Making a Mint: Comparative Studies in Late Iron Age Coin Mould [1 ed.] 9781784914073

This book presents the first large-scale comparative study of Iron Age coin mould. The subject of Iron Age minting techn

205 45 6MB

English Pages 217 Year 2016

Report DMCA / Copyright

DOWNLOAD PDF FILE

Recommend Papers

Making a Mint: Comparative Studies in Late Iron Age Coin Mould [1 ed.]
 9781784914073

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Making a Mint Comparative Studies in Late Iron Age Coin Mould Mark Landon

Making a Mint Comparative Studies in Late Iron Age Coin Mould

Mark Landon

Archaeopress Archaeology

Archaeopress Publishing Ltd Gordon House 276 Banbury Road Oxford OX2 7ED

www.archaeopress.com

ISBN 978 1 78491 408 0 ISBN 978 1 78491 409 7 (e-Pdf)

© Archaeopress and Mark Landon 2016 Cover: Conjoining fragments of a Verulamium form mould tray from Ford Bridge, Braughing

All rights reserved. No part of this book may be reproduced, in any form or by any means, electronic, mechanical, photocopying or otherwise, without the prior written permission of the copyright owners. This book is available direct from Archaeopress or from our website www.archaeopress.com

Contents

List of Figures������������������������������������������������������������������������������������������������������������������������������������������������������� v Acknowledgements��������������������������������������������������������������������������������������������������������������������������������������������� xi Chapter 1: Starting point��������������������������������������������������������������������������������������������������������������������������������������1 1 The background to this study���������������������������������������������������������������������������������������������������������������������������������1 2 What is coin mould?�����������������������������������������������������������������������������������������������������������������������������������������������3 Chapter 2: The Literature��������������������������������������������������������������������������������������������������������������������������������������5 Chapter 3: Recording coin mould: aims and methodology����������������������������������������������������������������������������������18 A Aims��������������������������������������������������������������������������������������������������������������������������������������������������������������������18 B Methodology��������������������������������������������������������������������������������������������������������������������������������������������������������18 C Resolving the theories into testable propositions������������������������������������������������������������������������������������������������18 D Coin Mould Recording Protocol�������������������������������������������������������������������������������������������������������������������������30 E Database Key version 2.6�����������������������������������������������������������������������������������������������������������������������������������32 Chapter 4: The Henderson Collection (Braughing) coin mould assemblage����������������������������������������������������������35 1 General observations��������������������������������������������������������������������������������������������������������������������������������������������35 2 Tray forms�����������������������������������������������������������������������������������������������������������������������������������������������������������37 3 Edge Profiles�������������������������������������������������������������������������������������������������������������������������������������������������������37 4 Edge markings�����������������������������������������������������������������������������������������������������������������������������������������������������38 5 Evidence of elaboration���������������������������������������������������������������������������������������������������������������������������������������38 6 Methods of hole manufacture������������������������������������������������������������������������������������������������������������������������������38 7 Predictable relationship between base and top hole diameters���������������������������������������������������������������������������39 8 Predictable relationship between base diameter and pellet module��������������������������������������������������������������������40 9 Hole depth����������������������������������������������������������������������������������������������������������������������������������������������������������40 10 Control of hole volume�������������������������������������������������������������������������������������������������������������������������������������41 11 Calcium carbonate traces�����������������������������������������������������������������������������������������������������������������������������������42 12 Proportions of used and unused mould fragments��������������������������������������������������������������������������������������������42 13 Grass marks, chaff marks and matting marks���������������������������������������������������������������������������������������������������43 14 Grain casts���������������������������������������������������������������������������������������������������������������������������������������������������������43 15 Inclusions and tempers��������������������������������������������������������������������������������������������������������������������������������������43 16 Clay caps and luting������������������������������������������������������������������������������������������������������������������������������������������43 17 Conclusions��������������������������������������������������������������������������������������������������������������������������������������������������������43 Chapter 5: The Ford Bridge (Braughing) assemblage�������������������������������������������������������������������������������������������46 1 General observations��������������������������������������������������������������������������������������������������������������������������������������������46 2 Tray forms������������������������������������������������������������������������������������������������������������������������������������������������������������47 3 Edge profiles�������������������������������������������������������������������������������������������������������������������������������������������������������48 4 Edge markings�����������������������������������������������������������������������������������������������������������������������������������������������������48 5 Evidence of elaboration���������������������������������������������������������������������������������������������������������������������������������������49 6 Methods of hole manufacture�������������������������������������������������������������������������������������������������������������������������������50 7 Predictable relationship between base and top hole diameters���������������������������������������������������������������������������50 8 Predictable relationship between hole base diameter and pellet module������������������������������������������������������������51 9 Hole Depth����������������������������������������������������������������������������������������������������������������������������������������������������������52 10 Control of hole volume�������������������������������������������������������������������������������������������������������������������������������������53 11 Calcium carbonate traces�����������������������������������������������������������������������������������������������������������������������������������54 12 Proportions of used and unused mould fragments��������������������������������������������������������������������������������������������54 13 Grass marks, chaff marks and matting marks���������������������������������������������������������������������������������������������������55 14 Grain casts����������������������������������������������������������������������������������������������������������������������������������������������������������56 15 Inclusions and tempers��������������������������������������������������������������������������������������������������������������������������������������56 16 Clay caps and luting�������������������������������������������������������������������������������������������������������������������������������������������57 i

17 ‘Raised platform’ mould������������������������������������������������������������������������������������������������������������������������������������57 18 Conclusions��������������������������������������������������������������������������������������������������������������������������������������������������������58 Chapter 6: The PuckeridgeAssemblage���������������������������������������������������������������������������������������������������������������62 1 General observations�������������������������������������������������������������������������������������������������������������������������������������������62 2 Tray Forms����������������������������������������������������������������������������������������������������������������������������������������������������������62 3 Edge Profiles�������������������������������������������������������������������������������������������������������������������������������������������������������64 4 Edge markings�����������������������������������������������������������������������������������������������������������������������������������������������������65 5 Evidence of elaboration���������������������������������������������������������������������������������������������������������������������������������������65 6 Methods of hole manufacture������������������������������������������������������������������������������������������������������������������������������66 7 Number of holes in a tray������������������������������������������������������������������������������������������������������������������������������������67 8 Predictable relationship between base and top hole diameters����������������������������������������������������������������������������67 9 Predictable relationship between hole base diameter and pellet module������������������������������������������������������������69 10 Hole depth���������������������������������������������������������������������������������������������������������������������������������������������������������71 11 Control of volume���������������������������������������������������������������������������������������������������������������������������������������������71 12 Calcium carbonate traces����������������������������������������������������������������������������������������������������������������������������������73 13 The introduction of metal into holes�����������������������������������������������������������������������������������������������������������������74 14 Proportions of used and unused pellet mould����������������������������������������������������������������������������������������������������74 15 Grass marks, chaff marks and grain casts����������������������������������������������������������������������������������������������������������77 16 Inclusions in mould fabric���������������������������������������������������������������������������������������������������������������������������������78 17 Clay caps or luting?�������������������������������������������������������������������������������������������������������������������������������������������79 18 Raised platform mould���������������������������������������������������������������������������������������������������������������������������������������80 19 Conclusions�������������������������������������������������������������������������������������������������������������������������������������������������������81 Chapter 7: The Wickham Kennels assemblage����������������������������������������������������������������������������������������������������83 1 General observations��������������������������������������������������������������������������������������������������������������������������������������������83 2 Tray forms������������������������������������������������������������������������������������������������������������������������������������������������������������85 3 Edge profiles�������������������������������������������������������������������������������������������������������������������������������������������������������85 4 Edge markings�����������������������������������������������������������������������������������������������������������������������������������������������������86 5 Evidence of elaboration����������������������������������������������������������������������������������������������������������������������������������������86 6 Methods of hole manufacture������������������������������������������������������������������������������������������������������������������������������86 7 Predictable relationship between base and top hole diameters���������������������������������������������������������������������������87 8 Predictable relationship between base diameter and pellet module���������������������������������������������������������������������87 9 Hole depth������������������������������������������������������������������������������������������������������������������������������������������������������������87 10 Control of hole volume�������������������������������������������������������������������������������������������������������������������������������������88 11 Calcium carbonate traces�����������������������������������������������������������������������������������������������������������������������������������89 12 Proportions of used and unused mould fragments���������������������������������������������������������������������������������������������89 13 Grass marks, chaff marks and matting marks����������������������������������������������������������������������������������������������������90 14 Grain casts����������������������������������������������������������������������������������������������������������������������������������������������������������90 15 Inclusions and tempers���������������������������������������������������������������������������������������������������������������������������������������90 16 Clay caps and luting�������������������������������������������������������������������������������������������������������������������������������������������90 17 Selective deposition considered�������������������������������������������������������������������������������������������������������������������������90 18 Conclusions�������������������������������������������������������������������������������������������������������������������������������������������������������91 Chapter 8: Small finds from Braughing/Puckeridge���������������������������������������������������������������������������������������������93 1 General observations��������������������������������������������������������������������������������������������������������������������������������������������93 2 Fragment RR/BC/5860����������������������������������������������������������������������������������������������������������������������������������������93 3 Fragment WB/SOG 5171�������������������������������������������������������������������������������������������������������������������������������������97 4 Fragments RR/RC 01 and RR/RC 02�����������������������������������������������������������������������������������������������������������������100 5 Fragment RR/BER 5881������������������������������������������������������������������������������������������������������������������������������������102 Chapter 9: The Bagendon study sample������������������������������������������������������������������������������������������������������������ 107 1 General observations�����������������������������������������������������������������������������������������������������������������������������������������107 2 Tray forms���������������������������������������������������������������������������������������������������������������������������������������������������������108 3 Edge profiles�����������������������������������������������������������������������������������������������������������������������������������������������������110 4 Edge markings���������������������������������������������������������������������������������������������������������������������������������������������������110 5 Evidence of elaboration�������������������������������������������������������������������������������������������������������������������������������������111 6 Methods of hole manufacture���������������������������������������������������������������������������������������������������������������������������111 7 Number of holes in a tray����������������������������������������������������������������������������������������������������������������������������������113 8 Predictable relationship between base and top hole diameters��������������������������������������������������������������������������113 ii

9 Predictable relationship between hole base diameter and coin denomination��������������������������������������������������114 10 Control of volume�������������������������������������������������������������������������������������������������������������������������������������������115 11 Calcium carbonate traces���������������������������������������������������������������������������������������������������������������������������������116 12 The introduction of metal into holes���������������������������������������������������������������������������������������������������������������116 13 Proportions of used and unused pellet mould�������������������������������������������������������������������������������������������������117 14 Grass marks, chaff marks and grain casts�������������������������������������������������������������������������������������������������������118 15 Inclusions in mould fabric�������������������������������������������������������������������������������������������������������������������������������119 16 Clay caps and luting����������������������������������������������������������������������������������������������������������������������������������������119 17 Selective deposition: a serious possibility?�����������������������������������������������������������������������������������������������������119 18 Conclusions�����������������������������������������������������������������������������������������������������������������������������������������������������121 Chapter 10: Coin mould from Old Sleaford in the British Museum�������������������������������������������������������������������� 123 1 General observations�����������������������������������������������������������������������������������������������������������������������������������������123 2 Tray forms���������������������������������������������������������������������������������������������������������������������������������������������������������124 3 Edge profiles�����������������������������������������������������������������������������������������������������������������������������������������������������125 4 Edge markings���������������������������������������������������������������������������������������������������������������������������������������������������126 5 Evidence of elaboration�������������������������������������������������������������������������������������������������������������������������������������126 6 Methods of hole manufacture����������������������������������������������������������������������������������������������������������������������������127 7 Predictable relationship between base and top hole diameters�������������������������������������������������������������������������128 8 Predictable relationship between base diameter and pellet module�������������������������������������������������������������������128 9 Hole depth���������������������������������������������������������������������������������������������������������������������������������������������������������129 10 Control of hole volume�����������������������������������������������������������������������������������������������������������������������������������130 11 Calcium carbonate traces���������������������������������������������������������������������������������������������������������������������������������130 12 Proportions of used and unused mould fragments������������������������������������������������������������������������������������������131 13 Grass marks, chaff marks and matting marks�������������������������������������������������������������������������������������������������131 14 Grain casts�������������������������������������������������������������������������������������������������������������������������������������������������������131 15 Inclusions and tempers������������������������������������������������������������������������������������������������������������������������������������131 16 Clay caps and luting����������������������������������������������������������������������������������������������������������������������������������������132 17 Conclusions�����������������������������������������������������������������������������������������������������������������������������������������������������132 Chapter 11: The Turners Hall Farm Assemblage������������������������������������������������������������������������������������������������ 134 1 General observations������������������������������������������������������������������������������������������������������������������������������������������134 2 Tray forms���������������������������������������������������������������������������������������������������������������������������������������������������������136 3 Edge Profiles�����������������������������������������������������������������������������������������������������������������������������������������������������137 4 Edge markings���������������������������������������������������������������������������������������������������������������������������������������������������138 5 Evidence of elaboration�������������������������������������������������������������������������������������������������������������������������������������138 6 Methods of hole manufacture����������������������������������������������������������������������������������������������������������������������������139 7 Number of holes in a tray����������������������������������������������������������������������������������������������������������������������������������139 8 Predictable relationship between base and top hole diameters�������������������������������������������������������������������������139 9 Predictable relationship between hole base diameter and pellet module����������������������������������������������������������140 10 Hole depth�������������������������������������������������������������������������������������������������������������������������������������������������������142 11 Control of hole volume������������������������������������������������������������������������������������������������������������������������������������143 12 Calcium carbonate traces��������������������������������������������������������������������������������������������������������������������������������143 13 Proportions of used and unused mould fragments������������������������������������������������������������������������������������������143 14 Grass marks, chaff marks and matting marks�������������������������������������������������������������������������������������������������144 15 Grain casts�������������������������������������������������������������������������������������������������������������������������������������������������������144 16 Inclusions and tempers������������������������������������������������������������������������������������������������������������������������������������144 17 Clay caps and luting����������������������������������������������������������������������������������������������������������������������������������������145 18 ‘Raised platform’ mould����������������������������������������������������������������������������������������������������������������������������������145 19 Conclusions�����������������������������������������������������������������������������������������������������������������������������������������������������145 Chapter 12: Conclusions����������������������������������������������������������������������������������������������������������������������������������� 148 Section A: Evaluating the protocol�����������������������������������������������������������������������������������������������������������������������148 Section B: Collating and comparing the data.�������������������������������������������������������������������������������������������������������148 1 General observations������������������������������������������������������������������������������������������������������������������������������������������148 2 Tray forms����������������������������������������������������������������������������������������������������������������������������������������������������������153 3 Edge Profiles�����������������������������������������������������������������������������������������������������������������������������������������������������156 4 Edge markings���������������������������������������������������������������������������������������������������������������������������������������������������157 5 Evidence of elaboration�������������������������������������������������������������������������������������������������������������������������������������160

iii

6 Methods of hole manufacture��������������������������������������������������������������������������������������������������������������������������� 164 7 Tray capacity���������������������������������������������������������������������������������������������������������������������������������������������������� 166 8 Predictable relationship between base and top hole diameter�������������������������������������������������������������������������� 167 9 Predictable relationship between hole base diameter and pellet module��������������������������������������������������������� 167 10 Hole depth������������������������������������������������������������������������������������������������������������������������������������������������������ 169 11 Control of hole volume����������������������������������������������������������������������������������������������������������������������������������� 170 12 Calcium carbonate traces������������������������������������������������������������������������������������������������������������������������������� 171 13 Introduction of metal into holes��������������������������������������������������������������������������������������������������������������������� 172 14 Proportions of used and unused coin mould�������������������������������������������������������������������������������������������������� 172 15 Grass marks, chaff marks and matting marks������������������������������������������������������������������������������������������������ 177 16 Grain casts������������������������������������������������������������������������������������������������������������������������������������������������������ 178 17 Inclusions and tempers����������������������������������������������������������������������������������������������������������������������������������� 179 18 Clay caps and luting��������������������������������������������������������������������������������������������������������������������������������������� 179 19 Raised platform mould����������������������������������������������������������������������������������������������������������������������������������� 180 20 Selective deposition considered��������������������������������������������������������������������������������������������������������������������� 180 Section C: Drawing the threads together�������������������������������������������������������������������������������������������������������������� 182 Appendix I: Some experiments in the manufacture of coin mould�������������������������������������������������������������������185 1 Preamble����������������������������������������������������������������������������������������������������������������������������������������������������������� 185 2 Experiment Series 1: Making trays.����������������������������������������������������������������������������������������������������������������� 185 3 Experiment Series 2: Making Holes with a Single-Pronged Dibber����������������������������������������������������������������� 187 4 Experiment series 3: Making holes with a multi-pronged dibber�������������������������������������������������������������������� 190 5 Experiment series 4: Examining how much force is required to make a hole in wet clay with a dibber.�������� 194 Appendix II: A List of British Find-Sites��������������������������������������������������������������������������������������������������������������196 Bibliography�����������������������������������������������������������������������������������������������������������������������������������������������������198

iv

List of Figures Figure 1.1: The Ford Bridge mint trench.������������������������������������������������������������������������������������������������������������������������������������������� 2 Figure 1.2: In situ coin mould at Ford Bridge.������������������������������������������������������������������������������������������������������������������������������������ 2 Figure 1.3: Presumed method of using pellet mould.����������������������������������������������������������������������������������������������������������������������� 4 Figure 3.1: A completed record card, front and back.��������������������������������������������������������������������������������������������������������������������� 19 Figure 3.2: ‘Verulamium’ form tray.������������������������������������������������������������������������������������������������������������������������������������������������ 19 Figure 3.3 Verulamium form tray from Merlin Works, Leicester����������������������������������������������������������������������������������������������������� 19 Figure 3.4: ‘Puckeridge’ form tray.��������������������������������������������������������������������������������������������������������������������������������������������������� 20 Figure 3.5: I-section profile������������������������������������������������������������������������������������������������������������������������������������������������������������� 20 Figure 3.6: An experimental ‘box-mould’ with one open end.�������������������������������������������������������������������������������������������������������� 21 Figure 3.7: ‘Lazy S’ Profile���������������������������������������������������������������������������������������������������������������������������������������������������������������� 21 Figure 3.8: An experimental ‘bowl-mould’ with one open end.������������������������������������������������������������������������������������������������������ 21 Figure 3.9: ‘Straight section’ profile.����������������������������������������������������������������������������������������������������������������������������������������������� 21 Figure 3.10: ‘Angled section’ profile������������������������������������������������������������������������������������������������������������������������������������������������ 21 Figure 3.11: ‘Rolled edge’ profile����������������������������������������������������������������������������������������������������������������������������������������������������� 21 Figure 3.12: ‘Overhang’ profile.������������������������������������������������������������������������������������������������������������������������������������������������������� 22 Figure 3.13: ‘Cut and tear’ banding.������������������������������������������������������������������������������������������������������������������������������������������������ 22 Figure 3.14: Results of an experiment to produce 26 holes with a controlled depth of 5mm.������������������������������������������������������� 27 Figure 3.15 Comparing the average depth and standard deviation for four experimental trays attempting to achieve a hole depth of 5mm.������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 27 Figure 4.1: Average number of holes in rows and columns for fragments with more than 5 holes.����������������������������������������������� 35 Figure 4.2: Tray average thickness in the Henderson Collection expressed as percentages.���������������������������������������������������������� 36 Figure 4.3: Comparing composition - position types expressed as percentages of the total number of individually listed fragments.��� 36 F igure 4.4: Record-card diagram of fragment HC/30���������������������������������������������������������������������������������������������������������������������� 37 Figure 4.5: Edge profile distribution.����������������������������������������������������������������������������������������������������������������������������������������������� 37 Figure 4.6: Average intra-fragment standard deviations in three hole parameters compared - How careful were the mould-makers?�� 39 Figure 4.7: Variability in relationship between top and base hole diameters in the Henderson Collection and the Ford Bridge assemblages expressed as percentages of the number of holes in each assemblage exhibiting both measurements.����������������� 39 Figure 4.8: Top diameter distribution in the Henderson Collection������������������������������������������������������������������������������������������������ 39 Figure 4.9: Fragment average base diameter distribution in the Henderson Collection.���������������������������������������������������������������� 40 Figure 4.10: Fragment average base diameter distribution in the Henderson Collection expressed graphically.��������������������������� 40 Figure 4.11 : Intra-fragment diameter variation in fragments with 2 or more measurable diameters������������������������������������������� 40 Figure 4.12: Fragment average hole depths in the Henderson Collection�������������������������������������������������������������������������������������� 40 Figure 4.13: Fragment average hole depths in the Henderson Collection expressed as percentages.������������������������������������������� 41 Figure 4.14: Hole size variation in the Henderson Collection tabulated.���������������������������������������������������������������������������������������� 41 Figure 4.15: Average hole volume plotted against average hole base diameter.���������������������������������������������������������������������������� 41 Figure 4.16: Average intra-fragment total variation in three hole parameters in the Henderson Collection compared with experimentally generated data.������������������������������������������������������������������������������������������������������������������������������������������������������ 42 Figure 5.1: Average thickness measurements from the Ford Bridge assemblage compared with study averages.������������������������ 46 Figure 5.2: Fragment average thicknesses expressed as percentages.�������������������������������������������������������������������������������������������� 46 Figure 5.3: Verulamium form pediment with horizontal incised guideline. Note the deformation of the apex hole.�������������������� 47 Figure 5.4: Edge profile distribution.����������������������������������������������������������������������������������������������������������������������������������������������� 48 Figure 5.5: Band and lines edge marking���������������������������������������������������������������������������������������������������������������������������������������� 48 Figure 5.6: Mould lining mark, possibly made by bast or bark�������������������������������������������������������������������������������������������������������� 49 Figure 5.7: Possible Puckeridge form fragment with 17mm diameter holes and an incised guideline.������������������������������������������ 50 Figure 5.8: Hole slighting in two axes – arrows show the characteristic flattening������������������������������������������������������������������������� 50 Figure 5.9: Variability in relationship between top and base hole diameters in the Ford Bridge Assemblage������������������������������� 51 Figure 5.10: Top diameter distribution.������������������������������������������������������������������������������������������������������������������������������������������� 51 Figure 5.11: Base diameter distribution by percentage.����������������������������������������������������������������������������������������������������������������� 52 Figure 5.12: Homogenous base diameter distribution. Context 00 forms the first cluster, context 03 the second, and the extended third cluster is formed from contexts 04; 06; 09 and VH.��������������������������������������������������������������������������������������������������������������� 52 Figure 5.13: Fragment average hole depths������������������������������������������������������������������������������������������������������������������������������������ 53 Figure 5.14: Fragment average hole depths expressed as percentages������������������������������������������������������������������������������������������ 53 Figure 5.15: Base diameter plotted against volume������������������������������������������������������������������������������������������������������������������������ 53 Figure 5.16: Variation within hole size groups tabulated.��������������������������������������������������������������������������������������������������������������� 53 Figure 5.17: Chalk wash in mould holes, approximately 1.5mm thick at the base.������������������������������������������������������������������������ 54 Figure 5.18: Signs of extreme heating - Ford Bridge and Puckeridge compared.���������������������������������������������������������������������������� 55 Figure 5.19: Grass marks on tray base.�������������������������������������������������������������������������������������������������������������������������������������������� 55 Figure 5.20: Grain cast in hole base (Frag. BRR/03/096).���������������������������������������������������������������������������������������������������������������� 56 Figure 5.21: Key to Figures 5.22 and 5.23.�������������������������������������������������������������������������������������������������������������������������������������� 56 Figure 5.22: Inclusions and tempers from Ford Bridge (BRR) and Puckeridge (PUC) expressed as % of total inclusions + tempers for each site.����������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 56 Figure 5.23: Chalk and shell tempers from Ford Bridge (BRR) and Puckeridge (PUC) expressed as % of individually recorded fragments.��������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 57 Figure 5.24: Moulded platform fragment (Frag. BRR/06/006). Note vesiculation. ������������������������������������������������������������������������ 57

