLFBERH+T inscription. This has been examined again more recently (see below). A large number of analyses (76) of early iron weapons and tools found in Norway were undertaken by Dannevig Hauge in the years before and 1 Lorange, A.L. 1889 “Den yngre jernalders svaerd” (Bergen, 1889). 2 Petersen, J. “De norske vikingesverd” (Kristiana, 1919).

Viking-Age Swords and Their Inscriptions

117

during the Second World War, and the results published in 1946.3 The analyses were carried out by combustion in oxygen, and melting- temperatures of samples (3–5g) were also determined. [It should be observed that the presence of high phosphorus contents might also have altered their melting temperatures.] This monograph seems to have been little noticed until recent years although the results are most significant.4 Twelve swords were analysed and found to have surprisingly high carbon contents, one being made of a 1%C steel. Photomicrographs of one of these swords were included and show the microstructure of a hypereutectoid steel—very fine pearlite and cementite globules. Despite the work of Hauge, until recently, the use of hypereutectoid steels in Early Medieval Europe does not seem to have been widely recognised. Some years ago, this author analysed a broken sword, bearing an “Ulfbehrt” inscription, in the stores of the Württemburg Landesmuseum, and found, to his surprise, that it was made of a hypereutectoid steel, of perhaps 1.4 %C (see below). “Ulfberht” Swords There are around 100 swords with ‘Ulfberht’, or variants of this name, inlaid into the blade. These have been found scattered all over Northern Europe. The largest concentration is in Scandinavia and the Baltic Sea, although it has been suggested that if Ulfbehrt was their maker then on linguistic grounds the source of their manufacture should lie in the Rhineland.5 From the different forms of these swords and their hilts, Ulfbehrt would have been active for 300 years, so it has been suggested that perhaps this was a family of smiths rather than an individual, or the name was a trade mark of some sort. The analyses of samples from some of these swords are presented below, together with some swords bearing different inscriptions (and a couple with no inscriptions). The results are here divided into five groups, of decreasing quality, according to their carbon contents:

3 Hauge, T.D. “Blesterbruk og myrjern” (Oslo, 1946) Tab 9, 179–182. 4 I am grateful to the late Prof.Maréchal for drawing my attention to this monograph. 5 Müller-Wille, M., 1970. “Eine neue Ulfbehrt –schwert”. Offa, 27: 65–91.

118 Group I Group II Group III Group IV Group V

chapter eight hypereutectoid steels (more than 0.8 %C) eutectoid steels (around 0.8 %C) hardened steel (generally around 0.4% C) edges on an iron core unhardened steel (generally around 0.4% C) edges on an iron core iron blades (less than 0.2 %C) Inscriptions

The metallurgical results may be compared with the swords’ inscriptions as follows: Group I 1 + V L F B E R H+T 2 + V L F B E R H+T 3 + V L F B E R H+T 4 + V L F B E R H+T 5 + V L F B E R H+T 6 + V L F B E R H+ 7 ……………….… H+T 8 V L F….. E R H+T 9 + V L F B E R H+T 9 swords Group II 1 V L F B E R H+T (hypereutectoid in places) 2 + V L F B E R H+T 3 + V L F B E R H+T 4 ……. F B E R H+T 5 + V L F B E R H+ 5 swords Group III: 1 +V L F B E R H T+ 2 +V L F B E R H T+ 3 ………………. R H T+ 4 V L F B E R N + T 5 + V L F B E R H +┴ 6 + V L F B E H + T 7 V L F P I R^ ^

Viking-Age Swords and Their Inscriptions 8 + V┌ ╘ B E R H T C  with another maker’s name, INGEFLRII (sic) 9 V .L… B….……… T 10 …..…… B E…. + 11 V…… B E.. H….. 12 ……………….…… 13 HARTO…FER 14 ……………….… 14 swords Group IV 1 +…… F B E …H…+.. 2 V L F B E R + + 3 + V L F B E R H T+ 4 + V L ..… E… H… + (+ Latin crosses) 5 + V L F B E R H C T 6 + V L F B E R H++ 7 …..F………T 8 H ┌ X I N T ┼ 9 + I I E.............. 10 ……………B R T 11 + V L F B E R H + 12 R E X and C O N S T A N IIE N S 13 H K I ΛI 14 F N H I ┘I T 15 V L F B E P H T 16 + I…F B E R H + T 16 swords Group V 1 + V L Г B E R H+ T 2 .…F.. V … 3 V L F B ……H +.. 4 + V I ┐ I F R I + ┴ 5 ....R T … 6 ┤ │┌ │ ┬ ┬ ┘┴ ┼ 7 8 9 10

+ V L E H B A H L N++ + V L Γ P… + M├ B E RIT + ………… B……..T… 11 swords

119

120

chapter eight

Groups I and II are clearly distinct from the others. They are made in part or in whole from steels which are much higher in carbon content (and lower in slag content) and which therefore would have been very serviceable swords. Their maker’s name is spelled +V L F B E R H+T and no hypereutectoid steels are found in any of the swords with a variant spelling, so it is evident that these were the originals. They are of the highest quality, and their starting material seems to have been a very unusual raw material, which could have been an imported crucible steel. Crucible steelmaking would usually have produced a hypereutectoid steel, if the contents were completely melted, although some of the speci­ mens show a variation in carbon content that suggests an imperfectly melted steel—one that remained in the “slushy” range, in fact. Some eutectoid steels may well have come from a crucible process also; there seems to be no other reason for their low slag contents. A hypereutectoid steel that retained cementite needles in the microstructure would have been unacceptably brittle, and some annealing was evidently called for. The sword from Stuttgart had no annealing, while the blade from Solingen had too much. Even so, these swords would have been better than the alternatives, and the best ones like Helsinki 9164.3 or Hamburg 1965/124 would have fetched much higher prices, and so the incentive to counterfeit them would have been considerable. The original maker of the “Ulfberht” swords was evidently a craftsman (or perhaps a craftsman/merchant) who had access to a source of highcarbon steel. This may have been ingots of crucible steel imported from the Middle East via the River Volga. In which case, his location was probably in the Baltic area, where this trade route terminated, and where most of these swords have been found (see map at end of this Chapter). The presence of primary graphite as well as a cementite network in the microstructure of the sword from Bergen 882 may have been a relic from the manufacturing process. That as described by al-Biruni (973–1048), involved heating cast iron with bloomery iron in a covered crucible for a matter of days. Eventually enough carbon would have been absorbed for the alloy to melt, and the broken crucible would yield a cake of cast steel, a convenient size for making a sword blade. The product he described was made around Herat and exported via North India to Persia & other