v

Figure 6.1: High-peaked Verulamium form tray fragment�������������������������������������������������������������������������������������������������������������� 62 Figure 6.2: Puckeridge form tray (PUC/Box 2/0008)����������������������������������������������������������������������������������������������������������������������� 63 Figure 6.3: Edge profile distribution in the Puckeridge assemblage������������������������������������������������������������������������������������������������ 64 Figure 6.4: ‘Lines and banding’ edge marking��������������������������������������������������������������������������������������������������������������������������������� 65 Figure 6.5: Lateral and double horizontal incised guidelines���������������������������������������������������������������������������������������������������������� 66 Figure 6.6: Boustrophedon hole making revealed by slighting������������������������������������������������������������������������������������������������������� 67 Figure 6.7: The distribution of differences between top and hole base diameter in the Puckeridge assemblage.������������������������� 68 Figure 6.8: Distribution of intra-fragment variation in the difference between top and hole base diameters.������������������������������ 68 Figure 6.9: Hole top diameter distribution.������������������������������������������������������������������������������������������������������������������������������������� 68 Figure 6.10: Fragment mean hole base diameter distribution.������������������������������������������������������������������������������������������������������� 70 Figure 6.11: Fragment mean hole base diameter distribution expressed graphically��������������������������������������������������������������������� 70 Figure 6.12: Mean intra-fragment variation in base diameter (in mm) correlated with fragment size.������������������������������������������ 70 Figure 6.13: Fragment average hole depths in the Puckeridge assemblage.����������������������������������������������������������������������������������� 71 Figure 6.14: Scatter graph plotting base diameter in mm against volume in mm3������������������������������������������������������������������������� 72 Figure 6.15: Range and distribution of volumes in the Puckeridge assemblage.���������������������������������������������������������������������������� 72 Figure 6.16: Calcium carbonate wash. Arrow shows incised guideline.������������������������������������������������������������������������������������������ 73 Figure 6.20: Evidence of trays adhering during heating������������������������������������������������������������������������������������������������������������������ 76 Figure 6.21: Occlusion of mould hole by heat-induced slumping��������������������������������������������������������������������������������������������������� 76 Figure 6.22: Two fragments melted together after breakage: a second heating episode?������������������������������������������������������������� 77 Figure 6.23: Evidence of heat applied by tuyère? A ‘plume’ of differential heating on the top surface of a fragment.����������������� 77 Figure 6.24: Grain cast on a tray base��������������������������������������������������������������������������������������������������������������������������������������������� 78 Figure 6.25: Large pebble inclusion������������������������������������������������������������������������������������������������������������������������������������������������� 78 Figure 6.26: Partial clay caps����������������������������������������������������������������������������������������������������������������������������������������������������������� 79 Figure 6.27: Intact clay cap�������������������������������������������������������������������������������������������������������������������������������������������������������������� 79 Figure 6.29: Possible luted hole from Puckeridge��������������������������������������������������������������������������������������������������������������������������� 79 Figure 7.1: Concordance attempting to reconcile Landon numbering with Cowell and Tite indexing.������������������������������������������� 83 Figure 7.2: Average number of holes in rows and columns for fragments with 5 holes or more.��������������������������������������������������� 83 Figure 7.3: Percentages of position types compared.��������������������������������������������������������������������������������������������������������������������� 84 Figure 7.4: Edge profile distribution.����������������������������������������������������������������������������������������������������������������������������������������������� 85 Figure 7.5: Intra-fragment standard deviations in three hole parameters compared - How careful were the mould-makers?����� 87 Figure 7.6: Fragment average hole depths in the Wickham Kennels assemblage.�������������������������������������������������������������������������� 88 Figure 7.7: Individual hole depth variation on frag. BR82/WK/3���������������������������������������������������������������������������������������������������� 88 Figure 7.8: Fragment average volumes and intra-fragment total variation in volume.������������������������������������������������������������������� 88 Figure 8.1: Comparing position type percentages.�������������������������������������������������������������������������������������������������������������������������� 93 Figure 8.2: Comparing average fragment size (Length 1 and Length 2).����������������������������������������������������������������������������������������� 93 Figure 8.3: Proportions of incomplete vs. complete holes compared.�������������������������������������������������������������������������������������������� 93 Figure 8.4: Fragment RR/BC/5860, showing where thickness measurements were taken.������������������������������������������������������������ 94 Figure 8.5: RR/BC/5860 – Thickness compared.������������������������������������������������������������������������������������������������������������������������������ 94 Figure 8.6: Patterns of slighting on RR/BC 5860. Arrows denote the direction of slighting. Slighting holes were made after slighted holes.����������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 95 Figure 8.7: Intra-fragment base diameter variability compared between fragments with 18+ holes.�������������������������������������������� 95 Figure 8.8: Hole base diameter variability on RR/BC 5860.������������������������������������������������������������������������������������������������������������ 96 Figure 8.9: Depth variation on RR/BC 5860 compared.������������������������������������������������������������������������������������������������������������������� 96 Figure 8.10: RR/BC/5860 - Variation in volume plotted against hole base diameter.��������������������������������������������������������������������� 96 Figure 8.11: Grass marks on the base of RR/BC 5860. ������������������������������������������������������������������������������������������������������������������� 97 Figure 8.12: Fragment WB/SOG 5171.�������������������������������������������������������������������������������������������������������������������������������������������� 97 Figure 8.13: Probable grain cast on WB/SOG 5171.������������������������������������������������������������������������������������������������������������������������ 99 Figure 8.14: Arrow shows massive flint inclusion.������������������������������������������������������������������������������������������������������������������������� 100 Figure 8.15: Fragment RR/RC 01��������������������������������������������������������������������������������������������������������������������������������������������������� 100 Figure 8.17: Selective deposition – fragment sizes compared.���������������������������������������������������������������������������������������������������� 102 Figure 8.18: Selective deposition – Average number of holes per fragment.������������������������������������������������������������������������������� 102 Figure 8.19: Fragment RR/BER 5881��������������������������������������������������������������������������������������������������������������������������������������������� 103 Figure 8.20: Hole numbering on RR/BER 5881.����������������������������������������������������������������������������������������������������������������������������� 103 Figure 8.21: Hole base diameter variation on RR/BER 5881��������������������������������������������������������������������������������������������������������� 104 Figure 8.22: Hole depths in mm across RR/BER 5881.������������������������������������������������������������������������������������������������������������������ 104 Figure 8.23: RR/BER 5881 - Hole volume in mm3 ������������������������������������������������������������������������������������������������������������������������ 104 Figure 8.25: Arrows indicate black crust, possibly silver salts, around hole mouth. Areas immediately above these marks are stained purple. Note also vesiculation.����������������������������������������������������������������������������������������������������������������������������������������� 105 Figure 8.26: Chaff cast on the upper surface of RR/BER 5881.����������������������������������������������������������������������������������������������������� 105 Figure 8.27: Incomplete grain cast on the base of RR/BER 5881�������������������������������������������������������������������������������������������������� 105 Figure 9.1: Correlating fragment numbering systems for the Bagendon study sample.��������������������������������������������������������������� 107 Figure 9.2: Average number of holes in rows and columns for fragments with more than 5 holes in the Bagendon study sample and the Ford Bridge assemblage.�������������������������������������������������������������������������������������������������������������������������������������������������� 108 Figure 9.3: Purposive trimming or accidental fracture?���������������������������������������������������������������������������������������������������������������� 109 Figure 9.4: Bagendon edge profile types tabulated.��������������������������������������������������������������������������������������������������������������������� 110 Figure 9.5: Edge profile percentages compared.��������������������������������������������������������������������������������������������������������������������������� 110 Figure 9.6: Hole numbering on BAG 81/44.93/None (1)��������������������������������������������������������������������������������������������������������������� 112 Figure 9.7: Slighting in two axes. Arrows indicate������������������������������������������������������������������������������������������������������������������������� 112 Figure 9.8: Intra-fragment variability in the difference between top and hole base diameter in mm������������������������������������������ 113

vi

Figure 9.10: Base diameter distribution���������������������������������������������������������������������������������������������������������������������������������������� 115 Figure 9.11: Average intra-fragment variation in base diameter in the Bagendon 1981 material compared.������������������������������ 115 Figure 9.12: Correlating hole base diameter in mm with hole volume in mm3���������������������������������������������������������������������������� 116 Figure 9.13: Ancient or modern? – Arrows indicate channels linking holes on this fragment of possible potin mould.�������������� 117 Figure 9.14: Differential signs of heating on base and top of a single mould fragment. �������������������������������������������������������������� 118 Figure 9.15: Grain cast on the base of a Bagendon coin mould fragment. BAG 81/20.83/Sample 9.������������������������������������������� 118 Figure 10.1: Fragment average thicknesses in the Old Sleaford study sample expressed as percentages������������������������������������ 123 Figure 10.2: Unusual orientation of hole-row to edge and corner on Fragment OS/1684����������������������������������������������������������� 124 Figure 10.3: Minimum number of corners in the Old Sleaford assemblage for a range of tray sizes.������������������������������������������� 125 Figure 10.4: Edge profile types in the Old Sleaford study sample.������������������������������������������������������������������������������������������������ 125 Figure 10.5: Edge marking on fragment OS/1693.������������������������������������������������������������������������������������������������������������������������ 126 Figure 10.6: Diastemas expressed as percentages of hole top diameter.�������������������������������������������������������������������������������������� 127 Figure 10.7: Average intra-fragment variation in three hole parameters compared: how careful were the mould-makers?������� 127 Figure 10.8: Comparing the range in variation between base and top hole diameters, expressed as percentages, in Old Sleaford and experimental material. ���������������������������������������������������������������������������������������������������������������������������������������������������������� 128 Figure 10.9: Possible range of base diameters for Old Sleaford coin mould.�������������������������������������������������������������������������������� 128 Figure 10.10: Fragment average hole depths in the Old Sleaford study sample��������������������������������������������������������������������������� 129 Figure 10.11: Fragment average hole depths in the Old Sleaford study sample expressed as percentages.�������������������������������� 129 Figure 10.12: Hole size variation in the Old Sleaford study sample tabulated.����������������������������������������������������������������������������� 130 Figure 10.13: Average hole volume plotted against average hole base diameter.������������������������������������������������������������������������ 130 Figure 11.1: Context numbering for coin mould in the Turners Hall Farm archive.����������������������������������������������������������������������� 135 Figure 11.2: Average thicknesses for the Turners Hall Farm Assemblage compared with study averages.����������������������������������� 135 Figure 11.3: Thickness distribution for the Turners Hall assemblage.������������������������������������������������������������������������������������������� 136 Figure 11.4: Turners Hall Farm position type distribution compared with the study averages.���������������������������������������������������� 136 Figure 11.5: Fragment 1012.55/1, after Nicky Metcalf.����������������������������������������������������������������������������������������������������������������� 136 Figure 11.6: Using the ‘minimum trays’ formula to demonstrate a shortfall in corners in the Turners Hall Farm assemblage.��� 137 Figure 11.7: Turners Hall Farm edge profile types tabulated.������������������������������������������������������������������������������������������������������� 137 Figure 11.8: Fragment 1012.53/1, after Nicky Metcalf, showing relation of incised guideline to corner.������������������������������������� 138 Figure 11.9: Variability in relationship between top and base diameters in the Turners Hall Farm material, compared with results from experimental tray manufacture.������������������������������������������������������������������������������������������������������������������������������������������� 140 Figure 11.10: Top diameter distribution at Turners Hall Farm compared with the study average.����������������������������������������������� 140 Figure 11.11: Average intra-fragment variation in hole base diameter at Turners Hall Farm compared with the study average and experimental hole-making.����������������������������������������������������������������������������������������������������������������������������������������������������������� 141 Figure 11.12: Distribution of hole base diameters in the Turners Hall Farm assemblage.������������������������������������������������������������ 141 Figure 11.13: Distribution of hole base diameters in the Turners Hall Farm assemblage.������������������������������������������������������������ 142 Figure 11.14: Distribution of intra-fragment average depth measurement from the Turners Hall Farm assemblage.����������������� 142 Figure 11.15: Variability in base diameter and hole volume in the Turners Hall Farm assemblage tabulated.����������������������������� 143 Figure 11.16: Intra-fragment average hole volume tabulated against intra-fragment average base diameter. ��������������������������� 143 Figure 12.1: Context and size of assemblages discussed in this work, where known.������������������������������������������������������������������ 149 Figure 12.2: Context and size of several other assemblages of British coin mould.���������������������������������������������������������������������� 149 Figure 12.3: Assemblage average fragment sizes in mm.�������������������������������������������������������������������������������������������������������������� 151 Figure 12.4: Effectiveness of two methods of retrieval used on the same Ford Bridge, Braughing, context.������������������������������� 151 Figure 12.5: Retrieval rates for larger assemblages expressed in terms of the percentage of very small fragments of coin mould.� 152 Figure 12.7: Shortfall of corners expressed as percentages for a range of tray forms.����������������������������������������������������������������� 154 Figure 12.8: Corners expressed as a percentage of all fragments in three assemblages.������������������������������������������������������������� 154 Figure 12.9: Shortfall in corners in the Ford Bridge and Puckeridge assemblages.����������������������������������������������������������������������� 154 Figure 12.10: Maximum number of trays of varying circumferences allowed by the total edge length in the Ford Bridge assemblage.155 Figure 12.11: Shortfall in edge in the Ford Bridge assemblage between actual edge length and edge length required by 46 trays of varying circumferences.���������������������������������������������������������������������������������������������������������������������������������������������������������������� 155 Figure 12.12: Edge fragments in the larger assemblages expressed as percentages of the total number of fragments.������������� 155 Figure 12.13: Edge fragments in the smaller assemblages expressed as percentages of the total number of fragments.����������� 155 Figure 12.14: Edge and corner percentages for both larger and smaller assemblages combined.����������������������������������������������� 156 Figure 12.15: Profile type distribution in the assemblages studied expressed as percentages of total profiles per assemblage (excluding ‘Cut & Tear’).���������������������������������������������������������������������������������������������������������������������������������������������������������������� 156 Figure 12.16: Study total of instances of Type 3 and Type 4 edge profiles in the Ford Bridge, Puckeridge and Wickham Kennels assemblages.��������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 158 Figure 12.17: The incidence of 17 categories of edge marking across six assemblages expressed as percentages of all edge fragments.������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 159 Figure 12.18: Total incidence in 6 assemblages of 17 categories of edge marking expressed as percentages of all edge fragments.���� 159 Figure 12.19: Number of assemblages in the study in which each of 17 categories of edge marking occurs.������������������������������ 160 Figure 12.20: The number of assemblages in which edge markings, resolved into 4 categories, occur.�������������������������������������� 160 Figure 12.21: Incised guideline codes used in the study.�������������������������������������������������������������������������������������������������������������� 161 Figure 12.22: Distribution of thirteen classes of incised guideline in four assemblages of coin mould.��������������������������������������� 161 Figure 12.23: Three possible orientations for an ‘Incised guideline’ parallel with a non-pedimental edge.��������������������������������� 161 Figure 12.24: Incised guideline parallel with Row 1.��������������������������������������������������������������������������������������������������������������������� 162 Figure 12.25: Incised guideline parallel with pediment edge.������������������������������������������������������������������������������������������������������ 162 Figure 12.26: Common combination – IGL + IGR.������������������������������������������������������������������������������������������������������������������������� 162 Figure 12.27: Common combination - IGL + IGL. ������������������������������������������������������������������������������������������������������������������������� 162 Figure 12.28: Guidelines on a near-complete Verulamium form tray originally part of the Puckeridge assemblage.������������������ 162

vii

Figure 12.29: Combinations of ‘incised guidelines’ at Ford Bridge and Puckeridge.��������������������������������������������������������������������� 163 Figure 12.30: Double incised guidelines parallel with a Non-pedimental edge. ��������������������������������������������������������������������� 163 Figure 12.31: Double incised guidelines parallel with Row 1.������������������������������������������������������������������������������������������������������� 163 Figure 12.32: Incised guidelines outlining pediment.�������������������������������������������������������������������������������������������������������������������� 163 Figure 12.33: The occurrence of hole slighting and boustrophedon dibbing in seven assemblages of coin mould.�������������������� 165 Figure 12.34: Variability in the relationship between top and base hole diameters for 7 assemblages, expressed as assemblage percentages, compared with results generated experimentally.�������������������������������������������������������������������������������������������������� 167 Figure 12.35: Amalgamated study average variation in hole diameter.���������������������������������������������������������������������������������������� 168 Figure 12.36: Hole base diameter distribution in the Ford Bridge and Puckeridge assemblages������������������������������������������������� 168 Figure 12.37: Average intra-fragment depth variation in mm for eight assemblages of coin mould.������������������������������������������� 169 Figure 12.38: ‘Depth signatures’ of seven assemblages.��������������������������������������������������������������������������������������������������������������� 170 Figure 12.39: Average intra-fragment variation in volume in 8 assemblages.������������������������������������������������������������������������������ 170 Figure 12.40: Inter-fragment variation in volume in six assemblages.������������������������������������������������������������������������������������������ 170 Figure 12.41: Percentage of fragments in each of the study assemblages classified as Burn Category 0: ‘Unclassifiable’����������� 173 Figure 12.42: Percentage of fragments in each assemblage studied showing no sign of heating beyond the temperature necessary for firing.���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 174 Figure 12.43: Heating Category B (Reddening only) expressed as percentages of each assemblage.������������������������������������������ 174 Figure 12.44: Heating Category C (Slight Vitrification and/or Vesiculation) expressed as percentages of each assemblage.������� 175 Figure 12.45: Heating Category D (Extreme Heating) expressed as percentages of each assemblage.���������������������������������������� 175 Figure 12.46: Vitrification, vesiculation and other signs of heating tabulated by location, and expressed as percentages of (Both+Top only+Base only).����������������������������������������������������������������������������������������������������������������������������������������������������������� 177 Figure 12.47: How were the trays dried? – Grass marks, chaff marks and matting marks tabulated by location on the fragment, and expressed as percentages of the total number of individually listed fragments in each assemblage.������������������������������������������ 177 Figure 12.48: The incidence of grain casts in eight assemblages of coin mould. Percentages are expressed in terms of the number of individually listed fragments in each assemblage.�������������������������������������������������������������������������������������������������������������������� 178 Figure 12.49: Inclusions and tempers expressed as percentages of the total number (2837) of individually listed fragments in the study.��������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 179 Figure 12.50: Rationalising edge marking categories�������������������������������������������������������������������������������������������������������������������� 158 Figure A.1: Box-mould with three sides.��������������������������������������������������������������������������������������������������������������������������������������� 186 Figure A.2: Bowl-mould with one open end.��������������������������������������������������������������������������������������������������������������������������������� 186 Figure A.3: Tray 7 – Horizontal diameters in schematic representation.��������������������������������������������������������������������������������������� 188 Figure A.4: Tray 7 – Vertical diameters in schematic representation�������������������������������������������������������������������������������������������� 188 Figure A.5: Tray 7 – Hole depths in schematic representation.����������������������������������������������������������������������������������������������������� 188 Figure A.6: Tray 8 – Horizontal diameters in schematic representation���������������������������������������������������������������������������������������� 188 Figure A.7: Tray 8 – Vertical diameters in schematic representation.������������������������������������������������������������������������������������������� 188 Figure A.8: Tray 8 – Hole depths in schematic representation.����������������������������������������������������������������������������������������������������� 188 Figure A.9: Tray 9 – Horizontal diameters in schematic representation.��������������������������������������������������������������������������������������� 188 Figure A.10: Tray 9 – Vertical diameters in schematic representation.����������������������������������������������������������������������������������������� 188 Figure A.11: Tray 9 – Hole depths in schematic representation.��������������������������������������������������������������������������������������������������� 188 Figure A.12: Variation in diameter in 63 holes made in wet clay with the same single-pronged dibber (horizontal diameter 14.45mm; vertical diameter 14.60mm)���������������������������������������������������������������������������������������������������������������������������������������� 188 Figure A.13: Control of depth – error margins expressed as percentages.����������������������������������������������������������������������������������� 189 Figure A.14: Control of depth – standard deviation and tray total variation for three experimental trays����������������������������������� 189 Figure A.15: Variation in experimental hole volumes������������������������������������������������������������������������������������������������������������������� 189 Figure A.16: Intra-fragment hole variability in different samples of coin mould compared.�������������������������������������������������������� 189 Figure A.17: The distribution of deviation in diameter in two axes of holes made in wet clay from the diameter of the dibber used to make them.������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 189 Figure A.18: Schematic representation of a sixteen-hole tray, with holes denoted by letters.����������������������������������������������������� 190 Figure A.19: An experimental multi-pronged dibber.�������������������������������������������������������������������������������������������������������������������� 190 Figure A.20: Experimental multi-pronged dibber – the spaces between prongs.������������������������������������������������������������������������� 191 Figure A.21: Experimental multi-pronged dibber – Horizontal and vertical diameters and prong lengths.���������������������������������� 191 Figure A.22: Tray 10 – Horizontal diameters in schematic representation������������������������������������������������������������������������������������ 191 Figure A.23: Tray 10 – Vertical diameters in schematic representation���������������������������������������������������������������������������������������� 191 Figure A.24: Tray 10 – Hole depths in schematic representation�������������������������������������������������������������������������������������������������� 191 Figure A.25: Tray 11 – Hole spacings in the horizontal axis in schematic representation������������������������������������������������������������� 192 Figure A.26: Tray 11 – Horizontal diameters in schematic representation������������������������������������������������������������������������������������ 192 Figure A.27: Tray 11 – Vertical diameters in schematic representation���������������������������������������������������������������������������������������� 192 Figure A.28:Tray 11 - Hole depths in schematic representation.��������������������������������������������������������������������������������������������������� 192 Figure A.29: Tray 12 – Hole spacings in the horizontal axis in schematic representation������������������������������������������������������������� 192 Figure A.30: Tray 12 – Horizontal diameters in schematic representation������������������������������������������������������������������������������������ 192 Figure A.31: Tray 12 – Vertical diameters in schematic representation���������������������������������������������������������������������������������������� 192 Figure A.32: Tray 12 – Hole depths in schematic representation�������������������������������������������������������������������������������������������������� 192 Figure A.33: Total variation in hole spacing across eight dibbing iterations for three gaps in a fixed axis.����������������������������������� 192 Figure A.34: Tray 11 – Deviation from dibber spacing: dibber spacing subtracted from hole spacing.����������������������������������������� 193 Figure A.35: Tray 12 – Deviation from dibber spacing: dibber spacing subtracted from hole spacing.����������������������������������������� 193 Figure A.36: Standard deviation in diameter of holes made with single- and multi-pronged dibbers.����������������������������������������� 193 Figure A.37: Tray total variation in diameter of holes made experimentally with single– and multi-pronged dibbers.��������������� 193 Figure A.38: Standard deviation in diameter for 6 fragments of coin mould with 18 or more holes.������������������������������������������� 193 Figure A.39: Tray Total Variation in diameter for 6 fragments of coin mould with 18 or more holes.������������������������������������������ 194

viii

Figure A.40: Average variation in depth – Comparing holes made in experimental trays using single- and multi-pronged dibbers with depth variation on real fragments.���������������������������������������������������������������������������������������������������������������������������������������� 194 Figure A.41: Dimensions of the dibber used in Experiment Series 4.������������������������������������������������������������������������������������������� 194 Figure A.42: The force required (in grams) to make five holes in wet clay using a single-pronged dibber.����������������������������������� 194 Figure A.43: Force required to use dibbers of various diameters and in varying numbers.���������������������������������������������������������� 195 Figures 12.6a & 12.6b: Average lengths compared – larger assemblages versus smaller assemblages.�������������������������������������� 152 Plate 8.16: Fragment RR/RC 02����������������������������������������������������������������������������������������������������������������������������������������������������� 101

ix

x

Acknowledgements This book has been a long time in gestation, and has been brought to delivery only with considerable assistance from many ‘midwives’. Without their kindness and patience, it is quite possible that this book might never have appeared, and therefore it is only fitting that I should thank them, each and every one. First and foremost must be Dr. Jonathan Hunn, now of Icknield Archaeology. But for his initial leap of faith, I would never have had the opportunity to do this work. I must further thank both him and Dr. Isobel Thompson of the Hertfordshire County Council Historic Environment Unit for undertaking the task of reading through the draft of this book to check it for lunacy and libel. However, I ought to point out that if either should remain, then the blame is mine alone. I also owe a special debt of gratitude to Jenny Glazebrook of the Norfolk publications unit. Not only has she offered advice and encouragement through the years, but she has also spent much time trying to teach me how to format text and images for publication. However, despite her best efforts, the credit for hammering the manuscript into a shape fit for printing must go to my editor at Archaeopress, Dr. Rajka Makjanic, whose patience, experience and skill have been much in evidence these last few months.The data used in preparing Chapter 11, ‘The Turners Hall Farm Assemblage’, was gathered by Nicky Metcalf, although the views, opinions and conclusions built on these facts are – except where stated – entirely my own. I would also like to thank the following people, who have all contributed either data or ideas which have helped me in the writing of ‘Making a Mint’: Dr. Richard Brickstock of Durham Castle; Dr. Stewart Bryant, formerly County Archaeologist for Hertfordshire; Professor John Collis of Sheffield University; Dr. Geoff Cottam; Professor Colin Haselgrove of Leicester University; Dr. Philip de Jersey of the Guernsey Archaeological Service; Dr. Ian Leins of the British Museum; Henrietta Longden, MA; Dr. Tom Moore of Durham University; Dave Parker, formerly of ULAS; Dr. Matthew Ponting of Liverpool University; Chris Rudd and Elizabeth Cottam of Celtic Coins; Dr. Hannah Russ and Dr. David Griffiths of Northern Archaeological Associates; Dr. John Sills of the Celtic Coin Index; Megan Stoakley of Wardell Armstrong Archaeology; John Talbot; Sara Taylor of Hertford Museum; Alison Tinniswood, MA, of the Hertfordshire County Council Historic Environment Unit HEU; Dr. Julian Watters, then of the Hertfordshire Portable Antiquities Scheme; Simon West, MA, of the Verulamium Archaeological Unit. I am beset by the suspicion that this list is not complete. If you ought to be there, and you are not, please forgive me. Finally, there is my family. To my wife. Sue, thank you for your patience; to my son, Jermyn, many thanks for the drawings of tray forms; my daughter, Esmée, should also be thanked for her help with the statistical work; my parents, Nick and Liz Landon, kept up my morale when it might have flagged; my brother, Tom, has provided invaluable help with researching French sources. My granddaughter, Jasmine Conway, is a constant source of joy and wonder to me, and it is to her that I would like to dedicate this book (although I do not expect her to read it for a few years yet!). Mark Landon April 2016

xi

xii

Chapter 1

Starting point Museum, but the mould is no longer with them: it has been suggested that it may have been retained by the British Museum following metal residue testing by Freestone.

1 The background to this study In 2006, the author of this book discovered a large deposit of clay coin mould fragments eroding from a river bank in one of the Scheduled Monument Areas south of Braughing. The find was reported, and funding was provided by English Heritage for a two-day, singletrench evaluation in advance of bank stabilization work,1 which was carried out under the direction of Dr Jonathan Hunn of ASC Ltd. In all, nearly 10kg. of mould was recovered, together with 6 kg. of pottery, bone and furnace debris. Since the deposit of coin mould was increasing in thickness as it disappeared into the trench section, it is clear that much still remains in situ. Braughing Local History Society and Dr Stewart Bryant of the Hertfordshire County Council Historic Environment Unit provided funding for a programme of Energy Dispersive Spectroscopy and electron microscopy,2 which was carried out as part of her Masters degree by Henrietta Longden, then of Liverpool University.

Since 2006, five isolated surface finds of mould have also been made in the Braughing/Puckeridge area, all at some distance from known mint sites. Then, in November 2008, the so-called ‘Puckeridge Assemblage’ first came to public notice.8 This assemblage comprises some 30 kg. of coin mould fragments, 17 kg. of associated pottery, around 2 kg. of bone, and some fragments of white stone. It is the second largest single find of coin mould ever made. It was found, allegedly in 1999, by an anonymous amateur under circumstances that remain unclear. An unknown quantity of the material was sold on eBay, but the bulk of the material was purchased from the finder by Chris Rudd, who commissioned the full evaluation which forms the basis of Chapter 6 in the current work.

However, the Ford Bridge site was not the first in Braughing to yield pellet mould.

Taking all of these assemblages together, the Braughing/ Puckeridge settlement becomes the largest9 known centre for the production of coin pellets in the whole of Europe, surpassing even Old Sleaford.10

The first assemblage to be found, the Henderson Collection, was unearthed at some point between 1935 and 1960. It comprises 64 fragments of coin mould, many of them small and abraded. Sadly, the finder has left absolutely no record of the context of the find, and only the vaguest indication of its location. A brief account, together with a short report by Craddock and Tite on the XRF analysis of the fragments for metal residues, is included in Partridge, ‘Skeleton Green’.3 Accounts which are in some respects more detailed are included in ‘Report on the Scientific Examination of Iron Age Coin Moulds’4 and ‘The examination of refractory ceramics from metal-production and metalworking sites’.5

It was felt that an assemblage of the size of Ford Bridge should receive proper evaluation, and that a comparative study should be made with the other pellet mould retrieved. However, it was discovered that the literature on the manufacture and use of pellet mould is sparse, and that this was a reflection of the small amount of primary research that had taken place. As a result, for many years there has been no real progress in the subject. Claims, contentions and controversies remain unsubstantiated or unsettled since they were set out by Elsie M. Clifford in her 1960 work ‘Bagendon: A Belgic Oppidum’,11 and little of the continuing debate has been data-driven.

Two small deposits of coin mould were discovered during the course of rescue excavations at Gatesbury Track6 and Wickham Kennels.7 Although most, if not all, of the Wickham Kennels material remains available for examination at Hertford Museum, it has proved impossible to locate the Gatesbury Track assemblage. Other finds from this excavation are held at Hertford

There are three main reasons for this stagnation. The first is that the study of pellet-mould morphology has been almost completely ignored. Instead, attention has been almost exclusively focussed on testing the material for metal residues. Why this should be is not clear. Were one to be facetious, one might suggest that the glamour and glitter of precious metals has perhaps bedazzled and distracted. However, even in a more serious vein, this

Hunn, J, 2007. Longden, H, 2009. 3  Craddock, P. and Tite, M. in Partridge, C. 1981: p. 326 4  Tite, M. and Freestone, I, 1983. 5  Tite, M, Freestone, I, Meeks, N, Craddock, P, 1985: p. 54. 6  Partridge, C, 1979: p. 99 and Freestone, I. in Partridge, C, 1979: p. 129. 7  Partridge, C, 1982: p. 41 and Cowell, M. and Tite, M, in Partridge, C, 1982: p. 57. 1  2 

Rudd, C, 2008: pp. 30 – 31. By weight. 10  In terms of weight: Braughing/Puckeridge coin mould is significantly thicker than that from Old Sleaford. 11  Clifford, E, 1960: p. 144 – 147. 8  9 

1

Making a Mint

Figure 1.1: The Ford Bridge mint trench.

Figure 1.2: In situ coin mould at Ford Bridge.

preoccupation is strange, not least because of the price differential: to test 1 kg. of mould for metal residue will cost more than the full morphological evaluation of 14 kg. It becomes stranger still when one considers that nobody has disputed that pellet mould was made for use in a process involving molten metal. Yes, it is interesting and useful to know exactly which metals are associated with a particular assemblage, but this is only a single aspect of a class of artefacts which were part of a process which was both technologically and socially complex. The testing of pellet mould for metal residues on its own can never address many of the questions surrounding the stuff itself.

proceeded methodically and carefully in the collection of data), at worst, nearly non-existent – to the point where it has been impossible to carry out even the most basic comparative study, because measurements have not been taken, or have not been taken using compatible methodologies.

A second, and perhaps greater, obstacle to progress has been the absence of an agreed protocol for recording pellet mould. The collection and recording of morphological data has for the most part been patchy and unsystematic (although Collis12 and Bayley13 both

The third reason has been the relative paucity of the material: little more than 200 kg. of coin mould has been found in the whole of Europe, spread between Štaštín near Bratislava in Slovakia and Bagendon in Gloucestershire, England. In the course of the present study, some 45 kg. – almost 25% of all the coin mould ever found – has been examined.

12  13 

In consequence, one of the principal aims of this study has been to evolve a recording protocol for pellet-mould designed to facilitate comparative work, and to address the major issues in the study of the manufacture and use of pellet mould.

Collis, J, in Tournaire et al. 1982.: pp. 422 – 423. Bayley, J, 1979: p. 1.

2

Chapter 1 : Starting point Nonetheless, it should be made clear that this book is not the final word on the subject. It had been hoped originally to integrate this study of the supramicroscopic morphology of coin mould with a comprehensive programme of electron microscopy and electron dispersive spectroscopy, to be carried out by Henrietta Longden MA, then of Liverpool University. Unfortunately it proved impossible to obtain funding in time, and this book is much the poorer for it. Indeed, given the current ‘consumption of the purse’ afflicting academe, it is all too possible that this work will never take place.

Made of clay, these moulds fall into three main types: ‘potin’ moulds, in which strips of coins joined together by sprues were cast at once, complete with the design, by pouring molten metal into holes linked by channels,17 and which seem in Britain to be associated with the earliest episodes of minting; ‘potsherd’ moulds, where mould-holes have been bored into a fragment of prefired ceramic,18 which seem to be associated with very smallscale coin manufacture; ‘pellet mould’, in which small quantities of metal were melted to make the precursors to coin flans,19 although this account has not been universally accepted.20

Moreover, although coin mould from eight different assemblages has been studied in this work, a far larger quantity remains unexamined. Only a minute sample of the very large Old Sleaford assemblage has been subjected to the techniques evolved during this study, many other assemblages have received no attention at all, and more are coming to light at a rate of around two a year.14 Perhaps more seriously, no comparable work has been undertaken on coin mould from mainland Europe, and so it is not yet possible to evaluate the similarities and differences across the whole area in which coin mould was used. It is to be hoped that the recording protocol which underpins the research on which this work is based will, in due course, contribute to the resolution of these uncertainties.

It is this last type of coin mould which has been the main focus of the study which underpins this work. It is also, by a very large margin, the most common type of coin mould found. In the most general terms, pellet mould was made by creating many cup-shaped depressions in a slab of wet clay 10mm – 28mm thick,21 itself usually formed in a mould and then fired. When completed, the slab is known as a ‘tray’. Several different forms bearing varying numbers of mould-holes are known.22 Some specimens appear to have been tempered with powdered charcoal,23 some with vegetable matter,24 others with shell and crushed chalk; yet more have had the inner surfaces of the mould holes coated with calcium carbonate.

In conclusion, far more information has been collected in the course of this study than has actually been used. It is to be hoped that this unused material will prove of use to future workers in this field: it will certainly be made available to anyone who wishes.2 What is coin mould?

Experiments in casting coin pellets using this type of mould25 have assumed that measured amounts of metal were placed in each hole; that the tray was then placed on a bed of charcoal, and that charcoal was then heaped over the holes. The temperature necessary to achieve melting was created using a tuyère. It is then assumed that the resulting pellet was again heated (annealed) and beaten into a flan, and then possibly reheated before being impressed with a design by being struck between two dies.

Coin mould has been found on sites across Europe. Collis, in his list of 1982,15 mentioned find-sites in Austria, Belgium, the Czech Republic, England, France, Germany, Luxemburg, Poland and Slovakia. Most of these discoveries have been made within the last hundred years, but the earliest recorded find of possible coin mould was apparently made in the eighteenth century at Haverhill in Suffolk16 - a ‘clay box’ containing coins is mentioned. It should be noted that England has so far yielded 23 assemblages, more than any other country in Europe. Possible reasons for this will be examined later in this work.

By no means all of the pellet mould retrieved shows to the naked eye obvious signs of use, here to be understood as the effects of extreme heat such as vesiculation and vitrification. On the contrary, many specimens Debord, J, 1989:p. 12. Collis,J, in Tournaire et al, 1982: p. 433. 19  Tylecote, R, 1962: p. 102; Collis, J, 1985: p. 237; Haselgrove, C, 1987: pp. 28 – 29. 20  Sellwood, D, 1980: pp. iii - vii; Casey, J, 1983: pp. 358 – 360. 21  Most specimens thicker than 30 mm tend to be puffed up with vesiculation, although a possible fragment of mould from Micheldever Wood Banjo Enclosure (Fasham 1987, Fig 34, C.5) is around 50 mm thick (inf. Dr T. Moore of Durham University) (Although Dr Leo Webley of the University of Bristiol disputes this identification). 22  See below, Chapter 3. 23  Tylecote, R, 1962: p. 102; Longden, H, 2008: p. 17. 24  Longden, H, 2008: p. 17 and Fig. 10, p. 18. 25  Tylecote, R, 1962: p. 102 and De Jersey, P. (with Burridge, N.), 2007: pp. 261 – 262. 17  18 

During 2014, a deposit of coin mould was reported from Switzerland by Dr Roy Trittschack of the University of Fribourg (Trittschack, R., 2014: pers. comm.), and a major deposit from Blackfriars, Leicester was examined by the author of this work on behalf of Wardell Armstrong Archaeology (Landon, M, 2015: unpubl. report). In March 2015 came the startling discovery of coin mould at Scotch Corner on the line of the widened A1 by Northern Archaeological Associates (Landon, M, forthcoming), which is forcing the reappraisal of so many of our ideas about the chronology and geographical extent of the use of coin mould. 15  Collis in Tournaire et al. 1982, Appendix I: p. 433. 16  Archaeologia, xiv, 1882, 72; cited in Elsdon, 1997. 14 

3

Making a Mint which have given positive results for metal residue have shown no evidence of extreme heating beyond a blackening of surfaces.26 This stands in stark contrast to the experimentally derived finding of Gebhard his collaborators,27 that ‘The surface usually shows a noticeable degree of vitrification’. Of those specimens which do show clear signs of having been subjected to extreme heat, the majority appear to have been heated on the upper surface only, a significant minority show these traces on both upper and lower surfaces, and a very few examples have been heated to such a degree that the fabric of the mould has begun to slump. It has been assumed by many that there is a direct link between the size of hole on a given fragment and the denomination of the coin to be produced from the blank cast in it, but this has not been demonstrated. Indeed, few subsequent accounts betray any awareness of the implications of the warnings of both Clifford28 and Tylecote29 about the variability they observed in the conformation of mould holes on a single fragment.

Figure 1.3: Presumed method of using pellet mould. 1. Charcoal bed; 2. Pellet mould; 3. Granulated charcoal heaped over hole; 4. Tuyère; 5. Granulated metal.

We know that people of the Iron Age imbued much of what we would regard as mundane and secular activity with symbolic significance and quasi-religious numen,31 and this clearly extended to coinage as well: not only are the designs upon the coins themselves freighted with symbols the meaning of which can often only be guessed at, but we can tell from the context from which many coin-hoards have been retrieved that the coin itself had a religious function. Given the relative complexity of the process chosen to make these coins, it would be unsurprising to discover that the process of manufacture was in turn enriched with symbolic meanings, and that the cultural and conceptual framework of the society producing the coins had influenced significantly the technique of minting. That the evidence suggests the continuance of this tradition until the last decades of the Roman occupation indicates, perhaps, the degree to which native British thinking remained unaffected by Romanitas – a useful counterpoint to the sneers of Tacitus.32

It should be noted that there is considerable variation in the material in several parameters. For this reason it would seem unwise to reason by analogy from one assemblage to another until it has been proved on the material that the analogy is valid. Finally, the minting of coin is not an activity that takes place in a social or economic vacuum. Money is the physical expression of a mutable conceptual construct peculiar to the social unit producing it. Ideas of value and worth, the projection of power and the expression of prestige can govern the choice of metal and the design upon the coin, while the economic context might decide the weight and precise composition. The function of money is not a constant. It is constantly defined and redefined by the culture within which it is used. Minting in the British Late Iron Age marks the start in these islands of a long road that leads ultimately to ideas of monetary value and function so complex and so abstract that it is impossible to generate a single, coherent definition. If we can begin to understand this initial point, we take a step towards understanding subsequent developments. That we now have good evidence that the use of this technique for minting persisted into the later C4th AD30 would seem to suggest that native ideas about minting (and hence the coin itself) continued as the dominant local tradition right through the Roman period.

If we start with the coin mould itself, note its archaeological context carefully enough, examine it closely enough, and consider it minutely enough, if we can reconstruct the process of which it was a part, we may begin to see points at which native British conceptualisation differed significantly both from the Roman and from our own, and from these hints perhaps glean some insights into the rich interconnectivity that informed areas of this vanished world of the mind.

Craddock, P. and Tite, M. in Partridge, C, 1981: p. 326. Gebhard et al. 2007, cited in Longden, H, 2008: pp. 9 – 11. 28  Clifford, E, 1960: p. 144. 29  Tylecote, R, 1962: p. 101. 30  Ponting, M, 2015: pers. comm. ‘...this method of blank production is very ‘Native’ and not at all Roman. But I also think that it continues as the preferred method of blank manufacture at least up to the period of the so-called Barbarous Radiates – I am pretty certain that the blanks in the Fenny Stratford hoard were produced this way.’ 26  27 

31  32 

4

Collis, J, 2008: pers. comm. Tacitus, Agricola, 21

Chapter 2

The Literature The literature on pellet mould falls into two categories. The more numerous of these categories is the excavation report, usually brief in the extreme, and often not very informative. The second category is much smaller, of analytical studies and specialist reports which have been intended to advance or summarize theories about the manufacture and use of pellet mould. The more significant of these are examined in detail below, in chronological order.

A more whimsical contribution is their use of the term ‘waffle’ to denote a mould-tray. 2 Clifford, Elsie M, 1960: ‘Bagendon: a Belgic Oppidum, Appendix VII: The Coin Moulds’, Heffer and Sons, Cambridge, pp. 144 – 149. Elsie M. Clifford, the first in Britain to deal with mould in any depth, includes an astute consideration of the topic in ‘Bagendon: A Belgic Oppidum’, pub. 1960. Not only has much of what she said become part of the accepted doctrine of pellet mould studies, but her experienced eye also picked out several important considerations overlooked by many subsequent writers.

1 Richards, E E and Aitken, M J, 1959: ‘Spectrographic and Magnetic Examination of Some Baked Clay SlabMoulds’, Archaeometry Vol. 2, pp. 53 – 7. This short piece consists of two parts, the first a very brief description of the material (‘...holes (varying in depth and diameter between ¼" and ¾")...’), followed by a list of British find sites (Camulodunum, Needham, Silchester, Verulamium and Bagendon), and a detailed exposition of their methodology. The second part lists their results by site.

She comments particularly on the irregular shape of the ‘cups’: ‘A few of the indentations, while certainly not rectangular, are squarish at the mouth of the hole, but are round at the bottom. Some of the indentations are well made with a horizontal bottom, but in most cases no great care has been spent making them and any blanks cast in them must have been uneven and erratic in shape. The holes taper slightly to the bottom.’

One batch of 15 samples from Bagendon, and two batches from Verulamium, one of 17 samples, the other of 10 samples, were tested.

This is an important observation, and it has been ignored by many later commentators.

Using residual magnetism, they are able to show that all the fragments tested were last subjected to heating with their holes uppermost. This result is important because it proves that none of the trays was used as the top half of a bivalve mould (as suggested by Jean Debord, Revue Numismatique 1989 Vol. 6 Issue 31, ‘L’atelier monetaire gaulois de Villeneuve-Saint-Germain (Aisne) et sa production’).

She also observes of the Bagendon material that the holes were arranged ‘chessboard fashion’ or else in diagonal lines, and that the trays were ‘always or almost always rectangular’. She notes that other tray shapes are known, citing a pentangular form from Verulamium, and a possibly hexagonal form from Colchester (although examination of photographs of the Colchester material suggests that this may be a misinterpretation of the pedimental apex of a Verulamium pentagonal tray).

Using a Hilger Large Quartz Spectrometer, they are then able to demonstrate the presence in various samples of coin mould of different combinations, and in varying proportions, of copper, silver, gold, tin, zinc and lead. However, the method they used was insufficiently sensitive to generate accurate percentages, and so instead was calibrated ‘high’; ‘medium’; ‘low’; ‘trace’; ‘not detectable’, with relatively wide margins of error for each metal. This relative scale makes problematic the comparison of their results with results obtained by the more sensitive techniques available today.

She mentions two sizes of hole, a smaller averaging 0.5", and a larger averaging ‘approximately’ 0.8", but she does not say whether these measurements were taken at the top or the bottom of the hole. She adds that the depth of the holes varies between 0.3" and 0.7". Clifford does not attempt to link the diameter of a ‘cup’ to a particular denomination of finished coin. Instead, she states ‘The sizes of the bottoms of the indentations... are in each case smaller than actual coins of gold or silver. This is as it should be, since the metal must have spread... when struck between the dies.’ – again, a point not always borne in mind by other recorders and commentators.

Their work, nonetheless, seems to have exerted some influence on their near-contemporaries, such as Craddock and Tite,1 who follow them in concentrating on the analysis of metal traces. 1 

In Partridge, C, 1981

5

Making a Mint Her salient hypothesis, derived from the work of Castelin,2 is the interposition of a process intermediate between the casting of the pellet and the striking of the coin, on the grounds that the edges of what she terms ‘Belgic’ coins bear no trace of the irregularities she has noted in the mould holes. This idea is repeated by most commentators, and is generally termed ‘annealing’. It is certainly an area in which the morphological analysis of coin mould has little – if any – part to play.

pencil into moist clay’ – in other words, the method used to make the holes in the samples he studied was informal and, as he makes clear elsewhere in the paper, inherently variable. He then observes that, although most of the holes have flat, rather than tassiform, bottoms, the coins and blanks which are evidently coeval with the coin mould ‘appear to have been struck from globules not cylinders’. This is a point vital to our understanding of the process employed to manufacture coin using clay mould trays: ‘how a spherical globule could be made in a cylindrical hole, or even in an irregular hole’, given that ‘an examination of the blanks suggests that they were made from spherical globules rather than cylinders’. Possibly driven by a need to explain the ‘bun’ shape of the trapped pellets found at Verulamium and Old Sleaford, he qualifies this conclusion by claiming that ‘It does not seem that spherical globules were always made, and it is possible to produce a flat blank by striking a cylindrical piece of metal provided the height/diameter ratio is not too great’, despite stating that there is no empirical evidence that coins were actually struck from cylindrical pellets.

In addition, she goes to some lengths to explain why she believes that pellet mould was used in the process of minting coin: ‘I had already come to the conclusion that the connection between the coins and the moulds was neither direct nor simple, and that the moulds might well have been used, not for making actual coin blanks, but for a purpose such as controlling the mixture of alloys used in coin making’. She then cites the work of Castelin as having convinced her that pellet mould was an integral, rather than ancillary, part of coin manufacture. He attempted to demonstrate that the metal was melted in the mould cups using a tuyère and bellows, rather than poured in the molten state. All of these ideas reappear – often uncredited – in later work.

In fact, his own experiments indicate a more likely resolution than the positing of lost coin series: the ‘bun’ shaped pellets found trapped in samples of coin mould represent instead a failure to maintain reducing conditions during heating: ‘an oxidizing flame caused the copper to effervesce on cooling and to stick to the sides of the mould’. This would imply that the trapped pellets are not typical or intentional, but are rather mistakes – and their extreme rarity a tribute to the skill and consistency of the native British smiths.

But it is to her contention that ‘the connection between the coins and the moulds was neither direct nor simple’ that attention should be directed, for this is a matter in which morphological study has a large part to play. It also stands in stark contrast to the almost automatic assumption of the contrary to be found in many reports dealing with finds of pellet mould. 3 Tylecote, R.F, 1962: ‘The Method of use of Early IronAge Coin Moulds’, Numismatic Chronicle, Seventh Series, Vol. II, pp. 102 – 109.

Tylecote demonstrates conclusively that the need to maintain a reducing atmosphere applies to copper and tin, not to gold or silver. Gold does not oxidize, and silver oxides act to reinforce the action of surface tension, while oxides of copper and tin do not, and furthermore have a tendency to ‘wet’ the mould, fusing with its fabric.

Tylecote reports on experimental work carried out to discover the means by which metal was introduced into the cups. He concludes that solid lumps, rather than granulated or powdered, metal were placed in each hole, perhaps covered with charcoal. He claims that powdered or granulated metal could not be used, because the blast from the tuyère tended to scatter it.

This is plain evidence that the manufacture of coin using clay pellet mould cannot be regarded as a single, homogenous technique: different metals have different requirements which affect the details of the process. He suggests that there is evidence that the method used by the British smiths to maintain a reducing atmosphere during the production of copper alloy pellets involved both the addition of powdered charcoal to the clay of the mould, perhaps as much as 22% by weight, as well as the heaping of charcoal over the mould tray during heating. However, as will be shown below, this is certainly not the only method used.

The importance of this well-known paper has not always been fully appreciated by subsequent writers. Cottam3 considers that Tylecote is ‘the only one who knows what he is talking about’, and it is hard to deny that Tylecote’s application of classic scientific method to the subject has given his conclusions considerable force. He begins with a brief description of the material, noting that mould holes ‘often look as though they had been made by pushing a round stick about the size of a wooden 2  3 

Tylecote conducted eight experiments using different casting techniques and mould materials, showing that very few combinations of material and technique are able

Castelin, K, 1960, summarised in Germania 38, pp. 32-42 Cottam, G, 2008: pers. comm.

6

Chapter 2: The Literature raw data it contains, its methodology, and the conclusions to which it comes.

to produce spheroidal globules of copper with consistency regardless of pellet size. This is a good indication that his conclusions about the process used cannot be far wide of the mark: had more of his experiments generated acceptable results, we should have been unable to choose between a multiplicity of variant processes.

There are seven main sections in this work. The first three deal with assemblages of coin mould from Levroux, Aulnat and Roanne. The fourth section summarizes these findings, and in it a number of conclusions are drawn about the minting process in Late Iron Age France. This is followed by three appendices: the first, a list of all coin mould sites then known in Britain and Europe; the second is a list of both sites then known where in situ pellets had been found; the third, a list of results of all testing for metal residues then carried out on samples of coin mould. The copious illustrations should also receive mention, as they are an important resource in their own right. Each section of the text will be reviewed in turn.

As an interesting postscript to these experiments, while examining a box of coin mould fragments from Old Sleaford which had found its way into the British Museum,4 the author of the current work discovered a battered cardboard box deposited unceremoniously amongst the bags of coin mould. A note on the lid stated that it contained the fruits of Tylecote’s experiments, and that it had been presented by John Casey. Inside were the two fireclay strip moulds illustrated in Figure 3 of Tylecote’s paper, together with the imperfect spheres cast in them and some cut-up fragments of Iron Age coin. It seemed to this author a less than fitting memorial to such an important piece of work.

Section 1: Finds from Levroux, 1977 and 1979 Subsection 1 – Introduction by O. Buchsenschutz Buchsenschutz gives a brief resume of the nature of the site and the contexts in which the coin mould fragments were found, and closes with a short summary of the main conclusions reached.

4 Sellwood, D, 1980: Numismatic Chronicle XX, pp. iii-vii. Casey, J, 1983: Britannia xiv, pp. 358-60. Sellwood concludes on the basis of the edge characteristics of finished coin that they could not have been made in pellet moulds, and feels that the number of finds of pellet mould is far greater than the possible number of mints operating in Britain. He suggests instead that flans were produced by pouring molten metal onto a flat surface, on the basis of his own experiments as well as on the existence of the ‘MOTVIDIACA’ coins. He managed – with practice – to achieve an accuracy of ±0.5g in the production of 17g pellets, and feels that - with more practice - he could have improved his performance. As an alternative to the production of coin pellets, he echoes the theory discarded by Clifford, and suggests that pellet mould was used to make pellets of consistent weight and composition for use in the production of alloys.

Subsection 2 – Description of the moulds by J. Tournaire The bulk of this subsection is taken up by a detailed catalogue of the whole assemblage, piece by piece. Unfortunately, Tournaire has not maintained a consistent recording protocol, which makes the comparison of fragment data difficult and, in some cases, impossible. These discrepancies extend beyond the minor - for Frag. V83 -010, we are given figures for incomplete and complete holes, whereas for Frag. V83 - 011 we are told only that there is one complete hole. For both of these fragments we are given the volume of the complete hole, but not for Frag. V83 - 027, which has two complete holes; and not for Frag. V.44, which has four. Measurements for hole base diameter are given for only two of the four fragments bearing complete holes.

Casey cites Sellwood’s work as ‘A truly important contribution to the study of these artefacts... which demonstrates once and for all, that coins were not made in these moulds’.