Viking-Age Swords and Their Inscriptions

121

Muslim lands. The Persians traded in crucible steel,6 and there was a wellestablished trade route from the Baltic to Persia via the Volga, exploited by the Vikings in the 9th–10th centuries, during the period of these swords’ manufacture. There are said to be more Samanid (815–1005) silver coins from their Afghan mines in Sweden than there are in Persia. After the fall of the Samanids, and the rise of the various Russian principalities, the use of this trade route by the Vikings declined.7 It is notable that, at this time, the manufacture of these “Ulfberht” swords apparently ceases, presumably because the raw material is no longer available. Such a high-carbon steel would have needed to have been forged, counter-intuitively, at a lower temperature than customary. If forged at the correct temperature and for the correct time, the swords produced would have been both hard and tough, and would have been highly valued. So it is not surprising that many other swordsmiths tried to copy these swords, and also copy their maker’s inscription. One possible way would have been by welding small pieces of bloomery steel onto a billet of iron, and forging that into a blade before quenching it. The sharp edge that could be formed might well fool the less discerning customer, but with a depth of only a few millimetres it would not have survived many sharpenings. Many workshops seem to have followed this method, which was already well-known to smiths, and their products make up Group III. One workshop in particular has the maker’s name identical apart from the placing of the terminal T. (+V L F B E R H T +) The close similarity of the edge hardnesses found in the swords having that particular variant spelling suggest that they may well have come from the same workshop, which was probably one of many endeavouring to copy the high-carbon blades of the original “Ulfberht” workshop. Unless a complete section or half-section of a blade could be examined, which was not always the case, it is not possible to say whether separate steel edges were welded on, or whether a bar of iron was simply carburised. In those cases where it was possible, welded-on steel edges greatly outnumber the merely carburised ones. (Only one such is found in Group III, no.13, and perhaps one in Group IV, no.7.) 6 Lang, J., Craddock, P.T. and Simpson, St J., “New evidence for early crucible steel” Historical Metallurgy 32 (1998) 7–14. and Craddock, P. & Lang, J. “Mining & Metal production through the ages” (2003) 231–257. 7 Mitchiner, M. “Evidence for Viking-Islamic trade provided by Samanid silver coinage” East and West, 37, (Rome, 1987) 139–150.

122

chapter eight

Group IV is the group of swords with (unhardened) steel edges. Where possible, the blade was sampled both on an edge and in the centre. When a sample could be taken only from the edge, one cannot be certain that the body was iron, but in some cases the welding lines were visible, and in other cases, a judgement was made on the basis of the homogeneity of the steel. Group V is made up of swords largely made of iron, or low-carbon steels. These swords were of the lowest metallurgical quality, and also generally show the greatest variations in spellings. As the quality of the metal declines, so does the literacy of the inscriptions—a situation not unknown to purchasers of counterfeit goods today. A number of the swords analysed here were from cremation sites, so the blades had, in effect, been annealed. This may cause some surface decarburisation, depending on the surroundings. It will also temper any martensite or bainite present into globules of cementite within a ferrite matrix, and the lamellae of pearlite will similarly form globules. If the annealing were to have been complete, then it would not be possible to say whether a martensitic or pearlitic microstructure had been the antecedent, but frequently some traces do survive, and it is possible to deduce whether the sword had been originally hardened, or not. The inscriptions given are as read by the author, unless another attribution is quoted. Many of the swords in Norway were illustrated by Lorange (1889) but not all the letters he recorded seem to be visible now. Group I Swords Containing Hypereutectoid Steels 1. Before the First World War, an “Ulfberht” sword 95.5cm.long with the inscription +VLFBERH+T was dredged from the Alster and given to the Museum for Hamburg History (M.1152, Type X, 11th century). This was published, with an illustration, by Schwietering in 19158 but then later an analysis (by one Dr.Schindler) was published in a Festschrift.9 It was found to have a carbon content of 1.2%. No other data was published, and the very high carbon content of a hypereutectoid steel was not remarked upon. Yet another “Ulfberht” sword, with a broken blade, found in the Elbe in 1957, is now in the same museum, and was published in 1970 by Müller-Wille. This also proved to be a hypereutectoid steel (see below). 8 Schwietering, J. “Ein Ulfberhtschwert des 11 Jahrhunderts” Zeitschrift für Historische Waffen- und Kostümkunde, 7, (Berlin, 1915) 107–8. 9 Jankuhn, H. “Festschrift für G. Schwantes”, (Hamburg, 1951) 212–229. Analysis of the Ulfberht sword is given on p. 224.

Viking-Age Swords and Their Inscriptions

123

Fig. 1.

2. Württemburg Landesmuseum (WLM) Stuttgart, inv.no.1973–70 This was acquired at auction in 1973, having been found in the river at Karlsruhe, with the inscription +VLFBERH+T . The microstructure taken from a half-section at the broken end, consisted of pearlite and cementite, some at grain-boundaries, and some in needle-like form. It is now evident that this was made from a billet of crucible steel (hypereutectoid )which had been folded twice and forged out into a blade (Williams, 1977). The presence of these cementite needles may have led to a certain brittleness in the blade, as its end had been broken off. This bore no resemblance to the microstructure of any other contemporary Western European sword so far observed, but Paul Crad­ dock later suggested that the blade was made from a crucible steel which had undergone sufficient hot-work to obliterate traces of dendrites (although cementite needles are still present mixed with areas of pearlite); and indeed as the section suggests it was folded from a billet in the course of forging out the sword.

Fig. 2. Sword I.2 from Stuttgart.

Fig. 3. Inscription on I.2.

124

chapter eight

Fig. 4. Microstructure of I.2: pearlite and cementite (scale bar 50 microns).

Fig. 5. Microstructure of 1.2: pearlite areas and cementite as network and as needles (scale bar 10 microns).

Viking-Age Swords and Their Inscriptions

125

3. Solingen, Deutsche Klingenmuseum; inv.no.1973.w.5. It has an inscription which may be read as +VLFBERH+T A sample taken from the damaged edge shows a microstructure of what seems to have been pearlite which has been mostly divorced into carbide particles, and a network of particles outlining the prior pearlitic areas. There are very few slag inclusions.10 Microhardness 243–277; average = 258 VPH. This seems to be a steel which has been annealed, or undergone an excessive amount of hot-working. Since it seems unlikely that any sword would be intentionally softened by annealing, one may speculate that this was another hypereutectoid steel but it was somewhat overheated in working.

Fig. 6. Microstructure of I.3: pearlite areas with network of carbide particles (scale bar 50 microns).