Tournaire concludes this subsection with a rather odd statement. He says that ‘it is possible to distinguish two types of mould differentiated by the diameter at the lip of the hole’, and states that one has a top diameter of 9 mm, the other, 14 mm. He agrees that this ‘is not logical’, but fails to give any evidence in support of his contention. Instead, he suggests that, since ‘the difference between lip and bottom diameters is not great (1 - 2mm)’, it ‘does not materially affect our classification’. In other words, it does not matter which measurement is used.

5 Tournaire, J, Buchsenschutz, O, Henderson, J. and Collis, J, 1982: ‘Iron Age Coin Moulds from France’, Proceedings of the Prehistoric Society 48, pp. 417-435. This jointly-written piece has become perhaps the most influential work in the field of coin mould studies in Britain as well as in France. The ideas that are put forward in it appear in almost every subsequent treatment of the subject, sometimes uncredited, sometimes misunderstood, but almost always without question. It should therefore be examined in detail, in terms of the 4 

In fact, there are a number of flaws in his approach which raise serious doubts about the validity of his conclusion. The most obvious of these is that the two fragments for which he gives base as well as top diameter measurements

See Chapter 10, below.

7

Making a Mint show a difference of 3mm between top and base, rather than the 2mm he claims. The significance of this error for his theory will be examined later.5 The shortcomings in his reasoning actually begin at a conceptual level, because his arguments rest on two unspoken - and hence unexamined - assumptions. That he has not recorded the data that would enable proper evaluation would seem to indicate that Tournaire did not even articulate these assumptions to himself.

which, as he acknowledges in section 4 (see below), was both the desired shape of pellet and the normal behaviour of molten silver, and which would require no direct relationship between the pellet and the wall of the hole in which it was formed. Although in theory all of these deficiencies are remediable, the paucity of the material means that the retrievable data set, even if it had been retrieved, would be statistically insufficient for the resolution of the significant problems and doubts affecting his conclusion.

The first assumption is that there is a predictable relationship between the diameter at the top of a hole and the diameter at its base, and that the range of any variability in this relationship is not of a significant magnitude.

Subsection 3 – Analyses of the Levroux moulds by M.J.R. Bourhis The Levroux mould fragments were subjected to analysis for metal residues using arc spectrophotometry, and the results tabulated, showing that the assemblage was used in the smelting of alloys in which silver was by far the greatest constituent.

In order to test this, one must first be able to answer the question of where this variability - the existence of which Tournaire acknowledges, but understates - manifests itself: at the lip of a hole, at the base, or do both exhibit a degree of variation? To resolve this would require a systematic record of both base and top diameters for all holes from which this data is retrievable on each fragment, but this record has not been kept.

Subsection 4 – Petrographic analysis by Thierry Odiot Odiot states that the clay from which the Levroux trays were made derives from the Eocene Bois Bezard complex, situated a few kilometres from the find site. He says that this clay is composed mostly of kaolin, and is highly refractory, and that this property probably influenced the choice to use it.

One must next be able to say in what way this variability is manifested - is it always negative, downward from a target value, or can it sometimes be positive? If, as is usually the case, actual values cluster around a target value, what are the plus/minus parameters of this range of variability? Once again, the absence of the necessary data prevents consideration.

The importance of this observation cannot be overstated, because it is a clear indication that the use of clay mould in the production of coin cannot be regarded as a single, homogenous tradition. The use of a highly refractory medium for the Levroux trays means that vesiculation of the tray fabric was not nearly as great a problem for the metalworkers at this site as it was for their British counterparts: the use of refractory clays has never been noted of any British coin mould. This would have had significant effects on the detail of the process of use.

Finally, one must be able to give a convincing account of what constitutes ‘significant variation’. The most obvious definition of significant variation would be ‘variability such that it would be possible to confuse mould intended for one coin module with mould intended for another’. A ± 2mm range of intra-fragment variation - as derived from experiments in hole-making - in both top and bottom diameters shows that the ‘two modules’ claimed by Tournaire would not, in practise, necessarily be distinguished with ease. The 2mm range of variation in the difference between top and base diameters recorded, but not noted, is certainly of a magnitude sufficient to warrant examination. Instead, it is misstated and then dismissed.

The Levroux mintmasters would not have had to operate under the same imperative for speed as in Britain, because they did not have to address the possibility of the ‘meltdown’ of a tray were extreme heating prolonged beyond the minimum duration necessary to achieve fusion of the metal into a pellet. The problem of fusion of pellet with tray fabric would also have been less serious for them, with the result that the maintenance of reducing conditions would not have been the absolute priority that it was for the British smiths.

The second assumption underlying Tournaire’s ideas about ‘module’ and the significance of top diameter in its determination is that there is a necessary and predictable relationship between the diameter of a hole and the diameter of the pellet to be cast in it. This would seem to imply that the mould was being to used to cast cylindrical pellets, which would require such a relationship, rather than to assist in the formation of ‘spheroid’ globules 5 

One consequence of these facts is that we cannot rely on analogy from minting in one area to explain minting in another, unless we first demonstrate homogeneity of practise in the two areas.

See Chapter 3, §6.

8

Chapter 2: The Literature (only one hole diameter in the whole assemblage of 25 fragments is not given as approximate). These findings are confirmed by the drawings provided: not a single hole in the assemblage has been preserved intact.

Subsection 5 – A mint at Levroux by J. Tournaire Tournaire attempts in this Subsection to establish a link between the Levroux coin mould and issues of coin found in the vicinity. This has never been satisfactorily accomplished elsewhere, since preferential leaching from metal residues in coin mould has a significant effect on the precise proportions obtained by analysis, but Tournaire does not mention the results obtained from the analysis of the Levroux mould. This may be because he is aware of the problem of preferential leaching, or it may be that none of the finds of coin he mentions in the text had themselves been analysed. Whatever the reason, he limits himself to attempting to establish the link between mould and issued coin by morphological means alone.

Wisely, the writer avoids any discussion of diameter and module. Instead, there is an imaginative approach to a resolution of the question of tray form and possible numbers of trays represented in the assemblage. First, the writer attempts to gauge the minimum number of trays from the differing thicknesses of the fragments. It is acknowledged that this is far from certain, but the suggestion that this could be confirmed by more accurate measurements, were it possible to obtain them from the material, is not correct. Experimentally derived evidence shows that all the holes save one fall easily within the range of variation arising from the use of a single dibber. Experiment also shows us that, even when every attempt to control hole depth has been made, holes across a slab can vary substantially and unpredictably in depth. We know from Tournaire and others that thickness can vary perceptibly across a single tray: could it be that the fragments from Aulnat are simply too small to exhibit this variation? On balance, therefore, the figure provided must be regarded very much as hypothetical.

This would seem a forlorn hope from the outset, in that the spreading of the flan during striking would have been perforce irregular, since the force with which the malleator wielded his hammer would have varied from stroke to stroke, and - as he grew more tired - more generally through the course of a striking session. Tournaire’s case is not helped by his failure to acknowledge the intermediate process between pellet and coin posited by Castelin,6 and supported evidentially by the hammered edges exhibited by many ‘Celtic’ coins.

The method used to try to diagnose the form of the trays from which the fragments of the Aulnat assemblage derive is considerably more sophisticated. Using the dimensions of the assemblage fragments as a template, the writer calculates how many corner, edge and middle fragments of this size would result from the breaking of square trays of varying capacity (2 x 2; 3 x 3, and so on). Expressing these theoretical figures as a ratio, the writer claims that the actual ratio of edge to middle fragments (there are no corner fragments in the assemblage) most nearly resemble the theoretical ratio for a 7 x 7 tray, ‘similar to the only known complete example from St. Albans which was 7 x 7 with an extra hole to make 50’. The problem with this is that the St. Albans tray was pentangular, not square, with five corners and five edges - and hence a very different ratio of corner:edge:middle fragments. The rectangular 6 x 10 tray from Saintes (Plate 31b. in the text) would perturb this method in much the same way, and the possibility that more than one tray form might be represented in an assemblage7 adds a new dimension of uncertainty to the equation. Unless the tray-form is known beforehand, the ratio of corner:edge:middle fragments from a complete tray in a real assemblage must be purely conjectural.

Section 2: Coin moulds from Aulnat-Gandaillat Subsection 1 – Introduction (anonymous) A short summary of the history of the excavations at this site is given, followed by an equally brief account of the contexts in which the coin mould was found. Subsection 2 – Description (anonymous) The technique adopted in this Subsection for recording coin mould is a very substantial improvement on that used for the Levroux assemblage. It uses a standard recording protocol, and tabulates the results, thus making the comparison of data between individual fragments quick and easy. Although the protocol adopted is not particularly detailed (we are not given separate totals for complete and incomplete holes, and measurements for individual holes on a fragment are not given), and one of the categories is not accurately defined (we are not told whether diameter measurements were taken at lip or base of a hole), it does make clear which measurements are approximate, and does enable conclusions to be drawn beyond those given in the text. For instance, even without looking at the illustrations provided, it is possible to tell that the fragments are very small (none has a dimension greater than 25mm, and none has traces of more than 4 holes), and that the state of preservation of the holes is not good 6 

In fact, the writer himself adumbrates yet a third way in which the composition of a retrieved assemblage might not derive from the original form of the trays from which the fragments derive. When he points out the scarcity of

Castelin, K, 1960, summarised in Germania 38, pp. 32-42

7 

9

See Chapter 5, §2 and Chapter 6, §2, below.

Making a Mint corner fragments in the Aulnat material, he is the first to notice a surprising fact, one which this assemblage has in common with many British coin mould assemblages. Elsewhere in the current work,8 it is suggested that this shortfall so often seen in larger assemblages is the result of the selective removal of corners for deposition elsewhere. Selective removal would mean that there could be no guarantee that the corner:edge:middle: ratio in the retrieved assemblage had any necessary relationship to the ratio in the pre-depositional assemblage.

Subsection 3 – Analysis of the Roanne fragments by J. Henderson

Subsection 3 – Analyses of the Aulnat moulds by J. Henderson

Collis suggests that the lack of either metal residues or vitrification means that Frag. B is either unused mould, or not coin mould.

Although Henderson does not say so, it is to be presumed that he used XRF to carry out the analysis, as the results are presented using the same numerical coding as the for the Aulnat assemblage. Only Frag. A yielded positive results, for copper and lead. Subsection 4 Discussion by J. Collis

Henderson tells us that the mould fragments were examined by means of X-Ray Fluorescence, giving details of the equipment and the method in which it was used.

Section 4: The Method of Using the Moulds Subsection 1 – The manufacture of the moulds by J. Tournaire

His results are then tabulated and discussed. He emphasizes that there are good reasons why these results might not be an accurate reflection of the original composition of the alloys smelted in the various fragments. He also devotes space to a demonstration that purely ocular examination cannot definitively diagnose which metals might be present as traces in a specimen of mould.

Tournaire begins this subsection by attempting to define the difference between the type of coin mould discussed in this work, and the type of coin mould used to make potin issues, although he does not inform us that potin issues were cast in strips, while the pellets cast in this mould were cast singly, and therefore fails to mention the major morphological difference between the two types of mould: potin issues were made in moulds with little channels connecting the holes, while pellet mould has no such channels.

Section 3: Finds from the Institution St. Joseph Roanne Subsection 1 – Introduction by R. Périchon

He then states ‘The need to reproduce identical flans demands a certain precision in the moulding of the hollows in the tray where the flans are cast’, but he does not say why there might be a need to reproduce identical flans when, as his own data shows, there is inherent variability in the finished coin. Furthermore, by leaving undefined the term ‘a certain precision’, he renders it effectively meaningless: is the nearly 20% variation observed in the difference between top and bottom diameters to be defined as ‘precision’? Given Castelin’s ‘intermediate process’ hypothesis (to which the hammered edges of many ‘Celtic’ coins would seem to lend evidential force), which would - if true - make even more tenuous the link between pellet and coin, and hence between mould-hole and coin, it is hard to see why these questions have not been addressed, especially since Castelin is cited in the following subsection.

Périchon gives a brief account of the history of the excavations carried out the site, the nature of the archaeology and its chronology, and of the context in which the two fragments of mould were found. Subsection 2 – Description by J. Collis Although the data is not tabulated, as in the previous report, and the protocol is not explicitly stated, since there are only two fragments this is not a serious problem. For each fragment we are given the size and thickness, though we are told the number of complete holes only for fragment A, we are given a diameter (although not at what point in the hole it was taken, or whether it is an average or a single, sample, measurement), and a depth. The fabric is briefly described, together with the presence or absence of signs of extreme heating. For Frag. B, Collis describes the tray form (circular), reconstructs the original capacity of the tray (12 holes), notes the presence of hole slighting, and hypothesises that the holes were made with the end of a finger.

8 

Tournaire is at pains to explain the bowed cross-section observed on the Levroux material (and on much of the coin mould found in Britain). He sees it as having been caused by the moulding of the holes on a tray all at once, using a peg-board. He theorises that, in order to avoid deformation of the tray during the hole-making process, the peg-board was applied to the plastic clay while the clay was contained in a ‘rectangular mould’, and that this process would mean that there was more clay to be

Chapters 4, 5, 6, 7, 8 and 9.

10

Chapter 2: The Literature displaced from the middle of a tray than at the edges. He then cites ‘a ridge caused by the insertion of the matrix’ at the lip of each hole as further evidence of the use of a peg-board, since these ridges show no sign of deformation. Yet careful examination of the photographs provided of the Levroux assemblage reveals what appear to be two holes on Frag. V83 - 027 with a D-shaped outline, the result of one hole slighting another. Holeslighting will not occur if a peg-board is used.

There is no reason to doubt the accuracy of this account, but there is equally no reason to suppose that it is universally applicable to all sites where coin mould was used, and Tournaire and Henderson are explicitly aware of this. If, as they argue subsequently, the great variation in the depth of mould holes across a slab means that the metal must have been weighed before it was placed in the hole, it is hard to see any need for an aequetor (with his file) once the pellet has been cast. Moreover, the account does not make any mention of weighing metal before smelting. Since the account is so precise about every other aspect of the process, it does not seem likely that this is simply an oversight.

Although without proper examination of the fragment in question this matter cannot be settled out of hand, it is clear that Tournaire’s demonstration of the use of a peg-board is not conclusive, and any of the ideas that flow from it - such as his claim that the edges of the box-mould in which the clay was supposedly contained may well have been proud of the clay in order to prevent the dibbers on the peg-board penetrating the clay slab altogether - must be seen as more speculative than evidentially derived.

We are left with three possibilities: first, that the metal was either not weighed, or at least not accurately weighed, before being placed in the mould, and that the aequetor routinely dealt with major corrections of weight; second, that the control of pellet weight was regarded as being of such importance that it was checked twice during the process; and third, that we have here an account of a process which differs significantly from the process in which the moulds described in this work were used.

In the final paragraph of this subsection, Tournaire makes a very important point: ‘other methods were used to produce moulds’. He notes two very different tray forms - a circular form (examples from Jublains and Roanne) on which the holes were scattered irregularly, and a form found at Saintes, on which holes have been made in ten regular rows of six, undeniably using a dibber with six protuberances in a line. This is further clear evidence that the use of coin mould in minting is not to be regarded as a unitary tradition, but as a group of similar or related techniques. This calls into question any tendency to rely on analogy between mould from different sites, and also makes more difficult any attempt to make valid general statements about coin mould.

It is not possible to decide between these possibilities without the establishment of standards of acceptable variation in coin weight for each site under consideration and, as we have seen, it is virtually impossible to establish a definite link between a particular specimen of mould and a particular issue of coin. However, it is interesting to note that the 22mm flans from Oizon cited in the text vary in volume from 532mm3 to 760mm3 - a variation of almost 50%. Either a possible disparity of a full unit’s weight - and perhaps more - between two random samples of three coins was considered acceptable, or these flans were awaiting the attentions of the aequetor.9

Subsection 2 – Casting the flan by J. Tournaire and J. Henderson

Tournaire and Henderson then proceed with an evaluation of the experimental work and theories of pellet manufacture of Castelin, Metzler and Tylecote, stating that the precise method used by the Iron Age smiths remains hypothetical.

This subsection opens with an examination of the textual evidence for the organization of the process of minting using pellet mould. A source for the Gallo-Roman period describes a complex and minute division of the task into five apparently distinct categories of specialization: the signator, who supervised the operation; the flatuarius, who cast the pellet; the aequetor, who corrected the weight of the cast pellet by filing; the suppostor, who (using tongs) placed the reheated pellet between the dies; and the malleator, who wielded the hammer.

Their point that vitrification would result from the mixing of silicates in the clay with an alkali, possibly deriving from wood-ash, is well made, as is their additional point that the presence of vitrification indicates a localized concentration of heat. They are emphatic that the frequent occurrence of vitrification on the upper surfaces of fragments, its infrequent occurrence on the edges, and total absence on the base of fragments indicates that heat was concentrated on the upper surfaces of trays: the temperature of the tray as a whole was not raised to smelting levels.

This is a valuable account, but it should be used with care. There is good reason to concur with Tournaire’s thesis that ‘other methods were used to produce moulds’, and good reason also to extend the scope of this statement to cover the entire process: that more than one method was used to mint coin using clay moulds.

See Williams, J, 2005: pp. 126-127 for an alternative theory – that coin weights were determined by obtaining a fixed number of coins from a given weight of metal. 9 

11

Making a Mint They are aware that oxidizing conditions during smelting can cause copper and tin to fuse with the fabric of the mould, but a lack of systematically gathered data about the presence and extent or absence of the signs of oxidization prevents a proper consideration of this. The related question, whether (and, if so, how) reducing conditions were created to avoid the problem of fusion, receives no consideration at all.

mould trays were each used only once, since they had to be smashed in order to extract the pellets.10 The claim that the edge characteristics of finished coin are incompatible with flans having been produced in a mould is countered by a restatement of the intermediate process theory derived ultimately from Castelin: one would not expect that the finished coin would show characteristics derived from the mould in which it was cast if it had been reheated almost to melting point and then substantially reworked before being struck.

Conclusions It is stated that the analyses of the Levroux and Aulnat assemblages appear to confirm their use in coin production. Negative results from Roanne Frag. B, Hod Hill and perhaps Tuchlovice may indicate a different function for these pieces. It is considered that the Scotton fragment may in fact be coin mould, despite the lack of metal traces, as by no means all the material from Levroux and Aulnat yielded positive results. It is unclear why this should not apply to the Roanne, Hod Hill and Tuchlovice material.

However, Collis feels that the strongest counter-argument to the Sellwoood/Casey position is to be derived from the evidence of metal residues in mould holes. He notes that many mould fragments have been shown to have traces of either silver or gold alloys, and points out that, apart from their use in coinage, the incidence of precious metal alloys is extremely rare in the Iron Age. He concludes that the use of pellet mould in the production of coin remains the most plausible hypothesis. In the course of his discussion, Collis asserts that, as well as functioning as a crucible, the mould was also in some way used to measure the quantity of metal in each pellet. He speaks of the mould ‘being used to produce pellets of known and consistent size and weight’, and adds ‘In cases where calculations have been made of the likely weight of pellets based on the size of the hollows in the moulds, these seem to correspond closely with that of the contemporary coinage’, although he does not detail the method used to calculate first the volume of a mould hole (the calculation of the volume of what is, after Clifford, effectively an irregular – even erratic - polyhedron, is one of the crowning achievements of the classical Greek mathematicians, and is no simple matter), and then the weight of metal that it might have contained. To support his contention, he refers to work carried out by Ščasnár et al, Slovenská Numizmatika viii (1984), 121-45, on mould from Šaštín, which showed that the ‘weight of silver cast on one of the hollows would be 3.85g if filled to the top, or about 3.00g if filled to 1mm below the lip.’ However, this reasoning can apply only if metal was poured into the mould holes in a molten state.

It is claimed that the low percentage of finds from oppidum sites, and the presence of coin mould on relatively minor sites such as Aulnat and Štaštin, would seem to indicate that - as suggested by Collis in his 1971 doctoral thesis - the minting of coin was not perhaps the centrally controlled operation that many had assumed. 6 Collis, J, 1985: ‘Iron Age ‘Coin Moulds’, Britannia 16, pp. 237 – 238. Collis, replying to Casey in Britannia xvi, 1985, 2378, deprecates ‘over-emotive responses’ to the subject of Iron Age coinage. He points to a confusion between experimentally-derived evidence of possibility and conclusive proof of actuality: some coins were made by pouring molten metal onto a flat surface, and Sellwood may have shown that it might be possible to achieve the required accuracy, but this does not prove that all coins were made by pouring. Indeed, he demonstrates with the aid of British and Continental examples that it is not only certain that Iron Age coin was made using more than one method, but also that, in the case of potin issues, it is undeniable that some were cast in multiple moulds.

It should also be borne in mind that there is as yet no evidential basis for an assumption of validity for

However, exception might be taken to his statement that ‘Experimental work cannot be taken as proof’. Rather, it is poorly-devised experimental work that cannot be taken as proof: one must show, not just that something was possible, but that it could not have been otherwise, and this is achieved by examining the points at which it is clear that alternative hypotheses would produce noticeably different effects on the retrievable data.

10  The actual position is somewhat more complex, because - unless there had been a failure to maintain reducing conditions or the tray had been overheated to the point of failure – pellets could be tipped out without any need for the least percussion to help them, and therefore breakage was not a necessary part of the process of use. Indeed, Cottam (pers. comm, 2008) feels that many of the assemblages of coin mould fragments found so far are the result of accidental breakage, and that we may yet find deposits of unbroken trays stored for reuse. However, mould trays are so fragile that breakage following use would seem more likely than not, and there is good evidence that deliberate breakage was regarded as an important part of a tray’s journey from manufacture to deposition (see Chapter 12, §20 for a fuller exposition of this idea).

He deals with Sellwood’s contention that find-sites for pellet mould are vastly more numerous than the possible number of functioning mints by suggesting that pellet 12

Chapter 2: The Literature analogies drawn from Continental forms, methods of manufacture or processes of use of pellet mould.

Of particular interest is his account of the ‘traces of calcium carbonate, probably from powdered chalk’ found in the mould holes of some of the fragments from Verulamium. He interprets this as a ‘mould-release agent’. If this is true, then we must accept an extra, preparatory, stage between the manufacture and the employment of the mould. If such traces are found without evidence of extreme heat (such as slumping, vesiculation and vitrification), then we might assume that such mould had been readied for use, and that fragments which are in the same deposit, exhibit no signs of extreme heat, and have no deposits of calcium carbonate, could perhaps be considered as not ready for immediate use. One possible interpretation of this might be that it is evidence that pellet mould was stockpiled in quantities greater than would be required for a single episode of casting.

7 Haselgrove, C, 1987: ‘Iron Age Coinage in SouthEast England: the Archaeological Context, Part i’, BAR British Series 174(i), pp. 28-29. Haselgrove includes a succinct summary of the major theories about pellet mould, including the ‘intermediate process’ hypothesis, and the Sellwood/Casey interpretation, which he describes as ‘over-contrived’. He highlights the difficulties involved in linking the occurrence of mould with the presence of a mint, citing the spatial complexities of African metalworking set out by Rowlands (1971) as an alternative to a ‘single location, start to finish’ model for the making and finishing of coin, before concluding that all of these issues remained unresolved at the time of writing.

Van Arsdell warns of a tendency towards circularity in the argument associated with identifying mint sites from the presence of pellet mould, whereby ‘mint sites yield mould and sites that yield mould are mint sites’. He proposes that this problem can be resolved by noting that all mints sites attested by mint-names on finished coin have also yielded pellet mould.

8 van Arsdell, RD, 1989: ‘Celtic Coinage of Britain’, Spink, London, pp. 46 – 48. van Arsdell includes a brief treatment of the subject. He describes what he terms the ‘Flat Rock’ and ‘Muffin-Mould’ methods for producing flans, and says that ‘the arguments for choosing between them are irrelevant because both were probably used’. This is doubtful, for if one is attempting to characterize Iron Age coin, the question remains as to which coins were made by each method.

He is aware of only one tray-shape, the pentangular, 7 x 7 + 1 hole, ‘square-with-pediment’ type found at Verulamium. He claims that ‘two sizes of cavity are known, one about the size of the larger gold staters and the other about that of the smaller silver pieces.’ He does not give even approximate size ranges for either category, nor does he say from where he derives this idea.

In his account of the ‘Flat Rock Method’, he mentions advantages this process would have had over production by means of a mould, since the shape of droplet produced would have been suitable for striking without the need for an ‘intermediate process’. However, he also highlights the great skill that would have been required of the pourer to achieve the consistent accuracy of weight that he had observed in the finished coin. He quotes an error-margin approaching ± 0.05g – a figure 10 times smaller than that achieved by Sellwood in his pouring experiments.

There is, however, one glaring error in his account. He mentions ‘other moulds from Villeneuve–SaintGermain’ with ‘little holed covers made of clay’. The original reports of this material make no mention of such ‘little covers’ (‘petits chapeaux’): instead, the description is of potin mould, with the holes linked by little channels (‘petits canaux’).11 It also mentions ‘petits creusets munis d’un bec verseur’ – tiny crucibles furnished with a pouring spout – and then goes on to say that these were rarely found in conjunction with coin mould, concluding that therefore they were probably not part of the same process.

Describing the ‘Muffin-Mould Method’, Van Arsdell notes the cumbrousness of the ‘intermediate process’ required to turn pellet into flan, but points out the advantages of the increased accuracy obtained by weighing the metal in each cup. He mentions that there is disagreement over the manner in which metal was introduced into the cups, either as granules to be smelted, or poured in the molten state. He points out that the pouring method is subject to just the same problems of accuracy and consistency as the ‘Flat Rock Method’.