4. Museum für Hamburg Geschichte; inv.no.1965/124 Inscription +VLFBERH+T. Another river find, from the Elbe. 10 Williams, A. “Crucible steel in medieval swords” Metals and Mines ed. LaNiece, S. Hook, D. Craddock, P. (2007) 233–241.

126

chapter eight

Fig. 7. Sword I.4 from Hamburg.

A sample taken from the damaged edge about 10cm from the end, shows a microstructure of mostly fine pearlite with some cementite at grain boundaries, but no visible slag inclusions. Another taken from near the centre, at the broken tip, shows a microstructure of very fine pearlite. This is a steel of perhaps 1% carbon, or more, and so it is almost certainly a hypereutectoid crucible steel.  Microhardness centre 337–388; average = 355 VPH. edge 439–476; average = 463 VPH. It seems that this blade underwent more hot-working than sword # 2, and perhaps had a lower carbon content, for no cementite was observed in the form of needles within the prior austenite grains, but only at grain boundaries, or as laths, in a more equiaxed form. By contrast, the sword # 3 seems to have undergone a little too much hot working. Electron microanalysis suggests that the inclusions are iron oxide.

Fig. 8. Microstructure of I.4: fine pearlite, cementite and very little slag (scale bar 10 microns).

Viking-Age Swords and Their Inscriptions

127

5. Oslo Historisk Museum c.4690 It has an inscription which may be read as + VI F B E R H + T An analysis of samples from this sword (by Refsaas) was published in Petersen (1919). The cavities left by his sampling are still visible on the surface of the blade. They were taken from the middle of the blade, which might explain why the carbon content in the published analysis (0.75%C) does not tally with that expected from the microstructure of the edge. A sample was taken from the edge, at the broken end of the sword tip. The microstructure consists mostly of pearlite, in areas surrounded with a network of cementite. In places, this cementite appears as laths within the areas of pearlite; in other places, the network has broken up into globules. There are a few slag inclusions. In parts, away from the edge, there are areas of ferrite also. These seem to become more frequent as one moves towards the areas from where the samples were taken. This appears to be a hypereuctectoid steel, in places, which has undergone considerable hot-working (or even perhaps a deliberate anneal to soften it) which has caused the cementite to spheroidise in some parts. This may also have led to surface decarburisation, deliberately or accidentally. This may explain why the carbon content, according to combustion analysis of samples away from the edge, is only 0.75%C; but away from the surface, according to the microstructure, it is well over 1.0%C, so this sword has been put in Group 1.

Fig. 9. Sword I.5: the detached end with earlier sample drillings may be observed.

128

chapter eight

Fig. 10. Microstructure of sword I .5: note ferrite grains near surface (scale bar = 50 microns).

Fig. 11. Microstructure of sword I.5 interior: pearlite and cementite (scale bar = 50 microns).

Viking-Age Swords and Their Inscriptions

Fig. 12. Microstructure of sword I.5 interior: pearlite and a cementite network (scale bar = 10 microns).

Fig. 13. Microstructure of sword I.5 interior: pearlite and cementite needles (scale bar = 10 microns).

129

130

chapter eight

6. Bergen Historisk Museum 882 The blade is in excavated condition and has neither hilt nor pommel. It has an inscription which may be read as V L F B E R H +… Three specimens were taken; (i) A specimen from the body of the sword was examined. The microstructure consists of areas of fine pearlite, with a network of cementite, and no visible slag inclusions. This is a hypereutectoid steel of carbon content at least 1%. Microhardness range 253–308 average = 282 VPH.

Fig. 14. Microstructure of I.6 body: pearlite and cementite (scale bar 50 microns).

(ii) Two smaller flakes from the edge were also examined. The microstructure of one consists of areas of fine pearlite, with some ferrite grains near the surface, and no visible slag. The microstructure of the other consists of areas of fine pearlite, with a network of cementite and some ferrite grains near the surface. There are also some unusual features, which on repolishing and re-examination at a higher magnification, turned out to be cavi-

Viking-Age Swords and Their Inscriptions

131

ties, some of which still contain graphite flakes, and some of which are empty. In the surrounding area, the microhardness is higher there. This would appear to be the relic of an area of very high carbon content, which has contained primary graphite. This is strongly suggestive of the method of making crucible steel described by al-Biruni (see Chapter 3). Microhardness range 226–279 VPH.

Fig. 15. Microstructure of I.6 body: pearlite and graphite (scale bar 10 microns). Local decarburisation near the surface might have occurred during forging.

(iii) A sample was also taken from an inlaid letter. The microstructure of this consists of ferrite (in large equiaxed grains) and very elongated slag inclusions only. This suggests that the inlay was applied hot during forging, rather than hammered into the blade afterwards when cold.

Fig. 16. Microstructure of I.6 inlay: ferrite and slag (scale bar 100 microns).

132

chapter eight

7. Bergen Historisk Museum 1483 This sword is also in excavated condition. Only part of the inscription can now be read; …H + T

Fig. 17. Microstructure of a sample from the edge of sword I.7: pearlite and cementite (scale bar = 50 microns).

Fig. 18. Microstructure of sword I.7: pearlite and cementite (scale bar = 10 microns).

The microstructure consists of areas of fine pearlite, with a network of cementite, and only very few slag inclusions. This is a hypereutectoid steel of carbon content at least 1%. Microhardness range 295–360; average = 327 VPH.

Viking-Age Swords and Their Inscriptions

133

8. Helsinki Kansali Museum 9164:3

Fig. 19. 

A sword with an inscription which may be read on one side as +VL F …E R H +T  the character between F and E is obscure; and on the other side │││ X │││ where X is within a hexagonal outline11

Fig. 20. Microstructure of sword I.8: very fine (almost irresolvable) pearlite (scale bar = 50 microns).

The microstructure shows large areas of pearlite, of a feathery appearance, and almost irresolvable. There are no visible slag inclusions, nor areas of ferrite. Microhardness range 417–476; average = 447 VPH. The absence of slag inclusions may be significant. If this is another hypereutectoid steel, that has undergone some form of accelerated cooling, to form irresolvable pearlite (or perhaps even upper bainite) then the absence of visible proeutectoid cementite might be explicable. 11 Leppäaho, J. “Späteisenzeitliche waffen aus Finland” (Helsinki, 1965) 12. This book of illustrations of blades was posthumously published as Finska Fornminnesföreningens Tidskrift, 61.

134

chapter eight

Samuels12 shows a photomicrograph of a 1.0 %C steel which has been isothermally transformed at 450ºC for 1 minute to produce a fine pearlite/ bainite microstructure with a hardness of 425 VPH. A similar microstructure is also shown in some 17th–18th century Indian swords from Hyderabad;13 one specimen showed very fine pearlite of 370 VPH. This is presumed to have been a much later crucible steel. 9. A sword from a Private Collection It has an inscription (which is unusually clear, perhaps due to over-cleaning in the past) which may be read as +VLFBERH+T.

Fig. 21. Sword I.9.

A sample taken from the upper edge has the microstructure of an unusual mixture of areas of ferrite and carbides, which resembles somewhat an overtempered martensite, a large lump of iron oxide, and an area of ferrite and pearlite with grains of cementite and small, ovoid, slag inclusions. This is very heterogeneous and may perhaps have been the result of attempting (but failing) to completely melt a crucible steel. The carbon content varies between around 0.8% and 1%C. Some of the pearlite areas are lamellar, and some are irresolvable. After forging, it may have been given some sort of accelerated cooling to harden it. The precise nature of the heat-treatment is difficult to discern, as an overtempered martensite will resemble a hot-worked crucible steel. Microhardness range 308–345; average = 321 VPH.

12 Samuels, L.E. “Light microscopy of carbon steels” (Materials Park, Ohio, 1999) 266–9. 13 “The metallurgy of some Indian swords from the Arsenal of Hyderabad and elsewhere” (Williams, A.and Edge, D.) Gladius, 27, (Madrid, 2007) 149–176.

Viking-Age Swords and Their Inscriptions

135

Fig. 22. Microstructure of sword I.9: pearlite and cementite (scale bar = 50 microns).

Another sword with edges of hypereutectoid steel has recently been analysed. It was found in a richly equipped 9th century grave at Stara Kouřim (Bohemia). No inscription has yet been detected. (Košta J., Hošek J. work in progress, 2010). Group II All-steel Swords with Eutectoid (0.8%C) or Hypoeutectoid (Lower than 0.8C%) Blades 1. Bergen Historisk Museum 3149 This sword fragment has an inscription which may be read as V L F B E R + H T or V L F B E R H + T Two specimens were taken: (i) from the edge The microstructure consists of areas of fine pearlite, with numerous small equiaxed grains of ferrite (?), and very few inclusions, which are not elongated, but there is no cementite network. This area is a medium-carbon steel of perhaps 0.5% overall. Local decarburisation must have occurred during forging. A low forging temperature might explain the small ferritic grain size. Microhardness range 220–274; average = 231VPH.

136

chapter eight

Fig. 23. Microstructure of sword II. 1 edge: pearlite and ferrite (scale bar = 50 microns).

(ii) from the body The microstructure consists of areas of fine pearlite, outlined with a network of cementite, and no visible slag inclusions. This part is a hypereutectoid steel of carbon content at least 1%, which has been subsequently annealed. Microhardness range 204–286 VPH. The lower carbon content of the edge may be the result of an imperfectly melted and homogenised crucible steel, or decarburisation due to excessive hot-working. 2. Helsinki Kansali Museum 2548:839 It has an inscription which may be read on one side as +VLFBERH+T and on the other side │││X │││ with X within an ovoid (as read by Leppäaho,14 although this does not now seem to be visible). The microstructure shows large areas of pearlite, lamellar in some places, but irresolvable in others. Some ferrite is present near one end of the sample as equiaxed grains, and in other places within the pearlite areas in a Widmanstätten formation. There are also a number of large, irregular, slag inclusions. This is a rather unusual microstructure, but it seems to be a steel of varying carbon content which has undergone a rapid air-cool. Microhardness range 236–286; average = 269 VPH.

14 Leppäaho, op. cit. 46.

Viking-Age Swords and Their Inscriptions

137

Fig. 24. Microstructure of II.2 pearlite, with varying amounts of ferrite, and slag (scale bar = 10 microns).

3. A sword from a Private collection On exhibition at the Metropolitan Museum of Art, New York (L.2006.57).

Fig. 25. Hemispherical pommel decorated with silver inlay—an unclear inscription found by subsequent X-radiography to read +VLFBERH+T —perhaps 1000 CE.

The microstructure consists of a very uniform mixture of very small grains of ferrite, mixed with areas of very fine pearlite irresolvable in place (and perhaps 0.5%C overall) with no visible inclusions. This is a remarkably uniform medium-carbon steel which has been quickly air-cooled after fabrication. It is possible that it is also a crucible steel in origin, since otherwise the absence of inclusions would be difficult to explain in a medieval steel. Microhardness range = 279–322; average = 297 VPH 4. Virumaa Museum, Rakvere, Estonia RM 587/A21 It bears an inscription which may now be read as (…...) F B E R H + T but was read as +VL F B E R H + T by Anteins,15 with the V and L joined. 15 Op. cit. 44.

138

chapter eight

Fig. 26. Microstructure of II. 3.: pearlite and ferrite (scale bar = 10 microns).

Fig. 27. Sword II.4. The inlaid letters are formed of pattern-welded strips. The blade may have been annealed, and there is a fracture (the result of a cremation ?) half way down the blade. Surface hardness of the blade was 160–320 VPH, the pommel was 130–150 VPH

Two samples were studied: (i) a sample from the edge 15 cm from the hilt. The microstructure consists largely of areas of pearlite with some ferrite grains, and very few slag inclusions. The carbon content is around 0.6%–0.8%.

  Fig. 28. Microstructure of II.4: pearlite and some small, elongated, slag inclusions (scale bar = 50 microns).

Viking-Age Swords and Their Inscriptions

139

(ii) a half-section (15mm across) detached where the blade was already fractured.

Fig. 29. Half-section of II.4. 

The microstructure here contains an area with more ferrite near the surface than in the centre. There does not seem to be any trace of a weld, although there are a number of small, elongated slag inclusions. Otherwise it is a fairly uniform pearlitic-ferritic steel, with carbon content varying between 0.4% and 0.9%. The pearlite is broken up in a way characteristic of annealing. Cremation may have caused some surface decarburisation, but only within a fairly narrow layer. The hardness ranges from 150 to 200 VPH. 5. Saaremaa Museum, Estonia K85–120 The pommel of this sword is incomplete. A hemispherical element seems to be missing from the end.16 Length 68 cm, with traces of a crack 10cm from the end. The inscription may be read as + VL F B E R H + (T) with the initial V & L joined. Anteins17 read the inscription as +VLFBERHI .

Fig. 30. The inscription on sword II.5. 16 Ebert, M. (op.cit, 1914) illustrates a sword (without an inventory number, then in the Arensburger Museum, Oesel) which could be this one, if the end of its pommel has since been lost. The inscription is recorded there as +VLFBERH+ without a final T. 17 Op. cit. 44.