9 Elsdon, S (with Reynolds, J, and Robbins, K. and Bayley, J.) , 1997: ‘Old Sleaford Revealed: A Lincolnshire settlement in Iron Age, Roman, Saxon and Medieval times: excavations 1882 – 1995’, Oxbow Monograph 78, Nottingham Studies in Archaeology 2, pp. 51 – 67. This is the longest and most comprehensive description and examination hitherto of any assemblage of British

In his account of the ‘Muffin-Mould Method’, he distinguishes two processes by which metal placed in the mould holes might have be melted: by localized heating using a tuyère, and by heating the entire tray in a furnace. He concludes that, from the evidence of vitrification, both methods were used.

Debord, J, 1989: p12. This linguistic lapsus has occasioned considerable embarrassment both to the author of the current work and to one of his clients. Under these circumstances, little in the way of charity can be extended to its originator. 11 

13

Making a Mint coin mould. The data on which it is based were collected by the excavator, Margaret M. Jones, at the time of retrieval – a remarkable feat of recording carried out under field conditions.

have a vitrified base, but that vitrified fragments of all types form only 1.7% of the total number of fragments. She concludes that heat was supplied from below as well as from above.

9.1 Coin pellet moulds, Crucibles and Minting Procedures – Elsdon, S.

The next section deals with ‘Size of hollows, trays and estimates of mint capacity’. Only general parameters of tray thickness are given, note being taken of the great variability shown in the assemblage, and we are told that the majority of fragments are between 14 and 18mm Elsdon notes that the range at Old Sleaford is not the same as at Bagendon.

Elsdon opens with a brief description of the contexts within which the deposits of coin mould at Old Sleaford were found. She continues with the basic statistics of the assemblage: number of fragments, number of fragments with measurable hollows and thicknesses, number of corner pieces and scraps, and number of ‘pieces with a flat base’. She tells us that each fragment was X-rayed, and that - as well as suffering flood damage – only half now survive. She also states that, apart from a single box, all the fragments from the 1960-3 dig are now stored ‘safely at Lincoln’.12

Elsdon also notes that there is considerable variation in the diameter of mould holes. Following Tournaire, and because of a policy to retain in situ all mould hole fills, measurements for the top diameter only of holes are given. She says that the smallest holes are 4-5mm diameter, the vast majority between 7-9mm, and a small number between 12-14mm, adding that ‘Other sizes of cavity are proportionally represented’. She concludes that ‘these three sizes relate to the three denominations of Corieltauvian coins’, adding that the one complete pellet (which only a page before was ‘damaged’) corresponds to a full denomination coin, and comes from a 10mm top diameter hole (which does not seem to fall within any of the size-bands she gives).

There is a brief discussion of finds associated with the deposits of coin mould. She continues with a detailed discussion of possible tray forms and methods of manufacture. Her first conclusion is that the Verulamium 7 x 7 + 1 tray form was almost certainly not present at Old Sleaford. She points out the presence on a small number of edge fragments of a horizontal groove, which she surmises was intended to facilitate cleaving the trays to assist pellet removal. In support of this, she cites the presence in the assemblage of ‘many’ fragments which appear to be the lower parts of trays which have cleft along this axis, and states that pellet retrieval was very efficient, since only a single, partial, pellet was found trapped in situ in the mould.

She then demonstrates how wide are the limits for the estimation of the capacity of the mint, given the wide range of variation in possible estimates of tray capacity and the imprecision of estimates of the number of fragments remaining unexcavated. 9.2 The Statistical Analysis – Reynolds, J. Reynolds commences with a count of coin mould fragments and possible coin mould fragments. He then supplies a ‘grand total’ of complete and incomplete cavities, appending a table in which the numbers of complete and incomplete mould holes are broken down into size bands (and ‘unmeasurable’ holes). The size bands are 4-6mm, 7-9mm and 10-14mm, although we are given no evidence that these bands are anything other than arbitrary. He tells us that these groups of diameters represent, respectively, ‘minims’; ‘half denomination’; and ‘full denomination’.

She mentions circular trays from France (as described in Tournaire et al.), before suggesting that the almostcomplete 6 x 10 tray from Saintes is perhaps a closer analogy for the largest fragment from Old Sleaford, which exhibits a maximum of eight holes in row, and six in column. She mentions that the holes in the Saintes example were undeniably made using what she terms ‘a matrix’, a multi-pronged dibber. Elsdon then proceeds with a discussion of the evidence for the way in which the trays were used. She mentions that some trays exhibit vitrification on the upper surface, showing that they were heated from above, and points out that the melting-point of clay is slightly lower than the melting-point of silver. Minute beads of metal on the upper surface of fragments revealed by X-ray also support the idea of heat applied from above. She also notes that a substantial minority of vitrified fragments 12 

A detailed breakdown of the distribution of top diameters is then given, followed by two histograms, all of which show a standard bell-curve of notable symmetry, with a single peak at 9mm. There are no subsidiary peaks. Instead, there is a continuous spectrum of diameters from 4-14mm. He demonstrates that the majority of recovered fragments are relatively small, concluding from this that ‘extensive destruction was necessary to free the pellets’.

For the story of this single missing box, see Chapter 10, §1, below.

14

Chapter 2: The Literature In the next subsection, Reynolds attempts to demonstrate that the distribution of different hole diameters is not uniform across the two main areas of deposition. Unfortunately, he gives the proportions of each diameter at either site as a percentage of the total number of fragments of that size rather than as a percentage of the total number of fragments at each site. Since the disparity in the number of fragments between ‘Site A’ and ‘Site H’ is very large indeed – 90 at Site A, against 4507 at Site H – except where there is a total absence of a particular diameter at a particular site, what his figures demonstrate is that one assemblage is much smaller than the other, and do not actually address hole diameter distribution at each site.

to coalesce into a pellet’. All three tested positive for copper and zinc, and two also tested positive for silver. Analysis of one sample revealed two phases, a silverrich phase and a copper-rich phase.13 In the final section of their report, ‘Interpretation and Conclusions’, Robbins and Bayley state that there is no evidence that ‘molten metal was poured into the coin pellet moulds’. They also provide tables detailing the results of their programme of testing. 9.4 Comparisons and Conclusions – Elsdon, S. The Old Sleaford assemblage is first set in a wider context, as the largest deposit of coin moulds so far found in Europe. Elsdon demonstrates that it is one example of a widespread (although uncommon) type of find, with many deposits found both in Britain and in mainland Europe. She also shows that the finding of coin mould deposits has a long history, stretching back possibly into the C18th.. She lists many of the find sites by name.

By way of example, we will look at three hole diameters for which Site A yielded 4 fragments: 5mm, 10mm, 11mm. If one were examining the distribution of diameters at Site A in order to compare them with the distribution of diameters at Site H, these should all form the same percentage - 4.44% - of the total number of fragments at Site A (90 fragments in all), whereas Reynolds gives figures of 30.80% for 5mm, 1.10% for 10mm, 2.70% for 11mm.

In the following section she compares the ‘three sizes of cavity found at Old Sleaford’ with three sizes of contemporary coinage found in the vicinity. She then compares these ‘sizes’ with those found at other sites, stating that ‘present evidence’ suggests that the largest diameter holes were used for the production of either gold or bronze coinage, and the smaller holes for the production of three denominations of silver coinage.

In the following section, a number of useful statistics are given, for ‘fragments with metal in cavity’; fragments with metal... on surface’; ‘fragments with charcoal in hole’; ‘fragments with vitrification’; ‘fragments with corners’; ‘fragments with straight edge’; ‘fragments with base. The thickness of fragments is addressed in the next section.

In the next subsection, ‘Associated Coins’, Elsdon mentions that finished coin seems to have been a relatively uncommon find at Old Sleaford. Dating the minting activity there, therefore, must be on the basis of associated finds. She mentions that great capacity of the Old Sleaford mint, and dismisses the suggestion that its production was subsequently exported to the Iceni. She then attempts to break down this production in terms of denomination, using the proportions of posited hole diameter groups. She concludes this section with the observation that, in the absence of very large hoards of Corieltauvian coinage, the production of the Old Sleaford mint must largely have been smelted down for re-use.

In the section ‘Conclusions and Discussion’, Reynolds explicitly links hole diameter with pellet module, and on this basis attempts to characterise the output of the mint. He also notes, on a more secure evidential basis, that trays were heated from the base as well as from above. 9.3 Metallurgical analysis of coin pellet moulds and crucible fragments – Robbins, K and Bayley, J, Ancient Monuments Laboratory, English Heritage. Robbins and Bayley commence with an exposition of the debates surrounding the use of pellet mould, before concluding that the most likely explanation for their use is the production of coin pellets. They mention that ‘wetting’ of the mould surface by the molten metal would not have occurred in the casting of noble metals.

In the following subsection, Elsdon sides with Collis in dismissing the Sellwood/Casey hypothesis that pellet mould was not used in the production of Iron Age coinage. In her final subsection dealing with coin mould, Elsdon again postulates a possible size of output for the Old Sleaford mint. She also postulates itinerant ‘skilled moneyers’ making mould trays in ‘a wooden box of predetermined size with the sides standing proud of

They state that SEM analysis of an ‘unaltered zone’ in the single trapped pellet found in the Sleaford material showed that it was composed of 64.2% silver and 35.8% copper. They then describe the 22 fragments of coin mould sent for analysis. Three of the 22 fragments showed possible traces of metal adhering, small droplets which had ‘failed

13  This may indicate that silver-rich metal and copper-rich metal were added to the mould hole as separate ingredients, as hypothesized by Longden of the Ford Bridge, Braughing, metal prill (see below).

15

Making a Mint the top...in order to control the depth of the hollows’ – although her own illustrations14 provide strong evidence that hole depth was not controlled. She then states ‘These holes would be made with a portable matrix, a row of possibly seven borers set vertically in a horizontal bar’ – although Plate 13 and Fig. 36, 2 and 4 all show irregular hole spacings in both row and column.

in conjunction with an Energy Dispersive Detector, Longden examines selected fragments of coin mould for metal traces and subtle variations in heat-induced changes in the fabric of the mould, with the intention of discovering the type of coin pellet manufactured and the finer details of the process of pellet manufacture. The author deals first with the vitrification commonly exhibited by mould which has been used. She notes that the clay from which the trays are made vitrifies at a temperature close to the melting point of the metals used in coin production, and is able to demonstrate both that heat was applied to the top of the tray rather than at the base, and that the effects of heating are not as uniform in their distribution through the body of the fabric as with mould from Verulamium. This she attributes to the lower melting point of the base metal smelted in the Ford Bridge assemblage compared with the melting point of the silver pellets produced in the Verulamium mould. She suggests that it might be possible to infer the type of metal smelted in a given mould sample from the degree of vitrification exhibited, but notes that extensive additional SEM examination of mould from several different sites would be necessary in order to verify this hypothesis. She notes also that the variation in the degree and duration with which heat was applied to the trays may indicate variations in the detail of the smelting process, which might be indicative of lower levels of centralized control of the minting process.

10 Chadburn, A, 1999: ‘Tasking the Iron Age: the Iceni and Minting’ in ‘Land of the Iceni’, ed. Davies and Williamson, Centre of East Anglian Studies, Studies is East Anglia History 4. Chadburn provides a lucid and forward-looking account of the logistics and possible techniques of Iron Age minting. She is clear that much of the thinking around the subject remains hypothetical, but she is equally clear that the way in which a society organizes a complex, quasiceremonial process such as the making of coin will, in its turn, have something to say about the society that made it, the way in which it perceived both coin and minting, and the way in which that society was itself organized. She analyses the issues surrounding the assembly of raw materials, underlining that trade, transport, coin and minting are inextricably linked in the Iron Age, and she also highlights the need for specialist skills in both the production of dies and the striking of coin.

Next, Longden considers the ceramic of which the trays are made. She notes that they were fired in a reducing atmosphere, and that their low density confirms that they were subjected to furnace conditions. Although her report does not mention this, work carried out by Cottam and Gebhard et al. has shown that a reducing atmosphere is vital to the successful manufacture of a cupreous pellet in a clay tray. She notes the presence of voids in the sample examined which may be indicative of the use of an organic temper, and suggests that the vertical orientation of these voids is evidence that the trays were formed in a mould. She notes that the clay used is not refractory, citing the absence of high levels of aluminium in the material.

She notes that Icenian coin was made of a standardised alloy, and stands with van Arsdell on the remarkable accuracy with which the Iron Age coinsmiths were able to measure the metal they used. She sides with Collis against Sellwood and Casey, echoing also his note of caution (‘If we are correct about pellets and blanks’). With regard to the ‘intermediate process’ theory, she notes the relative awkwardness that this would introduce into the minting process, but cites both Jeffrey May (pers. com.) and finds of both pellets and blanks together as reason to allow the hypothesis. 11 Longden, H, 2008: ‘Coin Moulds from the Iron Age Oppidum of Braughing: An investigation of Celtic coinage production techniques.’ Unpubl. report, University of Liverpool.

The author then proceeds with the analysis, using SEMEDS, of a metal prill large enough to be located with the naked eye. She reports a generally homogenous composition throughout the prill, which is a 1.6% tin bronze with a high level of copper purity and very slight traces of iron, manganese and sulphur. Noting the preferential degradation of certain elements, she emphasises the difficulty of associating the Ford Bridge mould with particular issues of coinage, although she states that the levels of tin fall comfortably within the ranges exhibited by the coinage of both Tasciovanus and Cunobelin. She states that the homogeneity of tin concentrations through the sample is good evidence

This study was commissioned and funded by Stewart Bryant of the Hertfordshire County Council Historic Environment Unit on behalf of English Heritage, and carried out by Henrietta Longden, then of Liverpool University, as part of her MA dissertation. Using Scanning Electron Microscopy in Secondary Electron Imaging mode, Back-Scattered Electron mode, and Fig. 36, fragments 2, 4 and 8 all show holes that differ widely in depth.

14 

16

Chapter 2: The Literature the colour of the finished coin. In addition she notes the presence of iron, mentioning the work carried out by the University of Munich Archaeometry Group using iron phases in the fabric of coin mould to discover data about the temperature applied to the trays and the time during which the temperature was maintained. She concludes that similarities in the way silver was added to the metal mixture indicate a continuity in tradition between Continental Celtic minting practice and its counterpart in Britain.

that the bronze was pre-alloyed before being placed in the mould, citing parallels with evidence from coin mould found at Manching, and hypothesizes that this results either from the use of recycled metal, or tight (centralized) control of the composition of coinage. She argues that such a degree of control would presuppose an ability on the part of the smiths to judge the composition of an alloy with great accuracy without access to modern techniques of analysis. In addition, she makes the point that the use of ready-made alloys manufactured to a predetermined recipe would add yet another complex operation to what – to us – appears an already complex, even needlessly convoluted, process, highlighting the likelihood that Iron Age concepts of money and worth were very different from our own. She concludes that it is entirely possible that different stages in the production of coinage were carried out at different sites.

In her concluding section, Longden first emphasises that the moulds from Ford Bridge were undeniably used to make metal pellets, which were then almost certainly made into Late Iron Age coinage,16 pointing to the similarity in composition between metal traces in the moulds and coins known to have been manufactured in the Catuvellaunian territory. She confirms that the evidence from vitrification supports Tylecote’s hypothesis that heat was applied to the top of the trays, charcoal being heaped up over the holes to help ensure a reducing atmosphere, rather than (as suggested by Castelin) burning charcoal being placed within each hole.

In the fourth section of her report, Longden sets out the results of her analysis of what she terms a ‘white residue’ found in some mould holes. She states that this residue is almost certainly calcium carbonate, in the form of chalk, and that it was used as a mould-release agent to overcome the porosity of the clay (which is often exaggerated by the tendency of the clay to vesiculate under extreme heat), which might otherwise result in the loss of significant quantities of metal, and make more likely the fusion of the pellet with the fabric of the mould.15 She feels that the use of the chalk within the mould holes would explain why metal residues were found only on the top surfaces of mould fragments. The fact that this ‘residue’ has been noted on coin mould from Verulamium but not at other mint sites across the country she attributes either to local ‘tradition’, or as a practical response to the lack in this area of more refractory clays.

Her statement that production at Braughing and Verulamium was on a similar scale was correct when her report was written, but should perhaps be revised with the coming to light of the Puckeridge Assemblage. Citing coin loss patterns at both Braughing and Verulamium, she suggests that Braughing may have been a centre for the production of base metal coinage, while silver tended to be produced at Verulamium. From this she adduces the presence at Braughing of a ‘thriving trade economy’, while major state expenditure was focused on the tribal capital.

Next, Longden sets out the results of her analysis of metal residues on the material. She notes the presence of copper, and that it has been actively absorbed into the glassy phases of the clay, a phenomenon which could only have occurred at temperatures exceeding 1000o C. She tells us that tin was found similarly embedded in the form of cassiterite (tin oxide) crystals, and that these crystals formed as a result of leaching from a pre-mixed alloy. Silver was also present, as discreet globules in the voids of the vitrified clay, but in very small quantitites. She feels that this is not simply due to the higher specific gravity and melting point of silver compared with base metals, but could well result from the addition of the silver as a separate ingredient. She argues that the presence of silver in such small quantities does not indicate that a silver coinage was being produced, but that the silver might well have been included to affect

16  This idea requires a degree of modification in the light of recent discoveries and thinking. These will be discussed at length through the remainder of this work, but in essence Haselgrove,C, (2015, pers. comm) first raised the query about the security of the identification of an Iron Age context for British coin mould assemblages, and Ponting, M, (2015, pers. comm) who confirmed that most British copies of Roman coinage seem to have been made using coin mould

15  Although Trittschack, R, University of Friborg, (2013, pers. comm) does not accept this account, pointing out that calcium carbonate begins to degrade at temperatures above 700oC. However, since on the strength of this alone he concludes that these deposits are taphonomic, his own views are clearly open to question.

17

Chapter 3

Recording coin mould: aims and methodology After some trial and error, it was decided that the best approach was to collate the various claims made for pellet mould, its manufacture and function,2 and to attempt to resolve as many as possible of these theories into hypotheses that could be tested against morphological data. In addition to these, the current researcher had additional propositions to test, and more questions arose during the course of data collection.

A Aims To attempt to establish by means of supra-microscopic examination; classification using a standardized protocol; and comparison of coin mould: a.

The details of the processes in which coin mould was used b. The ways in which coin mould was made c. The possibility of local variations in the manufacture and use of coin mould, and the existence of local or regional traditions of minting d. The way in which minting at various sites may have been organized1 e. The social structures and value systems within which the manufacture of coin took place.

Early in the process it was realized that one of the theories central to the idea that pellet mould was used in the production of coin, the ‘intermediate process’ theory, could not be addressed by means of mensuration or observation. Rather, this is a question which can only be answered by the examination of the coins themselves, and then by a programme of experimental minting like that carried out by de Jersey and Burridge.3

B Methodology

The methodology devised is non-destructive, and requires no equipment more sophisticated than digital callipers, a strong desk lamp and a x8 handlens. The numerical data is retrieved according to the protocol set out below; additional features are listed of each fragment using a standardized terminology, as explained in the ‘Notes’ section of the protocol. Data from each sample of coin mould was listed on a pre-printed record card; specimens with more than five holes were drawn schematically on the record card, and the holes numbered. Fragments with no retrievable data beyond Burn Category and incomplete holes without measurable diameter or depth were bagged together according to the number of incomplete holes, and the number of fragments in each ‘bulk bag’ noted on a record card.

Unless it is possible to characterize a particular piece of mould in a standardized format, it will be impossible without the physical juxtaposition of the samples to compare one fragment with or differentiate it from any other fragment of mould. This lack of a standard procedure has been the single greatest obstacle to progress in the field of pellet mould studies during the last fifty years. The evolution of a useful recording protocol for pellet mould could be characterized by the two phrases ‘chicken and egg’ and ‘feedback loop’. Until a deal of mould has been examined, it is not possible to say which physical parameters should be measured – but without some decision as to which parameters to measure, it is difficult to make any meaningful examination of mould morphology.

C Resolving the theories into testable propositions 1 Tray forms

The basic criterion adopted was that as little as possible should be based upon assumption, and that primacy should be accorded to the data. Until its validity was demonstrated, reasoning by analogy would not be acceptable: uniformity of practice cannot be assumed, it must be demonstrated.

The deduction from a fragment of mould of the original form of the parent tray can be carried out by observing first, the angles of tray corners; second, the relationship between tray edges; third, the relationship between tray edges and the rows and columns of holes on the fragment; fourth, the number of holes in rows and columns,4 where this is possible.

1  Chadburn, A, 1999: ‘Tasking the Iron Age: the Iceni and Minting’; in ‘Land of the Iceni: the Iron Age in Northern East Anglia’, ed. Davies and Williamson; Studies in East Anglia History 4; Centre of East Anglian Studies; pp. 144 – 149. Her statement ‘...I want to briefly examine the range of tasks which might be required to make an Icenian coin, and in doing so try to illuminate the society which produced these objects’ encouraged the basic assumption behind the present study that the way in which a complex and socially-rooted task like minting is organized will reveal much about the structure and ideation of the society in which it takes place.

While this is a relatively simple process, it will only be possible to carry it out either if these features are to be See above, Chapter 2. de Jersey, P, 2007: pp. 257-269. 4  For the purposes of this study, ‘Row’ is defined as ‘on the same axis as Length 1’ and ‘Column’ is defined as ‘on the same axis as Length 2’. 2  3 

18

Chapter 3: Recording coin mould: aims and methodology

Figure 3.1: A completed record card, front and back.

found on the fragment, or if the features are in themselves sufficiently distinctive to permit diagnosis. A rightangled or rounded corner is a feature which could be possessed by several, entirely different, tray forms, and cannot therefore be considered diagnostic of a particular form, whereas the oblique corners of the pentangular Verulamium tray form can be distinguished with ease from both right-angled corners and from the wider oblique angles one would expect from a hexagonal tray.5 The single hole one finds at the apex of the Verulamium form is also distinctive; however, if this is missing from a fragment which has one edge and no corners, the angle between the edge and the hole-row will enable the researcher to decide with confidence that the parent tray was not rectangular, if the hole-row and the edge diverge significantly from the parallel.

The Verulamium form is the most widespread of all for which we have conclusive evidence in Britain. As well as at Verulamium, Turners Hall Farm and Braughing/ Puckeridge, examples have been found at Sheepen (Colchester), Merlin Works and Blackfriars (Leicester), and Scotch Corner (Yorkshire).

Figure 3.3 Verulamium form tray from Merlin Works, Leicester

The Puckeridge form, a rectangular tray with five rows of five holes, is more problematic. Both of the diagnostic examples of this type have holes with a diameter greater than 15mm, and no corner fragments from the Puckeridge Assemblage with holes of this diameter exhibited oblique corners or apex holes that one would associate with a Verulamium form tray. It is therefore suggested that the Puckeridge form was reserved for larger diameter holes, but this cannot be demonstrated with absolute certainty.6 Figure 3.2: ‘Verulamium’ form tray.

As hypothesized, in Clifford, E, 1960: p. 144, of examples from Colchester. However, no hexagonal form has been observed in the material in the course of the present study.

5 

6  For a full discussion of the Puckeridge form, see below, Chapter 6, §2.

19

Making a Mint While this occasional feature may not have been of great significance to the contemporary users of pellet mould, in cases where it does occur, bowing can serve as a good indicator of the location and orientation on the parent tray of fragments without edge profiles. 3 Methods of tray manufacture There are five obvious ways in which a tray might be made: f. Using a box-mould, as with tile-making; g. Using a bowl-mould, like a very shallow jellymould; h. By cutting the desired shape from a larger sheet of clay rolled to the appropriate thickness, like cutting pastry; i. Freehand, without any device; j. By beating to shape using a paddle. A sixth is attested at Bavay, France:12 balls of clay, one per hole, appear to have been forced into some sort of mould. However, the distinctive signature of this technique – cruciform creases in the base of the tray where the balls of clay meet – has not been observed to date on British material.

Figure 3.4: ‘Puckeridge’ form tray.

However, while it is not a simple matter to define the precise details of a tray form, it is a much easier task to prove that it is either impossible or at least very unlikely that a given fragment could have come from a known tray form. The very large fragment of pellet mould from Old Sleaford described by Elsdon7 falls into this category.8 The best hint we have for this third tray form comes from Scotch Corner, where a number of conjoining fragments derive from a tray with 10 x 10 holes, which seems to have been reserved for making smaller pellets with a diameter less than 9mm.9

Experiment has revealed13 that some of these methods will produce a distinctive signature on a finished tray. These are all related to edge characteristics, involving the edge profile and markings on the side-face. However, it should be pointed out that these experiments assumed that an edge profile was the product of a single process. In fact, it seems possible that some profile forms result from two intentional processes, moulding and handfinishing; others represent accidental modification of a profile during either the later stages of manufacture or during use.

2 Tray Profiles Tournaire10 claims that the bowing of a tray profile results from the displacement of clay by the insertion of the dibber. Elsdon11 also refers to the bowed profile of trays, although she advances no reasons for this characteristic. It has been noted that this is by no means universal, so it would seem unlikely that there is any processual imperative for bowing. Instead, it should perhaps be viewed as either an accidental effect of a particular method of tray manufacture, or the personal taste of the tray-maker.

Edge characteristics: a. ‘I-Section’ Profile

7  Elsdon, S, 1997: pp. 53-54 and p. 53, Plate 13. This corner fragment has 8 rows and 6 columns. 8  And a large middle fragment from the Blackfriars site in Leicester, which also bears 8 rows and 6 columns (Current Archaeology 292, July 2014, pp. 6-7; Daffern, N, 2014: pers. comm.). This is significant because it has been suggested that Leicester may have overtaken Old Sleaford as the major tribal centre of the Corieltauvi (Daffern, N, 2014: pers. comm.). 9  Landon, M, forthcoming: ‘The Scotch Corner Coin Mould Assemblage’. 10  Tournaire, J, in Tournaire et al, 1982: p. 429. 11  Elsdon, S, 1997: p. 54.