140

chapter eight

A sample was detached from the edge 20cm from the tip. The microstructure consists of a very uniform mixture of small grains of ferrite and areas of divorced pearlite, with very few slag inclusions. The carbon content varies between 0.3 % and 0.5 %.

Fig. 31. Microstructure of II.5: pearlite and ferrite, with very few slag inclusions.

This unusual characteristic is the reason for including this sword in Group II despite its hypereutectoid carbon content. Microhardness; range 189–217; average = 205 VPH. Group III Swords with Hardened Steel Edges 1. Olomouc Regional Museum, Czech Republic A sword from a 10th century grave at Nemilany, excavated without hilt or pommel, but with an inscription +VLFBERHT+.

Viking-Age Swords and Their Inscriptions

141

Fig. 32. Sword III. 1 Inscription

It was examined by the staff of the Conservation Laboratory of the Brno Technical Museum.18 It has a core of low carbon content (< 235 VPH) and welded-on steel cutting edges, which were hardened to 320–460 VPH by quenching . 2. Helsinki Kansali Museo 18402:1c +V L F B E R H T+ Average surface hardness = 210–220 (tang 110) edge 310–390 VPH.

Fig. 33. Sword III.2.

The microstructure of a sample from the edge consists of uniform tempered martensite, with a few slag inclusions. There appear to be no separate grains of ferrite, nor areas of pearlite.

Fig. 34. Microstructure of sword III.2 Tempered martensite (scale bar = 50 microns). 18 Selucká, A. Richtrová, A. & Hložek, M. 2001 “Konservace železného meče Ulfberht” Sbornik z konservátorského a restaurátorského semináře, České Budějovice, 65–68 and 103–4.

142

chapter eight

Microhardness range 388–476; average = 437 VPH. The low surface hardness of the body suggests that this was made of iron, which will not harden on quenching. 3. A sword from a Private Collection: Possibly 9th century. A sword with the inlay of pattern-welded metal forming the letters .....RHT+. 

Fig. 35. Sword III. 3.

Two samples were taken. One sample had only ferrite and some circular slag inclusions. The other sample (taken from nearer the edge) had a microstructure of a fairly uniform mixture of areas of nodular pearlite mixed with lightly tempered martensite (perhaps 0.5%C) and some small grey slag inclusions. This is a medium-carbon steel which has been hardened by slackquenching. It may well be that only the cutting edges were made of steel.

Fig. 36. Microstructure of sword III.3: tempered martensite and nodular pearlite (scale bar = 50 microns).

Viking-Age Swords and Their Inscriptions

143

Fig. 37. Microstructure of sword III.3: tempered martensite and nodular pearlite (scale bar = 10 microns).

Microhardness range = 264 (pearlite)—564 (martensite); average = 378 VPH 4. A sword in a Private British Collection19

This had a very worn inscription which might be read as VLFBERN+T.

Fig. 38. Sword III.4.

19 Edge, D. & Williams, A 2003 “Some early medieval swords in the Wallace Collection and elsewhere” Gladius, 23, Madrid, 191–210.

144

chapter eight

Fig. 39. Detail of inscription.

A sample was taken from the broken end of the blade. The microstructure consists largely of ferrite with a layer of martensite corresponding to a cutting edge which is approximately 6mm deep. Evidently a thin layer of (perhaps 0.5%C) steel was wrapped around a much larger piece of iron, or low-carbon steel, and attached by forge-welding. This composite billet was then forged out into a sword-blade and hardened by quenching. This gave a hard (microhardness 423–540, average = 467 VPH) edge on a much less hard core (236 VPH). The similarity of the edge-hardnesses attained in the first four swords from this Group is noticeable. 5. A sword found in the Danube at Abbsbach and now in a Private Collection20

Fig. 40. It has the inscription +VLFBERH+┴ with the final T inverted.

20 Szameit, E. “Ein VLFBERHT-Schwert aus der Donau bei Aggsbach, Niederösterreich” Archaeologia Austriaca, 76 (Vienna, 1992) 215–221.

Viking-Age Swords and Their Inscriptions

145

Fig. 41. Microstructure of sword III. 5: acicular martensite (?) and bainite (scale bar = 10 microns).

A sample taken from the upper edge (near the end of the inscription) has a microstructure which consists of areas of carbides, which have a feathery appearance in places (probably bainite). No free ferrite or pearlite is visible. This seems to be a low/medium-carbon steel which has been given an accelerated cooling to harden it. There is no way of knowing whether this is an all-steel blade, or whether only the edges are steel. Microhardness range 295–322; average = 307 VPH. 6. Germanisches National Museum, Nürnberg FG.2187 A sword (from Mannheim) with the inscription +VLFBEH+T

Fig. 42. 

146

chapter eight

A sample was taken from the middle of the broken end. The microstructure consists of very fine pearlite, irresolvable in places, with a little proeutectoid ferrite, spiny in places. There are very few slag inclusions. This is a medium-carbon steel (around 0.7%–0.8%C) which has been given an accelerated cooling after forging, such as for example, a lead quench. Microhardness range 286–345; average = 310 VPH

Fig. 43. Microstructure of sword III. 6 (scale bar = 50 microns).

Fig. 44. Microstructure of sword III. 6: pearlite, irresolvable in places, and spiny ferrite (scale bar = 10 microns).

Viking-Age Swords and Their Inscriptions

147

7. Institute of Archaeological Science of the Academy of Sciences of the Czech Republic, Prague A sword from the richly furnished grave 43821 at the Great Moravian burial site of the 9th century at Mikulčice had an almost illegible inscription which might be read as / | Γ Γ | Λ ^ ^ (or perhaps to the eye of faith, VLFPIR^ ^ ) . The blade was made up of at least six pieces of steel (some of which had themselves been folded) forge-welded together, and then slackquenched to harden it. The microstructure shows a piled structure showing bainite, pearlite and ferrite, as well as iron silicate inclusions. Microhardness varied between 205 VPH and 476 VPH at the (bainitic) edge.

Fig. 45. Section halfway along the blade. Section near to the hilt. (Metallography by Dr. J. Hosek).

8. Wisbech & Fenland Museum; inv.no.1860.5 A find from the River Nene. This sword has the inscription +VΓ╘BERHTC+ (with L and F inverted) on one side and another inscription INGEFLRII, on the other side. A sample was detached from a damaged edge, at about 8cm from the tip. The microstructure consists of tempered martensite, without free ferrite or pearlite visible. Microhardness range 329–379; average = 356 VPH 21 Hošek J. Košta J. Barta, P. “The metallographic examination of Sword No.438” (2010, in press): This summarises in English the results of the other swords excavated from this site, but published in Czech.

148

chapter eight

Fig. 46. Microstructure of III. 8: tempered martensite and bainite (?) (scale bar = 50 microns).

Fig. 47. (scale bar = 10 microns).

Viking-Age Swords and Their Inscriptions

149

This sword has been described and illustrated recently.22 It is a sword whose steel edges have been hardened by a successful heat-treatment. It is not possible without sampling the interior of the blade, to know whether the body of the sword was made of iron, with steel edges welded on, or whether it was all made of steel. 9. Helsinki Kansali Museo 6066:1 A broken sword, excavated without hilt or pommel. It has an inscription on one side V….…T (the intermediate letters might include L and B) and on the other side │││ XX │││, where the two X are interlaced (reading of Leppäaho, 44). Average surface hardness = 330- 380 VPH The microstructure shows a mixture of ferrite and martensite, in bands, without any visible pearlite, and a few slag inclusions. This sword had a steel edge, possibly welded onto an iron core, although there is no row of slag inclusions along the boundary between the ferrite and martensite. The sword was subsequently hardened by a full quench. Microhardness range 261–429; average = 350 VPH.

Fig. 48. Microstructure of sword III. 9: tempered martensite and ferrite (scale bar = 50 microns). 22 Gorman, M.R. 1999 “Ulfberht: innovation and imitation in early Medieval Swords” The XVIth Park Lane Arms Fair Catalogue, 7–12.

150

chapter eight

Electron microanalysis The metal matrix is virtually pure iron. There are some inclusions of iron silicate (with Mn) and some of iron silicate (without Mn). 10. Saaremaa Museum, Estonia K85–108 Three VLFBERHT swords which all came from the island of Saaremaa, and which were stored in the Institute of History, Tallinn University, are now back in the Saaremaa Museum, housed in Kuressaare Castle.23 Traces of silver are left on the hilt and pommel. The inscription is very unclear, but the letters (……. ) B E (.…)+ may be discerned. Anteins24 read this as +VLFBERH+

Fig. 49. Sword III. 10. The surface hardness was 170–220 VPH in the centre.

Two samples were detached: (i) from the broken end near the tip The microstructure consists of ferrite and slag only. This sword was evidently made with an iron core and steel edges.

Fig. 50. Microstructure of III. 10 body (scale bar = 50 microns). 23 Ebert, M. “Ein Schwert mit tauschierte Klinge von Lümmada auf Oesel” Baltische Studien zur Archäologie und Geschichte, (Berlin, 1914) 147–158. It should be observed that Oesel is now called Saaremaa. 24 Anteins, op. cit. 44

Viking-Age Swords and Their Inscriptions

151

(ii) from the edge, halfway down the blade (92 cm). The microstructure consists of a very uniform mixture of fine carbides, apparently divorced pearlite and an acicular material, probably bainite. There are no slag inclusions visible, nor separate grains of ferrite. This sword was made from steel edges probably welded onto an iron body, and then was given some sort of heat treatment to form a mixture of fine pearlite and bainite, which was finally over-tempered. Microhardness (edge); range 230–248; average = 240 VPH.

Fig. 51. Microstructure of III. 10. edge (scale bar = 50 microns).

Fig. 52. Microstructure of III. 10. edge (scale bar = 10 microns).

152

chapter eight

11. Another Sword from Olomouc This fragment of a late 10th or early 11th century sword comes from the city of Olomouc, and should not be confused with the +VLFBERHT+ sword which was discovered at Nemilany, 5 km from the centre of Olo­ mouc—published by Selucka et al. (see above). It bears an inscription on one side and a pattern-welded panel on the other side. The inscription is made up of letters which, might be read as (V)(B) EH although damage to the fragment makes the V and B doubtful. This sword has been analysed by Hošek,25 who found that the blade had cutting edges of steel, hardened to 545 VPH while the core was composed of a heterogeneous steel of 197 VPH on one side and the patternwelded panel on the other side. This panel was made up of high- and low- phosphorus layers, whose hardness reached 259 VPH in the high P% areas. Fragments of a “silver”inlay on the pommel were also analysed and found to consist of: Cu 14% Zn 1% Ag 82% Sn 1.4% Pb 1.6% One might also mention the so-called “sword of St. Stephen” in the treasury of Prague cathedral. It dates from the 10th century but the blade might be older. The inscription is I = B E R I T I. Unlike the +VLFBERH+T swords, the letters are inlaid with short pieces of pattern-welded billets.26 It has not yet been analysed. 12. A Viking-age Sword with a Silver-inlaid Hilt, in a Private Collection perhaps 8th or 9th century, with no inscription.

Fig. 53.  25 Hošek, J. 2007: Meč i.č. A94696—Olomouc. Metallographic report No. 11403/07, deposited in: archive of Institute of Archaeology CAS Prague, v.v.i. 26 Pleiner, R. “Staré evropské kovářství—Alteuropaisches Schmiedehandwerk”, (Praha 1962) 237, fig. 48; 168 gives other names of swordsmiths.

Viking-Age Swords and Their Inscriptions

153

A specimen was taken from the sword, near the edge. The microstructure consists of fairly uniform fine pearlite, irresolvable in places, with a few small elongated slag inclusions. There are also several large, lumpy, slag inclusions. Around the edges, in the vicinity of areas of corrosion products, there are layers containing ferritic grains as well as the pearlite. The carbon content of the edge is around 0.7%. The hardness of the pearlite suggests that it has undergone some form of accelerated cooling.

Fig. 54. Microstructure of the edge of sword: pearlite and irregular inclusions (scale bar = 50 microns).

The microhardness of the edge ranges from 311 to 357; average = 330 VPH, except away from the edge it ranges from 159–215 VPH suggesting a lower carbon content. 13. Sword from a Private Collection (on Loan to the Metropolitan Museum of Art, New York) Possibly 9th century It has an inscription which might be read as HARTO( ) FER

154

chapter eight

Fig. 55. Sword III. 13.

Fig. 56. Microstructure of sword III. 13 (scale bar = 50 microns).

The microstructure of a sample taken from the edge consists of a mixture of blocky ferrite and areas of untempered martensite. This appears to be a low-C% steel which has been quenched. The microhardness reaches 296 VPH in places. The microstructure of a sample taken from near the surface, as well as that of samples from the tang and pommel, consists of ferrite and slag; these parts are made of wrought iron. The blade has apparently been carburised (but only in a very thin layer) and then quenched (but not tempered) in an attempt to harden it. 14. Sword from a Private Collection

Fig. 57. Possibly 10th century. Formerly in the Gwynn Collection, now in another Private Collection.