Figure 3.5: I-section profile The ‘I-Section’ profile suggests that a box-mould has been used to form a tray. Its distinguishing features are the ‘serifs’ on both top and bottom edges, and result from the use of a mould open at both top and bottom. That 12  13 

20

Geoffroy, J-F, 2000. See Appendix I, ‘Experiment Series 1’.

Chapter 3: Recording coin mould: aims and methodology both serifs are preserved undamaged further suggests that the slab may have been extracted horizontally from a mould with an open end, as illustrated below.

Figure 3.8: An experimental ‘bowl-mould’ with one open end.

Edge Characteristics: c. ‘Straight-Section’ Profile

Figure 3.9: ‘Straight section’ profile.

This edge profile is suggestive, but not diagnostic, of a cut edge. Edge Characteristics: d. ‘Angled Section’ Profile

Figure 3.6: An experimental ‘box-mould’ with one open end.

Edge characteristics: b. ‘Lazy S’ Profile

Figure 3.10: ‘Angled section’ profile

Again, this profile is suggestive, but not diagnostic, of a cut edge’ Edge Characteristics: e. ‘Rolled-Edge’ Profile

Figure 3.7: ‘Lazy S’ Profile

This profile is consistent with the use of a bowl-mould. The features which distinguish it from profiles produced by other methods of manufacture are the smoothly rounded upper edge and the ‘foot serif’ at the base only. This ‘foot serif’ is caused when the clay is smoothed flush with the top of the mould.

Figure 3.11: ‘Rolled edge’ profile

The tapering of the slab as it approaches this type of profile, and the distinctive ‘rolling’ of both upper and lower edges seems to indicate that the edge was not formed in a mould or by using a paddle. On fragments retaining more than one edge, this form of profile may appear in conjunction with any of Profiles 1 – 3. For Profiles 1 and 2, this possibly indicates the use of a mould with one open end. However, it should be noted that experiment has not been able to resolve this point beyond doubt. The form of rolled edge noted on some material from Old Sleaford seems by its regularity and consistency to have been produced by the intentional modification of a Type 2 profile while the clay was still wet. 21

Making a Mint to grease the mould, which worked very satisfactorily; however, the use of a mould-lining would also be a practical way of dealing with the problem. The ‘band and lines’ marking is very similar to the impression caused in wet clay by a length of Iris pseudocorus leaf, and the occasional occurrence of sections of ‘band and lines’ terminating in a clear diagonal cut would seem to support this interpretation.

Edge Characteristics: f. ‘Overhang’ Profile

5 Evidence of elaboration This refers to features such as the ‘cleavage grooves’ noted in the Sleaford material by Elsdon,14 and the ‘incised guidelines’ observed on a very small proportion of the fragments from Ford Bridge. They are noteworthy in that they reveal a degree of attention and care during manufacture beyond the norm, but closer study is required before it can be assumed that these features reveal any further information.

Figure 3.12: ‘Overhang’ profile.

Although many examples of this type of profile have been made using a mould (especially at Scotch Corner), other examples clearly result from extreme vesiculation of the top surface of the fragment during heating, and other again seem to have arisen through displacement of clay during the hole-making process.

With regard to the ‘cleavage lines’ found at Sleaford, Elsdon15 suggests that they represent improved functionality, by enabling the cleaving of trays along the rough line of the base of the mould holes, which she claims would have enhanced the retrieval rate of pellets. On the face of it, the benefits of this practise would seem to have been minimal: cleavage lines have not been noted at Verulamium, and have been found on none of the material from Braughing, yet only three in situ pellets have been found at Verulamium, as opposed to one at Sleaford and one in the Puckeridge assemblage16 – and none at all have been found at Ford Bridge. It is for this reason that the present survey classes these grooves as ‘elaboration’.

Edge Characteristics: g. ‘Cut and Tear’ Banding on SideFace

Figure 3.13: ‘Cut and tear’ banding.

Concerning ‘incised guidelines’, it should first be noted that these features have so far firmly been identified only on material from the Braughing/Puckeridge Complex, with two probable examples from Turners Hall Farm near Verulamium. These lines were made using a point, and occasionally some sort of straight edge, on the upper surface of the tray while the clay was still wet. To date, they have been found in three places on the upper surface of a tray: lateral, between the outer column of holes and the side edge of the tray, running more or less parallel to the edge; horizontal, below the apex hole, and close to, and roughly parallel with, the top hole row; and pedimental, so far noted only once, with a line following the sloping edges of a Verulamium form pediment.

Experiment has shown that this distinctive marking – smooth bands at the top and bottom of the face, with a band of rough, torn clay in between - on a side-face is produced by cutting the clay with some sort of blade. 4 Edge markings A number of markings of uncertain import have been noted on both the Ford Bridge and the Puckeridge material, the most striking of which has been termed ‘band and lines’. Since this marking is not prominently displayed, and since it is often very faint and fragmentary, it should perhaps be concluded that this is not a decorative motif, and ought rather to be considered as a minor variant in the tray manufacturing process. It was noted during the experimental manufacture of mould trays that when using a wooden bowl-mould without modification the clay tended to adhere to the mould, resulting in serious malformation of the tray during extraction from the mould. The solution adopted in the experiments was

Elsdon, S, 1997: p. 53. Elsdon, S, 1997: p. 53. 16  A substantial fragment of trapped pellet was noted by Metcalf in the Turners Hall Farm Assemblage (see below, Chapter 11, §13). 14  15 

22

Chapter 3: Recording coin mould: aims and methodology The precise purpose of these lines is not certain.17 They have been termed ‘guidelines’, yet were they essential to the placing of holes on a tray, then one might reasonably expect their presence on a much higher proportion of the recovered material. Moreover, experiment has shown that it is perfectly possible to position holes in rows and columns with sufficient accuracy to enable all 50 holes to fit onto a Verulamium from tray without the use of any guidelines at all, using the apex and the two oblique corners of the ‘pediment’ as reference points. For a fuller discussion of possible methods of controlling hole spacings and alignments to enable the inclusion of the desired number of holes on a tray, see below, ‘Methods of hole manufacture’.

To modern minds, conditioned by two hundred years of Industrial Revolution, effort expended on the means of production offers clear benefits in terms of speed, ease and standardization. However, it must be remembered that the savings offered by this approach to massproduction in time and effort are often only apparent if the production process in question is carried out regularly and on a large scale. If production is episodic or spasmodic, and relatively small-scale, then it is perfectly possible that such savings will be insufficient to justify any great outlay of labour on complex or ‘sophisticated’ manufacturing systems. It should also be remembered that ‘standardization’ can be applied to any or all of the different parameters of an object: one may have a ‘standard diameter’ without a ‘standard depth’. The notion that the mass-production of a type of object should result in examples which are physically identical is not the only approach which will yield acceptable results. An alternative concept would be ‘functional identicalness’ – the mass-production of objects which are capable of fulfilling the same function, while not being physically identical. An example of this would be the production of pottery in the Late Iron Age: this was produced in quantity, but pots of the same type are never exactly identical. The height of the shoulder will vary slightly from pot to pot, as will the thickness and profile of the rim and the thickness of the wall, but these variations did not prevent the pots from fulfilling the same function. So long as the pellets were identical in the parameters that affect their functionality, precision in other parameters would have been unnecessary.

Furthermore, for horizontal guidelines at least, the evidence is ambiguous as to which came first – the top hole-row or the guideline. Slighting has been observed, but it has not proved possible to discriminate priority. Taken together, these observations would seem to show that ‘incised guidelines’ were neither essential to the successful completion of a tray, nor to its function in the minting process, and that they should therefore be classed as elaboration. 6 Methods of hole manufacture A word first about terminology: Elsdon18 has used the word ‘matrix’ for the implement used to make the holes. Since ‘matrix’ means ‘womb’, it seems a wonderfully inappropriate term for a tool that is essentially a blunt prong, and which operates by piercing. The term ‘dibber’ is therefore to be preferred as much more apposite and accurate.

The determination of whether mould holes were made all at once, in multiples, or singly, can be made irrefutably of a given fragment under certain circumstances using morphological data.

It has been claimed (by Reynolds in Elsdon)19 on the basis of Continental examples20 that the mould holes in a tray were made in multiples, but this has never been tested against British material; nor was consideration given to two possible alternative methods of hole-making: that the holes were made all in one go, using a sort of pegged board; or that they were made one at a time using a single point. Indeed, until very late in the process of preparing this book for publication, a third method of hole-making had not occurred to anyone: that the holes on a tray might be made all at once in a bowl-mould with dibbers set in its base.21

If a peg-board has been used to make the holes in a fragment, then one would not expect to find instances of one hole slighted by another, nor evidence of ‘abortive’ holes; and all holes would have an identical angle of insertion. However, none of these conditions save the last can be considered irrefutable evidence that a pegboard has been used. The finding of several fragments with identical hole spacings in two axes would constitute proof, but presupposes a standard of preservation better than that observed in the Puckeridge assemblage. If a dibber with more than one prong has been used, then one might expect that this would result in repeated patterns of spacings between holes, in either rows or columns, depending on the orientation of the dibber.22 One would expect to find an identical angle of insertion

But see below, Chapter 12, §5, for a full discussion of their import. Elsdon, S, 1997: p. 54. 19  Reynolds, J. in Elsdon, S, 1997: p. 56. 20  Tournaire et al, 1982: plate 31b. The almost complete tray from Saintes has undeniably been made using a dibber with six prongs in a row: the repeated patterns of spacing in one axis are plainly visible, as are variations in the orientation of each row and irregularities in the alignment of columns. 21  See Appendix I, §5, ‘Interpretation’. 17  18 

See Appendix 1, §4, for the testing of this ‘common-sense’ assumption.

22 

23

Making a Mint and/or extraction in that axis, and any hole slighting would similarly take place only in a single axis.

The sole exception to this would be the ‘tassiform’ hole profile caused by the use of chalk wash, which has the effect of making more alike hole profiles that were originally quite dissimilar.26

Conversely, if a single-pointed dibber has been used, then one would expect to see instances of hole-slighting, abortive holes, and random spaces between holes. It is also likely that holes would exhibit different angles of insertion.

7 Number of pellets in a tray It must be accepted that, in many cases, it will not be possible to answer this question. If trays can be proved to be of the Verulamium form, then the matter is simple enough. It is very likely that all trays of this form contained fifty pellets in a 7 x 7 + 1 conformation. As well as the famous near-complete example from Verulamium, this is supported by a fine specimen from the Merlin Works excavation in Leicester,27 as well as a nearly-complete example formerly part of the Puckeridge Assemblage, now in the hands of a private collector.28

It has been suggested by David Parker,23 then of ULAS, who was working on the material from Merlin Works in Leicester, that it might be possible to track the path of the dibber across a fragment by taking a second top diameter measurement at right angles to the first. The orientation of the longer axis on each hole, relative to the orientation of the longer axis on the other holes on the fragment, would show the orientation of the dibber when each hole was made, and therefore might also show the order in which both holes and hole rows were made.

In other cases, involving other tray-shapes such as the Sleaford and Bagendon presumed rectangular forms, the matter cannot be settled in the absence of more complete specimens than have been found hitherto. As has been noted above for the Puckeridge5 x 5 tray form and the Scotch Corner 10 x 10, even when a rectangular form is fully known, it does not possess sufficiently distinctive features to enable attribution with any certainty if the fragments are too small to define row and column size.

It was felt that this idea was good, certainly good enough to warrant investigation, and so five of the larger fragments were selected on which to test the theory. However, two out of the five fragments generated results for dibber orientation that looked almost random. It was realized, after much thought, that while the research design had modelled dibber orientation during the process of holemaking with a single variable, angle of insertion, there were in fact three, independent, variables affecting top diameter: angle of insertion; angle of extraction; shape of dibber. It was decided that this rendered the technique too undependable to justify its employment.

8 Predictable relationship between base and top hole diameters Some writers29 have felt that it is sufficient to measure the diameter of the mouth of a mould-hole in order to ascertain the diameter of the pellet it would have produced. However, without concrete data to support it, this idea would seem to be unsound for a number of reasons.

These considerations also affect the validity of the ‘angle of insertion’ method of determining the way in which holes were made, and therefore this technique has not been pursued.

First, one should consider that the metal was melted – or poured – in the bottom of the mould-hole. This entails that, if any inference about pellet-size is to be drawn from measurement of the top of a hole alone, there must be a predictable relationship between the diameter at the top and the diameter at the bottom of a mould hole.

However, instances of hole-slighting and abortive holes have been noted, and the measuring of hole-spacings on fragments large enough to be able to provide reasonable evidence of repeated patterns of spacing has taken place.24 Slighting introduces another source of unintentional distortion, producing both D-shaped and ‘squarish’ (Clifford)25 outlines. D-shaped outlines are informative, in that they can sometimes be used to discern the order in which holes were made: the slighted hole will undeniably have been made before the ‘slighting’ hole.

Second, it should be remembered that Clifford30 noted the presence of ‘tapered’ holes among the Bagendon material. This means that a predictable relationship

See below, this chapter, §11. Kipling, R. and Parker, D, 2008: p. 3, plate 3. 28  Cottam, G, 2008: pers. comm. 29  Reynolds, J. in Elsdon, S, 1997: p. 59. Tournaire (in Tournaire et al,1982: p. 419) goes even further, claiming that the ‘coin module’ of a given tray is defined wholly by top diameter. However, not only does he offer no evidence in support of this extraordinary claim, it is further undermined by his own (incomplete and unsystematic) data set, which shows that variation between top and base hole diameter is actually greater than he claims in the text. 30  Clifford, E, 1960: p. 144. 26 

Observed variation in hole profile across single fragments is so great that it is clear that there is very little relationship between dibber profile and hole profile.

27 

Parker, D, 2009: pers. comm. Although see Appendix I for a discussion of the parameters of variation in the spacing of holes made using a multi-pointed dibber. 25  Clifford, E, 1960: p. 144. 23  24 

24

Chapter 3: Recording coin mould: aims and methodology hole and the denomination of the coin derived from it. The argument behind this assumption would seem to run thus:

between top and bottom diameters cannot be assumed, but must be demonstrated by measurement. Third, it should be pointed out that, in the light of basic mechanical principles, it is most likely that unpredictable and irregular variation in diameter will occur at the mouth of a mould-hole. Any obliquity in the angle of insertion or removal of the implement used to make the hole, whether as part of a single or as part of a multiple ‘dibber’, will cause the greatest variation at the hole mouth. Any ‘wobble’ during insertion will similarly be greatest at the hole mouth.

‘In a given assemblage, the smallest hole diameter is ymm and the largest hole diameter is (y + z) mm, therefore the hole diameter range within the assemblage is zmm. If there are x number of denominations of coin known in the vicinity of the assemblage, then the diameter range for each denomination will be (z ÷ x)mm.’ The most obvious problem with this approach is its circularity: there are three denominations of coin locally, therefore there are three groups of hole diameters, which means that three denominations of coin were manufactured at this site. However, we can never be certain that all, or some, or just one denomination of coin was being manufactured at a given site, unless this can be demonstrated on the material from the site itself. Until that demonstration has been made, any reference to actual coinage is misleading and irrelevant.

On balance, it would seem that it is more likely that the relationship between base and top hole diameters will not be predictable. However, since a uniformity of practice in pellet mould manufacture across the country cannot be assumed, both measurements should be taken (where possible)31 in order to put the matter beyond doubt. Several measures have been used to quantify the relationship between base and top hole diameters. The first compares the total range in each assemblage of both base and top diameters, to determine whether or not the prediction that top diameter would be more variable than base diameter is true.

How might one prove a predictable relationship between hole diameter and coin denomination? Few writers33 since Clifford34 have mentioned, let alone considered, the irregularity and variability of the material, and noone since Clifford has considered the effect that this variability will have on the relationship between hole diameter and coin denomination. Therefore, in assigning diameter-ranges to denominations there has been no appreciation of the possibility of overlapping diameterranges, or of the possibility of ‘general purpose’ mould, where any diameter of hole will do, so long as the depth is sufficient to accommodate the metal. Were either one of these hypotheses to be true, there would be, to echo Clifford’s words, no direct or simple relationship between hole and coin – and any attribution of denomination based on the contrary assumption would be of very little value indeed.

The second measure requires that the difference between base and top diameter for individual holes on a fragment be obtained, and then expressed as standard deviation for each fragment from which the necessary data can be obtained. The third measure expresses the variation in difference between base and top diameter for individual holes on a fragment as ‘Fragment Total Variation’, and is obtained by subtracting the smallest difference between top and base diameter on a fragment from the largest. Although it has not always been carried out in this work, by combining these last two measures, and expressing ‘Fragment Total Variation’ in terms of σ, it is possible to see whether Fragment Total Variation falls within or without the standard 2 σ, 95% confidence interval. This enables us to determine whether or not the range of variation in the difference between base and top diameter is statistically significant.

Assuming that the sample be large enough to give a valid reflection of the distribution of the original population, the easiest way to eliminate these possibilities would be to measure the diameter on two axes of each hole on every fragment, noting intra-fragment diameter variation (a difference in diameter between holes on the same fragment) and intra-hole variation (the difference in diameter on two axes across a single hole) on fragments

9 Predictable relationship between hole base diameter and coin denomination

and Tite in Partridge, C, 1982: p. 57. J. in Elsdon, S, 1997: p. 59. S.S.; 1983: p. 30. Tournaire, J, in Tournaire et al, 1982: p. 419. This list is not exhaustive. 33  Bayley, J, 1979: ‘Rochester: Belgic Coins and Associated Finds’; Ancient Monuments Laboratory Report N. 2811 Wilthew, P, 1985: ‘Examination and Analysis of Coin Pellet Moulds from Rochester, Kent’; Ancient Monuments Laboratory Report No. 4541. 34  Clifford, E, 1960: p. 144. Cowell

It has been assumed by many32 that there is a direct and predictable relationship between the diameter of a mould

Reynolds, Frere,

31  In all the assemblages examined the number of holes from which both base and top diameters can be garnered forms a relatively small proportion of the total number of holes. 32  van Arsdell, R.D, 1989: pp. 46 – 48.

25

Making a Mint with more than one measurable hole. If the hole-diameter series were not continuous, and the discontinuity were greater than the greatest intra-fragment and intra-hole variation found in the assemblage, then it would be less likely that either ‘overlapping groups’ or ‘general purpose’ mould were present in the assemblage. Instead, it would become more possible that a direct relationship between diameter and denomination could be inferred.

demonstrate that it offered some sort of benefit of ease or accuracy over weighing. The consistent composition of various types of coinage indicates that Iron Age smiths were capable of reproducing with great accuracy particular alloys, so the Sellwood/Casey hypothesis must also demonstrate that it is capable of that accuracy. If it is argued that the metal was weighed before being placed in the mould, then the production of pellets for alloying offers no advantage over simple weighing – rather it adds a wholly unnecessary stage between ‘pure’ metal and alloy. If, on the other hand, it is suggested that the metal was introduced by pouring, then it is a necessary corollary of this that the volume of the mould hole was controlled sufficiently to produce pellets ‘of known and consistent size and weight’,41 and this is also true of the ‘pouring’ hypothesis.

However, it should always be remembered that the behaviour of the molten metal within the hole (where it coalesces into a globule,35 rather than being cast as a sub-cylindrical pellet), and the need to minimize contact between metal and hole wall in order to avoid the fusion of metal with clay (as observed in the pellets retrieved at Verulamium and Old Sleaford, and the partially preserved trapped pellet found in the Turners Hall assemblage) mean that, while some holes would be too small for certain denominations, no hole can be considered too large for even the smallest size of pellet. In practice, this means that conclusions drawn from an assemblage exhibiting the full hole diameter range, or a diameter range restricted to the larger hole sizes, about the denominations being manufactured and their relative proportions within the assemblage will be much less secure than similar conclusions drawn about an assemblage with a diameter range restricted to the smaller hole sizes.

The best method for demonstrating that the volume of mould holes was controlled is to measure the depth of every hole on each fragment with more than one measurable hole depth, and note the intra-fragment depth variation. It is then possible to work out a rough volume for a given hole, using the formula: ‘Vol. = π r2 x h’, where r = the radius of that hole, and h = its depth. Since 1cm3 of silver weighs 10.49g, then 1mm3 will weigh approximately 0.01 g. The increase in weight per millimetre increase in depth will therefore equal (π r2 x 0.01)g.42

In order to determine the distribution of hole base diameters in a given assemblage, a fragment average base diameter was generated for each fragment from which data could be derived, and these average diameters were then listed in 1mm increments.

Three examples will suffice to show the relative magnitudes of variability that this would entail for pellets at the lowest, middle and upper points of the known hole diameter range.

10 Hole depth and the control of hole volume

For a 4mm diameter pellet, π r2 x 0.01 ≈ 0.16g per millimetre of depth, which is more than three times greater than the variability observed in the Icenian 1.25g silver coin.

The idea that mould-holes were in some way measuring devices lies behind both the Sellwood/36Casey37 hypothesis that pellets were used in alloying rather than coin manufacture, and the idea that metal was introduced into the holes in molten state by pouring, rather than in weighed amounts of solid metal which was then melted in situ.

For an 11mm diameter pellet, π r2 x 0.01 ≈ 0.66g per millimetre in depth – very nearly the half the weight of the same Icenian coin. For an 18mm pellet, π r2 x 0.01 ≈ 2.55g per millimetre of depth, which is more than twice the weight of the Icenian 1.25g silver unit.

Since we know from van Arsdell38 and Chadburn39 that the technology existed in the Late Iron Age for the very accurate determination of weight, to ±0.05g., better than or equal to 1 in 250 (Talbot is able broadly to confirm this figure, with important provisos, for certain Icenian silver unit issues),40 the Sellwood/Casey hypothesis must

distorting effect of wear and damage, he adopts a convention of excluding the lower 30% of the range of values, as well as excluding the upper 5% to remove anomalous excess coin weights. He provides data for all currently known examples of each of three series: ECEN silver units (‘all die linked and therefore likely to be from one continuous production sequence’); ANTED units (‘from a single die linked sequence’); ECE B units (‘from a number of die groups and therefore these may have been minted at separate locations’). 41  Collis, J, 1985: p. 237. 42  Since the figures in this calculation are based upon the weight of pure silver, it is clear that the results are illustrative rather than precise.

Tylecote, R, 1962: p. 104 and de Jersey, P, 2007: p. 262, Figure 5. Sellwood, D.G, 1980: pp. iii – vii. 37  Casey, J, 1983; pp. 358 – 60. 38  van Arsdell, R, 1989: p. 47. 39  Chadburn, A, 1999: p. 168. 40  Talbot, J, 2008: pers. comm. The provisos are: first, that the ±0.05 g figure is a statistically derived average, and should not be regarded as a constant standard; second, that in order to minimize the downward 35  36 

26

Chapter 3: Recording coin mould: aims and methodology 7

6.5

6

5.5

5 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

4.5

Series1

4

3.5

3

2.5

2

Figure 3.14: Results of an experiment to produce 26 holes with a controlled depth of 5mm.

In order to examine the range and distribution of interfragment variability within and between assemblages, a mean depth is derived for each fragment in an assemblage, and the results for each assemblage are then tabulated in 1 mm increments.

From these examples it is possible to see that, if a degree of accuracy in any way approaching that achieved by contemporary weighing technology were required, depth would have to be controlled to within fractions of a millimetre, and that this control of depth would have to become ever more stringent as the diameter of the pellet increased.

The measure adopted to quantify intra-fragment hole depth variability for each individual fragment was Fragment Total Variation,44 and from these individual intra-fragment variability measurements an assemblage intra-fragment variation is then derived so that variability in different assemblages can be compared. The value derived from the experimental trays was ±0.69mm. As with the relationship between base and top hole diameter, Fragment Total Variation can be expressed in terms of SD units, so that it can be determined whether or not Fragment Total Variation is statistically significant.

In fact, experiment43 has shown that it is extremely difficult to control the depth of a hole made in wet clay using a singlepronged dibber with any precision. Figure 3.14, above, shows that only 9 holes out of 26 fell within ± 0.5mm, or ± 10%, of a target depth of 5mm – which means that 17 out of 26 holes, or 65.4%, varied from the target depth by more than ± 10%. Three holes, 11.5% of the total, each deviated from the target depth by more than 30%, which equates to a total range of variation of close on 70%, a figure which is confirmed by the measurement of real coin mould. This degree of variability is clearly far beyond the narrow limits which, as has been shown above, would be required were the coin mould to be used as a measuring device. Tray No 6 7 8 9

Av depth 4.48 4.80 4.61 4.54

Fragment Total Variation in volume is expressed in terms of the percentage of the fragment average volume. 11 Calcium carbonate traces

St dev 1.35 1.18 1.29 1.49

Van Arsdell45 notes the presence in some of the mould from Verulamium of ‘particles of calcium carbonate, probably from powdered chalk’, which he interprets as a ‘mould release agent’, although Robbins and Bayley in Elsdon,46 state that ‘wetting’ would not have taken place were the moulds being used for the casting of

Figure 3.15 Comparing the average depth and standard deviation for four experimental trays attempting to achieve a hole depth of 5mm.

Defined as ‘Fragment maximum depth – fragment minimum depth’. van Arsdell, R, 1989: p. 48. Robbins, K. and Bayley, J. in Elsdon, S, 1997: pp. 60-61.

44  45  43 

See below, Appendix I, ‘Experiment Series 2’.

46 

27

Making a Mint noble metals. Tylecote47 points out that ‘wetting’ occurs when a layer of oxide is allowed to form on the surface of a pellet, which then fuses with the clay. He shows that Iron Age smiths were well aware of the need to exclude oxygen from the casting process by demonstrating that mould fragments from Old Sleaford were originally fired under reducing conditions, and surmises that this was achieved by adding charcoal to the clay.

This means that the presence of calcium carbonate traces in mould holes is capable of more than one interpretation. If the application of calcium carbonate was a routine procedure at those sites where its presence is noted, regardless of the metal intended to be smelted in it, it could be seen as a process intermediate between the manufacture of a tray and its use. As noted earlier, if unused fragments with and without calcium carbonate are found, one might reasonably speak of the stockpiling both of fired trays, and of trays ready for use. If the application of calcium carbonate to mould holes was related to the metal intended to be cast in them, we may see its presence or absence as an indication that particular fragments were intended for the casting either of base, or of noble, metals. This question is, in fact, one that can only be resolved by means of testing for metal residues.

Longden48 sees the chalk coating as providing a physical barrier between the molten metal and the vesiculating clay, but this is disputed by Roy Trittschack of the University of Fribourg.49 He points out that calcium carbonate (CaCO3) begins to break down into CaO + CO2 at a temperature below 800oC, while pure copper only begins to melt at 1084.5oC, and is emphatic that therefore CaCO3 could not have survived the heating process. He suggests that the CaCO3 may well be a phase formed by high carbonate activity in the soil following deposition, stipulating only that the soil pH be greater than 7. However, his argument would seem to lose force on two grounds. The first is that his conclusions derive from only the first two stages of the calcium carbonate cycle: when heated, calcium carbonate does indeed break down in to calcium oxide and carbon dioxide – but calcium oxide is not a stable compound, and readily combines with water to form calcium hydroxide, which recombines easily with ambient carbon dioxide to form calcium carbonate again. There is nothing about this that precludes the initial presence of calcium carbonate as a deliberate application of chalk. The absence of chalk wash from any of the broken surfaces of the fragments on which it appears would seem to suggest that this is not a taphonomic phenomenon, just as its overwhelmingly preferential location within mould holes would seem to point towards deliberate placement. The second point at which his theory does not appear to be supported by experimentally derived evidence is his assumption that the temperature at the base of the mould hole was raised to the melting-point of the metal being cast, whereas the work of Gebhardt et al.50 suggests that the temperature at the base of a mould hole rarely exceeded 700oC.51

The question of how the calcium carbonate was applied to holes is, however, capable of resolution by simple observation. It has been surmised (van Arsdell)52 that powdered chalk was pressed into the wet clay, but the feasibility of this has never been considered. How would the powdered chalk have been introduced into the hole? How would it have been pressed into the wall and base of the hole? Remembering that some mould holes are very tiny indeed, 5mm and less, it is clear that a human finger would not be able to do the job. The use of a stick might be posited, but experiment has shown that this would cause observable and distinctive irregular distortion of the hole, leading to a very large increase in intra-tray variation in hole diameter, depth, profile and plan. The inference to be drawn would seem to be that the calcium carbonate was applied as a liquid wash, using either a brush or the simple expedient of pouring a small quantity of the wash into each hole and then agitating the tray with a swirling motion to coat the walls of the hole. Brush marks have been observed, both inside holes and on the tray surface. In other cases, drips, splashes and dribbles have been observed on the tray surface, as well as swirl-marks on the chalk coat at the base of a hole. These differences in process are significant, in that they provide evidence of different ‘hands’ at work within a particular assemblage.

The fact that calcium carbonate, when heated, emits carbon dioxide tends to suggest that it was used by some Iron Age smiths to create and maintain reducing conditions within the mould hole, thus overcoming the potential oxidizing effect of the blast of air from a tuyère.

Chalk wash also has consequences for the measurement of other hole parameters. The coat is typically at least 0.5mm thick on the wall of a hole, and more than 1.0mm thick at the hole base. Not only are the diameter and depth significantly affected by this, but the profile of a hole can be changed from straight-sided, narrow taper or broad taper into tassiform.

Tylecote, R, 1962: pp. 103-104. Longden, H, 2008: p.26 49  Trittschack, R, 2014: pers. comm. 50  Gebhardt, R, et al, 1996: p. 1. 51  The picture is, alas, complicated by the work of Tite et al, 1983 and 1985, which adduces hole-base temperatures in excess of 1000oC (but no greater than 1050oC) in some coin mould from Braughing. However, the fact that they also claim surface temperatures of between 1100oC and 1150oC, with continuous vitrification and medium to coarse bloating pores suggest that the fabric was approaching what they term ‘terminal distortion’, a condition associated in this work with a second episode of heating following use. 47  48 

52 

28

van Arsdell, R, 1989: p. 48.

Chapter 3: Recording coin mould: aims and methodology At Bagendon and at Old Sleaford, by far the greater proportion of the mould retrieved apparently shows no signs of use at all. This may also be seen as further evidence of the stockpiling of mould, which in turn may be taken to imply that pellet manufacture was carried out at a given site on a sporadic basis and in unpredictable quantities.

When one considers the extreme fragility of the chalk coating – since it does not always bond well to the clay, it can peel away in sheets, leaving no trace behind it – this means that, in the words of the well-worn apothegm, the absence of evidence is not evidence of absence. The fact that chalk wash is not observable on the mould is no indicator that chalk wash was never present. It may reflect only that conditions have not favoured its preservation. As well as destruction during use, or by weathering and abrasion following deposition, acid soil conditions could also result in the total disappearance of calcium carbonate from an entire assemblage.

This might suggest that the occasions for making pellet were governed by specific need, rather than as part of the routine maintenance of a monetized economy: we might posit the linking of pellet manufacture to the payment of taxation or tribute, to a need to pay manpower hired from outside the territory of the issuing authority, or to the advent of seasonal trading opportunities.

This adds an entirely new dimension of uncertainty to the problem of whether it is possible to deduce from the dimensions of a mould hole the size of pellet produced in a given specimen of mould. If it cannot be said with certainty of the large proportion of mould so far examined that does not show traces of chalk wash that it never contained chalk wash, then – bearing in mind the variable thickness of chalk wash – the hole dimensions obtained from such mould cannot be related to the hole dimensions of the mould in use, except in very broad terms.

For these reasons, it would seem important to record of each fragment whether or not it has been used. However, this may not be as straightforward a matter as might be supposed. While some fragments show undeniable traces of use, such as vitrification, vesiculation and slumping, and it has been presumed by many authorities that these are the diagnostic signs of use, by no means all fragments which have yielded positive results under metal trace analysis exhibit any of these characteristics.

12 Introduction of metal into holes The debate about the introduction of metal into the mould hole, whether this was achieved by melting solid metal in situ in the hole, or whether it was achieved by pouring in metal melted elsewhere, has been touched on in Section viii. above, where it was treated in terms of the control of the volume of a mould hole.

Writing of the Henderson Collection material, Craddock and Tite55 state that none of the six samples examined showed signs of vitrification, yet all tested positive for metal residues, mainly silver. Instead, they note that the fragments tested all show ‘signs of strong heating, being red on the base from oxidization, but quite black on the top around the depressions actually containing the metal, showing that the metal had been covered in charcoal to prevent its oxidization whilst molten’.

However, there is another observable trace that might be expected to occur were metal to have been poured into the holes in a tray. It is inconceivable that this operation could have been performed many times without showing some evidence of splashing or spillage. These would involve prills on the surface of the mould surrounding the mouth of a hole, and would be quite distinct from the metalliferous staining noted on many fragments of used mould, which would seem to arise from the condensation of metal-rich vapour53 spread by the blast from a tuyère, and from the occasional droplets of metal found adhering to the lip of a hole.

They surmise that this could be because the vitrification point of clay is higher than the melting-point of silver (960oC). Yet Elsdon states that the vitrification of clay takes place at around 950oC, a statement largely supported by Gebhard et al.56 Although Gebhard and colleagues are careful to say that vitrification is the ‘usual’ effect of use, it is hard to reconcile this acknowledgement of occasional exception with the complete absence of the phenomenon on the material tested by Craddock and Tite.

13 Proportions of used and unused pellet mould Collis54 states firmly that ‘while some have proved negative, the majority (of analyses of trace elements in the moulds) have produced some traces of metal’. This might be taken to imply that the majority of pellet mould found has been used, but this is not necessarily the case.

It is beyond credence that the basic physics of the process could be at fault: the melting point of silver is invariant under normal conditions, and so is the vitrification point of ceramic, except in the case of certain types of clay with an unusually high refractive index, such as the bentonite used for gas mantles.57 There can be no doubt

Opinion is divided as to whether the purple staining noted on many fragments of mould is a compound of the copper cast in it, or of the manganese occurring naturally in the clay of which the tray is made. 54  Collis, J, 1985: p. 237. 53 

In Partridge, C, 1981: p. 326. Gebhardt, R, et al, 1996: pp. 1 - 5. 57  Or the kaolin-rich clay used to make the Levroux moulds (Odiot, T, 55  56 

29

Making a Mint that the clay used to make the Henderson mould is utterly unexceptional, and the presence in the Collection of two fragments exhibiting vitrification would seem to provide irrefutable confirmation of this.

alteration, so the loss of this layer would result in a fragment without any obvious signs of use. Unfortunately, this merely serves to complicate the attribution of use on the basis of supra-microscopic evidence alone, as exemplified in the recording protocol by the ‘Burn Category’ classification. It has been noted on material from Ford Bridge that blackening can occur simply because the mould has been deposited in close contact with charcoal, and reddening can be caused during the firing of ceramic by a failure to exclude oxygen from the kiln, and requires no exceptional heat. Add to this the fact of differing degrees of reddening on various examples, ranging from complete reddening, top to bottom, through reddening of one or other surface, to very slight and localized reddening, and it becomes clear that these are at best equivocal signs of use.

This apparent conundrum may be resolved by close examination of the variables in the process. First, there is the question of how heat was applied to the trays. Many examples show reddening (and even vitrification) of the base, and this has always been assumed to be evidence that heat has been applied by placing trays in a furnace preparatory to the actual smelting process. We know from Gebhard that temperatures at the base of mould holes in the Manching material rarely exceeded 700oC, well below the temperature required to vitrify clay, so there is no need to assume that heat applied to the whole tray would have exceeded this. Gebhard has also demonstrated that the fusion of metal granules occurs relatively quickly, requiring the maintenance of a temperature high enough to melt the metal for between three and five minutes. However, the results given in the paper for Mössbauer spectroscopy and the alteration in iron-bearing species during heating show that the experimental samples were maintained at temperature for between 3 and 48 hours.

As a result, it was felt necessary to supplement the ‘Burn Category’ classification with a standardized system of verbal description of heat effects on coin mould, and in Chapter 12 this system of description is rationalised into four new categories of macroscopic heating effects. However, it must be emphasized that attributions of use can only be made without the employment of SEM-BSE and SEM-EDS61 if prills of metal exist on the sample large enough to be detected with the use of a handlens.

It seems possible that this lack of vitrification in the Henderson assemblage could simply be an indication of how efficient the pellet makers had become, in that they had learned not to prolong intense temperatures beyond the bare minimum required to smelt metal granules or powder, and that this time was not long enough to initiate vitrification in the solid mass of the mould.

D Coin Mould Recording Protocol Site Code: Enter the site code, followed by Context: The context number from which the find came, followed by ID Number: If the piece has an individual find number, otherwise enter ‘*’ Number of pieces: How many bits are in the bag? Weight: Weight of bag contents in grams. Burn category:62 0 Unquantifiable 1 No trace of burning 2 Yellowing 3 Partly reddened 4 Fully reddened 5 Vitrified Thickness 1: Taken at one end of Length 1 (in millimetres). If the fragment lacks one or both surfaces, then enter ‘*’.

Tournaire and Henderson58 state firmly that vitrification will occur only in the presence of an alkali, either from wood-ash or from alkaline earth metal compounds present in the clay of the mould. It should be noted that calcium is an alkaline earth metal. However, Tylcote59 notes only that a ‘well-fired layer’ would have been produced by the ‘fluxing action’ of wood-ash; while Tite, Freestone, Meeks and Craddock60 make no mention of alkali or wood-ash, citing only temperature as the cause of vitrification. A further factor to be considered is the fragility of the vitrified layer on used mould. This layer is often very thin and, given the often friable nature of the mould fabric, is extremely susceptible to removal by both abrasion and weathering processes such as frost. Vitrification has been observed on fragments with little or no signs of vesiculation or other heat-induced surface in Tournaire et al, 1982: p. 421). 58  Tournaire et al, 1982: p. 432. 59  Tylecote, R, 1962: p. 106. 60  Tite, M.S, Freestone, I.C, Meeks, N.D.; Craddock, P.T, 1985: pp. 50 – 55.

Longden, H, 2008: pp. 1 – 5 for a detailed exposition of the methodology. 62  This is the original ‘Burn Category’ system, and as it proved unequal to the task of categorising the effects of heat on mould fragments, has been replaced by the new four-category system set out in Chapter 12. 61 

30

Chapter 3: Recording coin mould: aims and methodology If it is not possible to obtain a particular measurement for a given hole, this should be represented in the table by ‘*’. Edge profiles: If the fragment has a Position Type code of 00 or 01, enter Edge Profile code 00. If the fragment has a Position Type code of 02 or greater, please enter the appropriate Edge Profile code: 00 No edge profile 01 I-section 02 Lazy S 03 Straight section 04 Angled section 05 Rolled edge 06 Overhang 07 Cut and tear 08 Other (supply profile diagram in notes) 09 Uncertain If two Edge Profile characteristics are exhibited by a single edge, for example Angled Section and Cut and Tear, then both codes should be entered, lower code first, thus: ‘04+07.’ If a single fragment has more than one edge, then Edge Profile codes for each edge should be included, the code for Length 1 first, thus: ‘04+07; 05’. Hole profile: 00 None 01 Straight 02 Narrow flare 03 Broad flare 04 Other/Indeterminate (Include diagram in Notes) 05 Tassiform 06 Circle and swirl Notes: Under this heading are included all the observable features of coin mould not covered in the categories above.

Thickness 2: Taken at other end of Length 1 (in millimetres). If the fragment lacks one or both surfaces, or if Length 1 is too short for variation in thickness to occur, then enter ‘*’. Thickness 3: Taken on the longest axis at 90o to Length 1, (in millimetres). If the fragment lacks one or both surfaces, or if it is too short along this axis for variation in thickness to occur, then enter ‘*’. Position type: 00 Unquantifiable 01 Middle of slab 02 Straight edge 03 Curved edge 04 90o Corner 05 Oblique corner 06 Curved corner 07 Corner Length 1: If the fragment is an edge, then this measurement is taken along the edge. If the fragment is a corner, then this measurement is taken in millimetres on the longer side. If it is a middle fragment, then measure the longest axis in millimetres. Length 2: If the fragment is an edge or middle, then measure the longest axis at right angles to Length 1. If the fragment is a corner, then enter the shorter side measurement in millimetres. Lengths greater than 2: In the rare event that a fragment has more than two formal edges, these should be labelled upon the diagram ‘Length 3’; ‘Length 4’, and so on. They should be measured in the same way as Length 1 and Length 2, and the measurements included in the Notes. Incomplete holes: Enter the number of incomplete holes on the fragment. If there are no incomplete holes, then enter ‘00’. Complete holes: Enter the number of complete holes on the fragment. If there are no complete holes, then enter ‘00’. Hole measurement: A diagram of the fragment should be made, indicating both complete and incomplete holes. These should be numbered for ease of reference. Four measurements of each hole should then be taken, and listed in tabular form using the individual hole reference numbers: i. Horizontal diameter (taken at the base of the hole on the axis of the hole row) ii. Vertical diameter (taken at the base of the hole on the axis of the hole column) iii. Depth iv. Top diameter.

Set out below is a comprehensive list of all the features observed on coin mould so far, together with the abbreviations used in the databases compiled for each of the assemblages examined in this work. The list is not entirely systematic, as the terminology it expresses was evolved during the course of research, much of which is unrepeatable, neither is it exhaustive, since it is wholly possible that other assemblages as yet unexamined will yield up a tithe of new features worthy of note.

31

Making a Mint E Database Key version 2.6 Note in Longhand Abraded Abraded base Abraded edge Abraded top Angle of insertion, skewed Annulus and pit marking on hole base Apex Apex edge (LH) Apex edge (LH), possible Apex edge (RH) Apex hole, entire Apex hole, part Apex hole, possible Base mostly gone Base all gone Base only Base partly gone Base possibly gone Bast marks, presumed. face Length 1 Bast marks, presumed. face Length 2 Black blotches/black spotting Blackened base Blackened core Blackened hole base Blackened top Blob, indeterminate, adhering Length 1 Blowhole on top Boustrophedon dibbing pattern Break, ancient Beak, modern Breaks sealed by melting, some Brick/tile in matrix Brown crust on surfaces is taphonomic Brown layer, base Brown staining on top Brush mark in chalk wash in hole Bulk bag, should be in Burring of Length 1 as it meets Length 2: possible sign of mould lining Burring of Length 2 as it meets Length 1: possible sign of mould lining Burring, bottom edge, Length 1 Burring, bottom edge, Length 2 Burring, top edge Length 1 Burring, top edge Length 2 Cap, entire Cap, most Cap, partial Cap, possible, trace CBM inclusions at base Chaff cast on top Chaff marks on base Chalk wash on top Chalk wash on top, green stained Chalk wash splashes on top Chalk wash on all side faces Chalk wash on face, Length 1 Chalk wash on face, Length 2 Chalk wash on face, Length 3 Chalk wash on base Chalk wash on base, possible Chalk wash on broken edges Chalk wash in holes Chalk wash in holes, all Chalk wash in holes, most Chalk wash in holes, some Chalk wash, possible Chalk wash, right-angled line on top. Channels joining holes Charcoal casts

Charcoal cast on base Charcoal cast on top Clay blob in hole Clay, blob, in hole: possible cap Clay blob, face Length 1 Clay blob on top Clay blob on top, vitrified Conjoining fragments; ancient break Conjoining fragments; ancient break; conjoins with (Code) Conjoining fragments: found on return by Henrietta Longden Conjoining fragments, modern break Coppery blob Cracked and very fragile Crazing on base Crazing, face Length 1 Crazing on top Crust adhering to hole wall Crust on base Dark brown base Deformation all surfaces Deformation of base, extreme Deformation of hole Deformation of Length 1 Deformation of Length 2 Deformation, pre-firing, Length 1 Deformation of top Deformation, unspecified Dimple on base coincides with hole above Edge bevelled Edge markings: band Edge markings: band, wide Edge markings: band and lines Edge markings: band and lines, Length 1 Edge markings: band and lines, Length 2 Edge markings: band over parallel lines over band Edge markings: band over unclear Edge markings: diagonal striations Edge markings: diagonal striations above horizontal striations Edge markings: diagonal striations above horizontal groove Edge markings: grass mould lining, possible Edge markings: groove above foot serif Edge markings: groove midway on face Edge markings: 2 parallel bands Edge markings: 2 parallel grooves Edge markings: groove parallel top over diagonal striations Edge markings: horizontal striations Edge markings: line, single Edge markings: parallel lines, Length 1 Edge markings: near-vertical striations, Length 1 Edge markings: wood grain cast, possible Fabric, very hard – conchoidal fracture Fine finish Fingertip impressions, face Length 1 Fingertip impression on base, possible Fingertip impression on top, possible Fired deposits in holes Foot serif, burred Fragment missing Fragments of superimposed tray adhering to top Fragment exceptionally thin Freehand manufacture, possible Fused fragments of mould, two Grass stalk cast on base Grass stalk cast on base. possible Grass stalk cast on top, possible Grass marks on base Grass marks on base, possible Grass marks, face Length 1 Grass marks in hole, possible

Abbr. AB ABB AE AT ASK APM A AL ALP AR AH AHP APOS BMG BAG BO BAP BPG BML1 BML2 BB NB BC BH BT BIL1 BOT BD BA BM BSS BTM BCT BLB BST BMH BUG BL1L2 BL2L1 BBEL1 BBEL2 BTEL1 BTEL2 CE CM CP CPT CBMB CHT CMB CWT CWTG CWST CWS CWL1 CWL2 CWL3 CWB CWBP CWF CWH CWHA CWHM CWHS CWP CWRL CJH CC

32

CCB CCT CH CHPC CL1 CBT CBTV CFAB CF/Code CF CFMB CB CVF CRB CRL1 CRT CAW COB DBB DA DB DH DL1 DL2 DPL1 DT DU DCH EB EMB EMBW EMBL EMBLL1 EMBLL2 EMBPB EMBU EMDS EMDHS EMDHG EMGL EMGF EMG EMPB EMG2 EMGDS EMHS EMLS EMPLL1 EMVSL1 EMWG FHCF FF FIL1 FIB FIT FDH FSB FM FAT FET FMP FFM GB GSB GST GMB GMBP GML1 GMH

Chapter 3: Recording coin mould: aims and methodology Grass marks on top Grain cast on base Grain cast on base, possible Grain cast in body of fabric, possible Grain cast in hole Grain cast, face Length 1, possible Grain cast on top, possible Grey base Grey core Grit on base Groove along top of Length 1, possible guideline Grooves in top Heat-affected surface flaking off Heated beyond use Hole, abortive Holes arranged in a very irregular chequerboard pattern Hole base only Hole base, odd fracture of Holes larger than 15mm Holes small Holes occluded Hole occluded by vitrified residue Hole mouths standing proud of top surface, possible Holes oval, possible Hole has pierced fragment base (possibly) Hole plugged with taphonomy Holes purposively oval Hole slighting Hole slighting in two axes Hole slighting in opposite senses in adjacent hole pairs Hole slighted by mitred corner Holes very deep (>14mm) Holes very shallow Hole has wide flare at mouth Holes widely spaced Impression on top, possibly another tray Incised guidelines, double, Length 1 Incised guideline, orientation uncertain Incised guideline parallel apex edge Incised guideline parallel Length 1 Incised guideline parallel Length 1, double Incised guideline parallel Length 1, possible Incised guideline parallel Length 2 Incised guideline parallel Length 2, possible Incised guideline parallel Length 3 Incised guideline parallel Row 1 Incised guideline parallel Row 1, possible Incised guidelines parallel Row 1, double. Incised guideline at 45o to Row 1, possible Incised guideline, right angled Incised line (more cut than guideline) Incised line on base Incised lines, double, on top Inclusion cast Inclusions, chalk Inclusions, chalk, large Inclusion, dark grey Inclusions, flint Inclusion, grog Inclusion, large, flint Inclusion, large, organic, burned out Inclusion, large, pebble Inclusions, massive Inclusion, massive, shell Inclusion, quartzite Inclusion, soft, dark red Irregular holes and spacings Irregular rows Irregular rows and columns Luting, possible, in hole Marks on base, matting or cloth Marks on base, unspecified May not be mould

GMT GC GCP GCF GCH GCPL1 GCTP GRB GRC GTB GTL1 GIP HFO HBU HA HIC HBO HOF HL HK HO HOV HPT HOP HPBP HPL HPO HS HS2 HSOS HSM HVD HVS HWF HWS IT IGDL1 IG IGA IGL1 IGL12 IGL1P IGL2 IGL2P IGL3 IGR IGRP IGR2 IG45R IGRA IL ILB ILD ICAS IC ICL IDG IF IGG ILF IO ILP IM IMS IQ IR IHS IRO IRC LP MBM MBU MNM

Mitred corner Modern abrasion Modern break Modern repair Modern repair, possible Moulded line parallel Length 1 Moulded platform for holes Moulded ridge on top parallel Length 1 Mould lining traces, Length 1 Mould lining, possible, Length 1 Mould lining traces, Length 2 Mould lining, possible, Length 2 Not coin mould Not Ver. form: angle of row to Length 1 wrong Old card only: fragment not returned by Henrietta Longden Orange glass in hole Overcleaned (damage resulting) Parallel striations on base: matting marks? Parallel striations on top Pellet detachment, possible scars of in hole Pellet, trapped in hole, possible Poorly made Pot fragment included in bag Puckeridge tray form: 5 holes in a row. Purple staining Reddening on base Reddening in core Reddening on corner Red cortex Reddening in holes Reddening on Length 1 Red staining on base Red staining on top Reddened surfaces Reddening, unspecified, slight Red deposit on base Red deposit in hole Red deposit on top Red top Residue on top Residue on Length 1 Ridging on base Rumex cast on base Sagging on base Sagging on top Sagging, Length 1 Sample taken from hole base, affecting depth Sectioned by Henrietta Longden. See old card for original dimensions. Serif, foot, moulded Shallow-peaked Verulamium form tray Shell, crushed, on base Shell temper Shell temper, sparse Slag, blob, on base Slag blob in hole Slag, blob, on top Some slumping in holes ‘Splatch’ marks on hole base ‘Stepping’ in holes Striations on hole base Silvery globules ‘Squidge’ mark in hole Straw cast, possible, Length 1 Surfaces mostly gone Taphonomy in all holes Taphonomy in one hole Temper, crushed chalk Temper, crushed flint Temper, grit Temper, grog Temper, shell

33

MC MA MB MR MRP MLL1 MP MRT MLTL1 MLTL1P MLTL2 MLTL2P NCM NVF OC OGH OD PSB PST PDP PTP PM POB PF PS RB RC RCR RCX RH RL1 RSB RST RS RUS RDB RDH RDT RT REST RESL1 ROB RCB SGB SGT SGL1 SAD SHL SFM SPV SCB ST STS SBB SBH SBT SH SMH STH SHB SG SM SCPL1 SMG TAH TOH TCC TCF TAG TGR TS

Making a Mint Temper, waterworn grit, coarse Tested by Henrietta Longden Too cracked and fragile to measure in any aspect Top gone Top mostly gone Torsion marks in holes, possible Tray form, new, probable Tray profile, concave, possible Twig cast on base Undistorted cortex Unvitrified cortex Vesiculation, all surfaces Vesiculation on base Vesiculation in core Vesiculation in holes Vesiculation in hole, possible Vesiculation, Length 1 Vesiculation, Length 2 Vesiculation, Length 3 Vesiculation, slight Vesiculation on top Vesiculation on top, possible Vesiculation, unspecified Vitrification, all surfaces Vitrification on base Vitrification in holes Vitrification, face Length 1 Vitrification, face Length 2 Vitrification, internal Vitrification on top Vitrification on top, possible Vitrification, minute traces on top Vitrification, unspecified Void, large Wash, possible, in hole Whitened base Whitened face, Length 1 Whitened top Whitish cortex Wipe marks on base Wipe marks, face Length 1 Wipe marks on top Yellowing, all surfaces Yellowed base Yellowing in holes Yellowed side Yellowed top

TWC THL TFM TG TMG TMP TFNP CTPP TCB UDC UVC VESA VESB VESC VESH VESHP VESL1 VESL2 VESL3 VESS VEST VESTP VES VS VB VH VL1 VL2 VI VT VTP VMT VU VOLE WPH WB WL1 WT WC WMB WML1 WMT YA YB YH YS YT

34

Chapter 4

The Henderson Collection (Braughing) coin mould assemblage contains 13 fragments (seven of which may not be coin mould) too small for individual measurement,5 or 19% of the total number of fragments. While this figure is significantly lower than has been noted of either the Puckeridge or the Ford Bridge assemblages, it is still evidence that the Henderson coin mould has not been subjected to a comprehensive selection process either during or after retrieval.