Viking-Age Swords and Their Inscriptions

155

The microstructure consists of a mixture of areas of pearlite, lamellar in places, and irresolvable in others, with carbide particles arranged in a network, areas of irresolvable material with a generally acicular appearance, and a few slag inclusions. No separate areas of ferrite seem to be visible. Microhardness range 308–345; average = 324 VPH

Fig. 58. Microstructure of a sample from the edge of III. 14 (scale bar = 50 microns).

Fig. 59. Microstructure of a sample from the edge of III.14: pearlite and irresolvable carbides (scale bar = 10 microns).

156

chapter eight

This microstructure is that of a low/medium-carbon steel which has under­gone some form of accelerated cooling, resulting in a mixture of pear­lite and (possibly) bainite, which has then subsequently been re­­ heat­ed to temper it. Group IV Swords with Unhardened Steel Edges 1. Estonian History Museum, Tallinn A.580:2020 Sword excavated in 1985 from a cremation site at Maidla, Western Estonia. The sword had been rolled up into a coil.

Fig. 60. View of sword IV.1 rolled into a coil.

It had apparently been polished and etched by Antiens, but subsequent corrosion had obscured the inscription. However, it was repolished by the Conservation Department, and the inscription became partially visible again. Traces of copper appear on the inlaid letters. It may be read as (doubtful letters in brackets) + (L) F (B) E (P) H + (I) although it was read as VLFBERHT by Mandel.27 27 Mandel, M. “Typology and dating of Estonian swords of the 8th–13th centuries.” Muinasaja Teadus I (Tallinn, 1991) 126–130.

Viking-Age Swords and Their Inscriptions

157

Surface hardness 150–170 VPH. A sample was detached from the broken edge, 4cm from the end of the blade.

Fig. 61. Microstructure of IV.1.: pearlite and ferrite with a large corrosion crack (scale bar = 100 microns).

The microstructure consists of fine pearlite with some ferrite grains arranged in bands. The carbon content reaches between 0.6% and 0.8%. There is a large corrosion crack visible. Cremation may have caused some surface decarburisation, but only within a fairly narrow layer, despite its history. The microstructure is pearlitic rather than overtempered martensite. Microhardness ranges from 178 to 246; average = 197 VPH. 2. Bergen Historisk Museum 1165 Inscription V L F B E R + +

Fig. 62. Sword IV.2.

158

chapter eight

Fig. 63. Microstructure of sword IV.2 (scale bar = 50 microns).

The microstructure consists of uniform very fine pearlite, feathery in some areas, and irresolvable in others. No separate ferrite grains are visible, but there are a few slag inclusions. This is a steel of perhaps 0.5% C, which appears to have undergone a rapid air-cooling after forging.  Microhardness range 234–261; average = 250 VPH. 3. Helsinki Kansali Museo 10390:2 This is a sword fragment without crosshilt or pommel. With an inscription on one side +VLFBERHT+28

Fig. 64. Microstructure of sword IV.3: ferrite and pearlite (scale bar = 50 microns). 28 Leppäaho, op. cit. 42.

Viking-Age Swords and Their Inscriptions

159

The microstructure shows a mixture of pearlite areas and ferrite grains both as a network and within the pearlitic areas. The latter grains are equiaxed rather than spiny in their formation. It is difficult to decide the previous heat-treatment, but this seems to be a steel which has undergone considerable subsequent hot-work­ing. The overall carbon content seems to be about 0.4 %. Micro­hardness range 116–223 VPH. 4. Saaremaa Museum, Estonia. K85–122 This is a blade without a crosshilt or pommel. It has been polished in the past, but only a few letters can now be discerned. + (NN) (…..) (P) E …H… P might be B and NN might be a joined VL. Anteins29 read the inscription as + VLBERHT + (and the + are Latin crosses and do not resemble the + on those swords described above) Three samples were detached: (i) the edge 1 cm from the tip

Fig. 65. Microstructure of specimen from tip of IV.4 (scale bar = 50 microns).

29 op. cit. 44

160

chapter eight

Fig. 66. Microstructure of specimen from tip of IV.4 (scale bar = 10 microns).

The microstructure consists of divorced pearlite in small areas with very little slag. There are no separate ferrite grains. Microhardness; range 234–266; average = 252 VPH. (ii) the edge 20cm from the tip

Fig. 67. Microstructure of specimen from edge of IV.4 (scale bar = 10 microns).

Viking-Age Swords and Their Inscriptions

161

The microstructure consists of slightly divorced fine pearlite with some isolated ferrite grains. Microhardness; range 210–234; average = 217 VPH. (iii) the rim of the central hole.

Fig. 68. Microstructure of specimen from centre of IV.4 (scale bar = 50 microns).

The microstructure consists of a mixture of ferrite and pearlite with some elongated slag inclusions. The carbon content is 0.4%–0.5%. Microhardness; range 137–185; average = 158 VPH. The higher carbon content of the edges (0.6%–0.8%) is probably in­ ten­tional, but the hardness is much less than one would expect. The spheroidisation of the pearlite in this steel, especially at the tip, suggests that it has been annealed, perhaps by plunging into a fire for some unknown ritual purpose.

162

chapter eight

5. Bergen Historisk Museum 2944 This sword was found damaged and bent somewhat at the end. It has an inscription [read as VLFBERHCT by Lorange].

Fig. 69. 

The microstructure of a specimen from the body consists of ferrite mixed with pearlite completely divorced to globules of cementite (iron carbide); there is some trace of a prior lamellar structure in places. This was probably a ferrite/pearlite steel, whose carbon content varied between perhaps 0.2% and 0.8%, before it was annealed.

Fig. 70. Microstructure of IV. 5.: ferrite and pearlite divorced to cementite (scale bar = 10 microns).

Microhardness range 163–289; average = 215 VPH

Viking-Age Swords and Their Inscriptions

163

6. Helsinki Kansali Museum 3601:2 A sword without crosshilt or pommel. It was found bent double when excavated. It has an inscription on one side +V L F B E R H ++ and on the other side X within a lozenge.30 The microstructure shows a mixture of ferrite and globular carbides. At higher magnification, it is apparently fairly uniform pearlite which has been almost completely divorced. There are also a few slag inclusions.

Fig. 71. Microstructure of IV. 6: ferrite and pearlite divorced to cementite (scale bar 10 microns).

This seems to be a medium-carbon steel (perhaps 0.5% -0.6% C) which has undergone a prolonged annealing. It view of the fact that the sword was found bent double, this would seem to be the explanation for such an unsuitable microstructure. Traces of a prior lamellar arrangement suggest that the original microstructure may have been pearlitic rather than martensitic, but this cannot now be confirmed. Microhardness range 195–217; average = 208 VPH. 7. Bergen Historisk Museum 2695 An excavated blade broken into three fragments. It had an inscription of which F….T can still be discerned. 30 Leppäaho, op. cit. 46.

164

chapter eight

Fig. 72. Microstructure of IV.7: pearlite and ferrite (scale bar 50 microns).

A specimen was taken from the edge of the blade. The microstructure shows mostly fine pearlite with some ferrite grains in a network, and some within the pearlite areas. The overall carbon content is perhaps 0.6% %. Microhardness (edge) range 175–243 VPH. A specimen was also taken from the centre of the blade. The microstructure (not shown) of this specimen consists of equiaxed ferrite grains and areas of pearlite as well as some fairly large slag inclusions. The overall carbon content is around 0.4%C. Although no attempt has been made to harden the blade, the higher carbon content of the edge is probably intentional. Microhardness (centre) range 160–215 VPH. 8. Bergen Historisk Museum 1069 Another fragmentary blade with an inscription H ┌ X I N T ┼ (broken-up letters which perhaps were a copy of Ulfberht). A sample was taken from the edge. The microstructure shows mostly pearlite with a few ferrite grains, and very few slag inclusions. The overall carbon content is perhaps 0.6%. Microhardness range 226–279 VPH.

Viking-Age Swords and Their Inscriptions

165

Fig. 73. Microstructure of IV. 8.: pearlite, slightly divorced, and ferrite (scale bar 10 microns).

9. Bergen Historisk Museum 3993 An excavated sword with the end is broken off near the hilt; it has an inscription which may be read + I I E.… The microstructure shows mostly pearlite with a few ferrite grains, and very few slag inclusions. The overall carbon content is perhaps 0.6%. Microhardness range 199–264 VPH.

Fig. 74. Microstructure of IV.9; pearlite with a little ferrite (scale bar 50 microns).

166

chapter eight

10. Bergen Historisk Museum 1622 This has an inlaid hilt, and an inscription which seems to read now …. B R T.

Fig. 75. Two specimens were taken.

(i) A specimen from the centre of the blade shows a microstructure of ferrite and slag. (not shown) Microhardness range 119–163 VPH (ii) A specimen from the edge of the blade shows a uniform microstructure of pearlite, which has been largely divorced to cementite by heating, together with some slag inclusions. There do not seem to be any separate areas of ferrite grains. This is a steel of around 0.8%C which has been annealed; for what reason can only be conjectured. Microhardness range 204–305; average = 231VPH.

Fig. 76. Microstructure of IV. 10: ferrite and cementite (scale bar 10 microns).

Viking-Age Swords and Their Inscriptions

167

11. Institute of History, Tallinn University AI 2643:108 This was a sword from Randvere, on the island of Saaremaa, which had previously been analysed by Anteins, who sectioned it transversely.31 He found that the central core was 0.15%C, and the outer layers were 0.4%– 0.8% C. The outer strips appeared to reach to the edges. It will be noted that the cutting edges were iron, and the backing strips steel—the opposite of any desirable arrangement. The surface was polished by Anteins, but the inscription was not then described in any detail. In a later publication (1973) he published his reading of this as +VLFBERH+ (without a final T). It is now difficult to discern this.

Fig. 77. Sword IV.11.

The hardness varies from 130–190 VPH in the centre. A section was removed from the broken end for re-examination. The microstructure consists of ferrite and pearlite only, with slag inclusions. One of the outer bands appears to be wider than the other. The cutting edges are formed from the low-carbon band, which does not suggest that it would have been a very successful sword. A division between high- and low-carbon bands is marked by elongated slag inclusions. The surface in this photomicrograph with more pearlite is the outer surface of the sword. A large corrosion crack has opened up.

31 Anteins, A.K. “Pattern-welded objects found in Baltic states” Journal of the Iron & Steel Institute, 206 (June 1968) 563–571. This is a much shorter (English) version of the 1973 paper of Anteins.

168

chapter eight

Fig. 78. Microstructure of IV. 11 (scale bar = 100 microns)

12. Helsinki, Kansali Museo 8911 (from Mynämäki). Possibly 10th century This is a fragmentary blade with no crosshilt or pommel, but with inscriptions; on one side REX and on the other side CONSTANIIENS with the N s reversed and the middle A and N run together (reading of Leppäaho)32. It might be a debased copy of CONSTANTINUS.33 A complete section was available from the research initiated by Leppäaho, but never completed.

Fig. 79. Cross section of sword 8911 (width 42 mm)

The microstructure consists of pearlite and ferrite with some slag inclu­ sions. There are also lines of inclusions where the outer pearlitic areas (of around 0.5%C) have been welded to the central, ferritic areas. This sword has been forged from a billet made up of a piece (or several pieces) of iron surrounded by pieces of steel, to give a harder outer part, and harder edges, but it has not been quenched. It seems to be representative of many swords of average quality. 32 Leppäaho, J. “Späteisenzeitliche waffen aus Finland” (Helsinki, 1964) 38. 33 Oakeshott, op. cit. 35, for a similar inscription.

Viking-Age Swords and Their Inscriptions

169

Fig. 80. Cross-section of an edge shows weld lines where three pieces of steel (dark-etching) and one piece of iron (lighter-etching) have been joined together here

Fig. 81. Microstructure of edge of sword 8911 (scale bar = 10 microns)

13. Saaremaa Museum K85–123 It has been polished in the past, but only a few letters can now be discerned. ….S…. A section was removed from the end of the broken blade. Anteins (op. cit. 51) described it as a “letter-like inlay” resembling HKIΛI

170

chapter eight

Fig. 82. 

The microstructure consists of pearlite with a lower-C% band near the surface. There is a sharp division between the two areas. There are long inclusions, apparently from folding during manufacture, as well as a few small elongated inclusions.

Fig. 83. Microstructure of IV.13 showing the join between higher- and lower-carbon layers (scale bar 50 microns).

14. The analysis of a sword found at Donnybrook (County Dublin) and subsequently displayed in Castle Museum, Nottingham (T608) was given in Hall.34 The inscription appears to be F N H I ┘I (T). 34 Hall, R.A. “A Viking-age grave at Donnybrook” Medieval Archaeology, 22 (1978), pp.64–83. 

Viking-Age Swords and Their Inscriptions

171

A sample was taken from the broken end, and analysed in the Ancient Monuments Laboratory, London, for elements by EPMA as well by metallography. Analysis of samples from Donnybrook sword

C

Si

Mn

S

P

Ni

Al

Centre Edge

0.2 0.3–0.4

* *

0.1–1.0 0.1

0.01 tr

0.02 tr

n.d 0.2–0.3