1 General observations The Henderson Collection coin mould assemblage comprises 68 fragments, 7 of which may not be coin mould. There appear to be no conjoining fragments in the assemblage as it survives. Although a location for its discovery is shown in the Hertfordshire H.E.R., this is largely conjectural, as the excavator left no record of either the date or place of its finding. Since there is no guarantee that this coin mould was found in association with any of the other items in the Collection, this assemblage must be regarded as being almost entirely without context, whether spatial or temporal – except for what can be gleaned from the mould itself.

The average size of fragment in the study sample is 24.03mm (Length 1) x 21.19mm (Length 2). This is significantly smaller than the figure for Puckeridge (33.60mm x 31.35mm), but is reasonably close to the Ford Bridge average (27.19mm x 25.15mm). This means that we can be reasonably certain that at no point between deposition and the present has anyone pursued a policy of ‘only keeping the big bits’. Since the standards of excavation and the retrieval policy for the Ford Bridge Assemblage are well documented,6 it is justifiable to regard the statistics obtained from this sample as a sort of ‘gold standard’ against which we can measure other groups of coin mould.

A report on this coin mould, written by Craddock and Tite, is included in ‘Skeleton Green’ by Clive Partridge.1 While it is clear from this report that some fragments were tested for metal residues, it is not stated in the report how many fragments were tested, neither is it made clear whether this testing involved the destruction of any of the samples. In fact there is no quantification of the assemblage beyond the vague statement that there were ‘very many’ fragments of coin mould. However, we are told that traces of silver were found, while no traces of vitrification were observed.

Of the 232 holes noted, 209 (90.1%) are incomplete, and 23 (9.9%) are complete. This compares with 92.3% incomplete and 7.7% complete holes for the Ford Bridge Assemblage, and 87.8% incomplete and 12.2% complete holes for the Puckeridge Assemblage. The very close agreement between these figures tends to reinforce the impression that the makeup of this assemblage has not been biased by selective retrieval.

The testing may be discussed in slightly more detail in Tite and Freestone,2 and again in Tite, Freestone, Meeks and Craddock3 but, as before, unambiguous quantification is lacking. Additionally, it is not made clear in these reports whether the mould tested came from the Henderson Collection or from one of Partridge’s own finds of coin mould.4

Site Ford Br. Hend. Coll.

It is therefore not possible to say whether any of the material originally retrieved has been lost in the years since its deposition in Hertford Museum. This means that any statistically-derived conclusions must come with the caveat that the sample may have been distorted both during and following retrieval.

Av. holes in row Av. holes in col. 3.38 2.83 3.06 2.88

Figure 4.1: Average number of holes in rows and columns for fragments with more than 5 holes.

The similarity between the average numbers of holes in rows and columns for the Ford Bridge Assemblage and the Henderson Collection,7 as shown in Figure 4.1, above, is startling, and demonstrates very clearly that, in terms of the conformation of larger fragments, the two samples are nearly identical. It may therefore justifiably be claimed that, in terms of the preservation of larger fragments, the Henderson Collection is also very close to the Ford Bridge Assemblage.

Unlike some of the smaller assemblages of coin mould examined in this book, the Henderson Collection Tite, M and Craddock, P in Partridge, C, 1981: p. 326. Tite, M and Freestone, I, 1983. Tite, M; Freestone, I; Meeks, N; Craddock, P, 1985: pp. 50-55.By far the most interesting information in this and the paper above concerns the experiments carried out to relate the signs of heating visible on the material to actual temperatures. Samples of coin mould fragments were heated to accurately measured temperatures, and then thin-sectioned and examined using SEM. 4  Wickham Kennels (Partridge, C, 1982) and Gatesbury Track (Partridge, C, 1979). 1 

2  3 

Defined as: ‘(lacking edge) + (lacking either base or top) + (having no measurable holes)’ Hunn, J, 2007. 7  The large number of fragments with modern mending makes more problematic the derivation of similar averages for the Puckeridge Assemblage. 5 

6 

35

Making a Mint However, despite these similarities, the Henderson Collection differs significantly from other finds of coin mould from the Braughing area in several parameters. Many of these will be examined later in this chapter, but at this point it will be sufficient to mention tray thickness8 as a parameter for which the Henderson Collection has returned distinctive results.

It should also be noted that the distribution pattern of tray average thicknesses is unusual: it does not form a bell-curve, but instead shows two distinct peaks. This is unique in the larger assemblages so far examined, and it could be construed as suggesting that this assemblage comprises mould from more than one episode of tray manufacture. Of the 63 individually listed fragments, there are 32 middle fragments; 20 straight edge fragments; 1 curved edge fragment; 3 90o corners; and no oblique, curved or unquantifiable corners. There are 7 fragments with an unclassifiable position type.

45 40 35 30 25

Series1

20

As Figure 4.3 below shows, the composition of the Henderson Collection is not typical. The proportion of ‘middle fragments’ is unusually high, there seems to be a proportionate shortfall in ‘edge fragments’, and a positive dearth of corners. Since we can be reasonably certain that there was little or no selectivity exercised during and after retrieval, except perhaps a little inefficiency in the retrieval of very small fragments, it seems highly possible that these compositional anomalies are the result of selective deposition. It is quite understandable that very small fragments of coin mould might be overlooked by an inexperienced eye, but it is much less likely that recognizable edges or corners would be missed in this way – if they are not present in the Collection as presently constituted, it is probably because they were not present in the material originally deposited.

15 10 5 0 10 to 11 11 to 12 12 to 13 13 to 14 14 to 15 15 to 16 16 to 17

Figure 4.2: Tray average thickness9 in the Henderson Collection expressed as percentages.

There are 52 fragments from which at least one measurement of thickness can be obtained, ranging from 10mm to 16mm. This 6mm range of variation is unusually restricted when compared with either the Puckeridge assemblage (a range greater than 29mm) or the Ford Bridge assemblage(a range greater than 13mm). In addition, the Henderson material is on average substantially thinner than the bulk of the fragments from either of the other two major assemblages from Braughing/Puckeridge. This could be related to the prevalence of a different tray-making technique to those used to make the other Braughing mould fragments. This idea is explored in greater detail in Section iii, below. Furthermore, the intra-fragment average standard deviation for tray fragments with more than one thickness measurement in the Henderson Collection is, at 0.84mm, also substantially lower than for these two assemblages (1.30mm for Puckeridge; 1.17mm for Ford Bridge). It is hard to explain why the Henderson coin mould should exhibit such low variability, but it does enable us to distinguish this assemblage from all other finds of coin mould in the area.

Pos. type Hend. Coll.

Ford Bridge

Puck.

00 01 02 03 04 05 05+05 06 07

11.9% 33.9% 41.7% 0% 4.4% 5.9% 0% 0% 2.2%

2.6% 36.9% 43.5% 0.2% 5.5% 8.0% 0.2% 1.2% 1.9%

11.1% 50.8% 31.7% 1.6% 4.8% 0% 0% 0% 0%

Figure 4.3: Comparing composition - position types expressed as percentages of the total number of individually listed fragments.

Given that a demonstrable shortfall in corners has also been noted in both the Ford Bridge and Puckeridge assemblages, where they constitute respectively 12.5% and 16.8% of all individually listed fragments, the meagre 4.8% of the Henderson Collection becomes even more obviously unusual. It seems undeniable that corners have deliberately been removed from the assemblage prior to deposition, and that what remains is in a sense the antonym of a structured deposit – it represents what is left of a body of material following the removal of the elements of a structured deposit.10

Tray thickness is a parameter which we can be virtually certain was of no great significance to the makers, users and owners of coin mould, since there seems to be no correlation between thickness and tray form or hole diameter. There could be a functional element to the choice of thickness for a given tray, in that – if hole depth was a significant choice – the tray had to be thick enough to accommodate that depth . This idea will be examined more fully in Chapter 7 . Otherwise the thickness of a tray was dictated either by the depth of the mould in which it was made, or by the whim of the maker if a mould was not being used. 9  Tray average thickness = average thickness per tray, obtained by adding together all thickness measurements for a given fragment and dividing by the number of thickness measurements for that fragment. 8 

10 

36

See below, Chapters 5, 6 and 7 for further exploration of this idea.

Chapter 4: The Henderson Collection (Braughing) coin mould assemblage would not be unreasonable to assume that some, at least, of the parent trays from which the Henderson Collection fragments derive were also of this form.

The overall condition of the Henderson Collection coin mould is very slightly worse than that of the Ford Bridge Assemblage, in that surface finishes are not as well preserved.

3 Edge Profiles

2 Tray forms

Of the 63 individually listed fragments, 24 have edge profiles. Twenty-one of these have one edge profile only; three have two edge profiles. No fragment in the assemblage has more than two edge profiles.

While there is no absolute proof of tray form within the Henderson Collection, there are nonetheless several pieces of evidence which, when taken together, form a reasonably coherent and convincing pattern.

Six edge profile forms have been noted in the assemblage:

First of all, there are the three right-angled corners. On their own, these are not diagnostic of any tray form, although they could not derive from circular or hexagonal11 trays. We are left, therefore, with three possible forms: rectangular (as seen at Saintes and – possibly – Old Sleaford), square (similar to the Puckeridge or Scotch Corner forms) or pentangular (Verulamium form).

Code 1 2 3 4 5 6 9

Profile Type I-Section Lazy S Straight section Angled section Rolled edge Overhang Unquantifiable

Freq. 1 5 10 4 1 1 5

% of Profile total 3.7 18.5 37.0 14.9 3.7 3.7 18.5

Figure 4.5: Edge profile distribution.

Next, there are two fragments, HC/04 and HC/12, which could be interpreted as bearing traces of a Verulamium form apex hole, although this identification cannot be regarded as certain.

Instances of ‘cut and tear’ banding were observed, but were not noted. Of the 24 edge fragments in the sample, 14 are of types (Straight Section and Angled Section) that are suggestive of cutting rather than moulding. By far the greater portion of the trays from which the Henderson assemblage derives to which a method of manufacture might be ascribed would seem, on balance, to have been made by cutting to shape from slabs of clay, rather than by the use of either type of mould. This is again unusual – occasional instances have been observed in both the Puckeridge and the Ford Bridge assemblages, but not (so far) in any other body of coin mould.

Finally, there is fragment HC/30 (see Fig. 4.4 below), on which the hole-row runs very clearly at an angle to the tray-edge, in a way that is usually characteristic of a pediment edge.

This preponderance is also significant because it indicates a certain homogeneity in respect of manufacture in the Henderson Collection. The concept of homogeneity will appear in different contexts later in this chapter. That there is evidence for the use of at least three different manufacturing techniques in such a small assemblage12 is surprising. It may be that, given the smallness of the sample, this can be explained as an accident of preservation, but there is good reason from the distribution patterns of other types of data drawn from the material to suppose that the Collection contains a genuinely representative sample. If it is assumed that this is evidence of three or more workers producing pellet mould at the same time, each with their own method of making trays, then one would have to explain the enormous difference in productivity between the worker(s) using a presumed cutting process, and the workers using either tray-mould type.

F igure 4.4: Record-card diagram of fragment HC/30

Although we should be clear that none of the above is much better than circumstantial evidence, nonetheless – bearing in mind also that by far the largest proportion of all the coin mould found in the Braughing/Puckeridge area to which a tray form can be ascribed with certainty is of the Verulamium ‘square and pediment’ type - it Posited for some fragments from Colchester (Hawkes and Hull, Camulodunum, pp. 129-133, cited in Clifford, E, 1960). It seems much more likely, however, that the fragments interpreted as deriving from hexagonal trays are actually pedimental fragments from Verulamium form trays (See Figure 4.9 above).

11 

‘Lazy S’ and ‘I-Section’ are the result of the use of different types of tray-mould (see above, Chapter 3, §3).

12 

37

Making a Mint A third theory is better able to explain the data pattern. First, there is no reason to assume that all the mould in the Collection was made at exactly the same time. Second, there is no reason to assume that the quantity of trays manufactured at a particular time was directly linked to the quantity of pellets cast during a particular episode. If one follows Haselgrove13 in his suggestion that the manufacture of pellets was not a process carried out on a continuous basis, but was instead episodic or periodic, and perhaps linked with specific occasions, such as the payment or receipt of tribute, or religious festivals, or the regular advent of fairs or other trading opportunities, then it is reasonable to assume that pellet mould could be made in advance of the occasion of use. If this were the case, then there is no reason why the amount of pellet mould made should have tallied precisely with the amount required on a particular occasion.

The complete absence of ‘incised guidelines’ in the Henderson Collection coin mould is a further point of difference between it and the Ford Bridge and Puckeridge assemblages. Given a the frequency with which incised guidelines occur in the Puckeridge coin mould (11.8% if individually listed fragments), one might have expected to see 7 examples in the Henderson Collection, and there absence might therefore be regarded as conclusive, were it not for the paucity of edge fragments noted above. 6 Methods of hole manufacture There are 3 fragments, or 4.8%, exhibiting hole slighting out of a total of 63 measurable fragments. No fragments exhibit signs of boustrophedon dibbing. The hole slighting percentage is in keeping with the 5.9% and 6.4% for hole slighting in the Ford Bridge and Puckeridge assemblages. All the evidence is, therefore, that the mould holes of the Henderson Collection assemblage were not made using a ‘peg-board’.

The tendency, surely, would have been to ‘make sure that there was enough’ – in other words, to manufacture a surplus above anticipated requirement, ‘to be on the safe side’. Were any mould to be left over at the end of an episode of casting, this would be no disaster: it could simply be saved until next time. It seems entirely possible that the evidence for at least three methods of manufacture in the Collection, and the small number of examples of two of these methods, is the result of stockpiling. The bulk of the Henderson material was made using one or other of the cutting methods, and the mould made forms represent some suitable oddments left over from a previous episode used to make up the numbers.

No fragment in the Collection has enough holes to enable meaningful examination of inter-hole spacings for repeated patterns. As a consequence of this and the absence of hole-slighting in more than one axis on a single fragment, it is not possible to prove conclusively that holes were made individually using a single-pronged dibber, although the presence of a single hole (out of three on the fragment) on HC/43 exhibiting an ‘annulus and pit’ marking generally associated with ‘double dibbing’ would also seem to indicate a lack of uniformity in hole making on a single fragment that is more consonant with individual, rather than multiple, manufacture. However, it is possible to state with certainty that, if a multi-pronged dibber was used to make the holes in the Henderson Collection coin mould, it was not used in the manner Elsdon suggests,15 since intra-fragment hole depth varies widely.16

That cutting to shape seems a somewhat ad hoc method of tray manufacture seems to accord well with a picture of expediency, and even haste. It is also a clear point of difference between the Henderson Collection and the Ford Bridge and Puckeridge assemblages.

It should also be noted that average intra-fragment standard deviation in all three hole parameters is significantly greater than was obtained from the experimental manufacture of holes using a singlepronged dibber. This would seem to suggest that not only was a single-pronged dibber used to make the holes in the Henderson Collection coin mould, but also that it was used with less care than was exercised in the experiments. The fact that average intra-fragment total variation in these parameters does not present the same picture perhaps reflects the small average size of fragments in the Collection, in line with the tendency noted elsewhere for variation range to increase in tandem with fragment size.

4 Edge markings Aside from the instances of ‘cut and tear’ banding noted above, no edge markings were observed on the Henderson Collection material. The absence of even a single instance of ‘band and lines’ markings is another point of difference between the Henderson Collection and the other two major finds of coin mould from the Braughing/Puckeridge complex. 5 Evidence of elaboration There were no instances of the ‘cleavage grooves’ observed by Elsdon14 on the Old Sleaford material.

15  Elsdon, S; 1997: p. 54: ‘...a stick with six protuberances of carefully regulated depth’ 16  See below, §9.

Haselgrove, C, 1987: p. 29. 14  Elsdon, S, 1997: p. 53. 13 

38

Chapter 4: The Henderson Collection (Braughing) coin mould assemblage Origin Base dia Top dia. Depth

Hend. Coll 0.60 0.82 0.91

discovered that the average difference between base and top hole diameters was 2.2mm – but as Figure 4.7 below demonstrates, the difference shown by individual fragments can be considerably greater.

Exp. 0.46 0.52 0.69

Figure 4.6: Average intra-fragment standard deviations in three hole parameters compared - How careful were the mould-makers?

Furthermore, although there is no absolute proof in this assemblage that all the holes were made individually, it is also true to say that there is not a single shred of evidence to suggest that holes were not made individually. It is at this point that reference must be made to the context of the Henderson Collection. Not only does it derive from an area where eight other finds of pellet mould have been made, as well as the evidence relating to tray form there is indisputable evidence that sets it firmly within the wellunderstood local tradition of pellet mould manufacture.17 Both the Ford Bridge and Puckeridge assemblages have yielded strong evidence that their mould holes were made individually, and we have no reason to suspect that the Henderson Collection material differed in this respect.

Range in mm

Hend. coll.

Ford Br.

0 to 1

0

7.83

1 to 2

34.62

40.00

2 to 3

38.46

33.04

3 to 4

23.08

17.39

4 to 5

3.85

1.74

Figure 4.7: Variability in relationship between top and base hole diameters in the Henderson Collection and the Ford Bridge assemblages expressed as percentages of the number of holes in each assemblage exhibiting both measurements.

Seventeen out of a population of 26 fragments, 65%, show a difference between base and top hole diameters larger than the range of one of Elsdon’s hole size groups, and there is a total range of variation of 4mm – equal to two of Elsdon’s hole size groups – in the relationship between base and top diameters in this assemblage.

7 Predictable relationship between base and top hole diameters Because the metal of a coin pellet would have been cast at the bottom of a mould hole, and not at the top, if it is claimed18 that there is a predictable relationship between top diameter and ‘coin module’, a corollary of this is that there must be a predictable relationship between base and top hole diameters, since it is the bottom diameter which dictates the maximum diameter of a pellet cast in a given mould hole, and that therefore it must, of necessity, be possible to infer top diameter from base diameter, and vice versa. Certainly it would be unreasonable to expect absolute precision, but it is not unreasonable to define limits to acceptable variability in the relationship which, if exceeded, would invalidate the claim. Since Reynolds19 follows Tournaire in assuming that top diameter alone is the measure of ‘coin module’, it has been decided to use the 2mm total range he has chosen for two of his hypothetical groups of hole diameters,20 which equates to a ± 1.0mm variation, and which is also the maximum variation between base and top diameter claimed by Tournaire21 of the pellet mould from Levroux on which he bases his ideas.

Even using the parameters espoused by the proponents of this idea, it is obvious that this relationship is not predictable, and therefore the claim that coin module is related to top diameter is not true of the Henderson Collection. Figure 4.8 below, showing the frequency with which various top diameters occur, demonstrates that, in the Henderson Collection, top diameter is more variable than base diameter: while base diameter has a 4mm spread, top diameter has a 5mm spread, nearly as great as the smallest top diameter, and 50% of the largest diameter. This means that there is absolutely no reason to suppose that there is any particular association between top diameter and coin module in the Henderson Collection assemblage. 18 16 14 12 10

In order to examine alone the variability in the difference between top and base hole diameters independently of actual diameter measurements, a figure for variation for each fragment was generated by subtracting average base diameter from average top diameter. It was

Series1

8 6 4 2 0

See below, §10. Tournaire, J, in Tournaire et al., 1982: p. 419 ; Reynolds, J in Elsdon, S, 1997: p. 59. 19  Reynolds, J, in Elsdon, S, 1997: p. 56 20  4 – 6mm and 7 – 9mm. 21  Tournaire, J in Tournaire et al., 1982: p. 419. 17 

6 to 7

18 

7 to 8

8 to 9

9 to 10

10 to 11

Figure 4.8: Top diameter distribution in the Henderson Collection

39

Making a Mint First, it should be noted that 9/14 fragments with 2 or more measurable holes show variation in diameter, and that of the 5/14 fragments which show no variation, none has more than 3 holes. The most obvious interpretation of these statistics is that intra-fragment variation in diameter is the norm, and that 7/9 fragments showing variation have at least one hole with a diameter of 6mm. This strengthens the impression already formed that 6mm is the central value of the assemblage, around which the other values cluster.

8 Predictable relationship between base diameter and pellet module As Figuress 4.9 and 4.10 below demonstrate very clearly, although the hole base diameter range of the Henderson Collection is relatively restricted when compared with either of the larger Braughing/Puckeridge assemblages, at 4mm it is still larger than any of the hole diameter groups posited by Reynolds.22 Range 3 to 4 4 to 5 5 to 6 6 to 7 7 to 8

Count 0 4 3 25 3

Despite this, it would seem extremely difficult to deduce from the base diameter alone what the pellet size might have been: the range of variation is nearly equal to the smallest diameter in the group. When the average intrafragment standard deviation in base diameter (±0.59mm) is factored in, the situation becomes even more complex and indeterminable. There is, as Clifford23 pointed out over 50 years ago, no clear and simple link between base diameter and denomination.

Figure 4.9: Fragment average base diameter distribution in the Henderson Collection. 30 25

9 Hole depth

20

Series1

15

There are 138 holes on 47 fragments in the Henderson Collection from which depth measurements can be obtained. The shallowest is 3mm; the deepest is 9mm. As set out in Chapter 3, §10, fragment average hole depth has been derived from individual hole measurements, and average intra-fragment variability derived from individual fragment total variation in depth, and it is these averages which form the basis of the discussion that follows.

10 5 0 3 to 4

4 to 5

5 to 6

6 to 7

7 to 8

Figure 4.10: Fragment average base diameter distribution in the Henderson Collection expressed graphically.

However, the distribution of diameters forms a coherent pattern – there is no ‘diastema’ like the 3mm gap in the base diameter distribution of the Ford Bridge assemblage. Once again, the Henderson Collection material is particularly homogenous. There is only a single significant cluster around 6mm – 7mm, and this, together with the restricted diameter range, tends to suggest that the coin mould in this assemblage was most possibly used to produce a single size of pellet. Frag. ID. HC01 HC02 HC04 HC05 HC06 HC07 HC12 HC14 HC18 HC22 HC24 HC32 HC35 HC42

4mm

1

1

Hole Diameter 5mm 6mm 1 4 2 4 1 1 3 3 1 1 2 1 2 2 2 2

As Figures 4.12 and 4.13 below demonstrate, interfragment variation in depth in the Henderson Collection is, in several respects, distinctive. First, at only 6mm, the total range of variation is significantly smaller than for either of the two major Braughing/Puckeridge assemblages. Additionally, the distribution of depths is biased appreciably towards the lower end of the scale, while the classic bell-curve formed would seem to suggest that, whatever selectivity was exercised prior to deposition, it has not significantly distorted the sample.

7mm

Range in mm 2 to 3 3 to 4 4 to 5 5 to 6 6 to 7 7 to 8 8 to 9

1 2 1

1

Figure 4.12: Fragment average hole depths in the Henderson Collection

1

Figure 4.11 : Intra-fragment diameter variation in fragments with 2 or more measurable diameters 22 

Count 0 1 11 24 8 2 1

Reynolds, J, in Elsdon, S: 1997: p. 56.

23 

40

Clifford, E., 1960: p. 146.

Chapter 4: The Henderson Collection (Braughing) coin mould assemblage Moreover, as Figure 4.15 below shows, when fragment average hole volume is plotted against fragment average hole base diameter, the assemblage clearly forms a single cluster with a continuous series of values.

60 50 40

Av vol

Series1 250

30 20

200

10

150 Av vol

0 2 to 3

3 to 4

4 to 5

5 to 6

6 to 7

7 to 8

100

8 to 9

50

Figure 4.13: Fragment average hole depths in the Henderson Collection expressed as percentages.

0 0

Average intra-fragment variability in hole depth in the Henderson Collection is, at 1.79mm, well within the normal range for Braughing/Puckeridge material. This gives additional assurance that the coin mould in this assemblage has not undergone significant modification since retrieval (the significance of this will become apparent when the Wickham Kennels assemblage is discussed). It also demonstrates that, despite its many untypical features, the Henderson Collection mould is still closely related to the main Braughing/Puckeridge minting tradition. It also suggests that the holes in all three of the larger coin mould assemblages from this area are likely to have been made in the same way.

4.0mm

Diameter range

4.0mm

Largest volume

205mm3

Smallest volume

50mm3

Volume range

155mm3

8

The most obvious point to make is that the largest pellet that could be made in the Henderson mould is very much smaller than the largest possible pellet which could have been cast in the larger holes of either the Puckeridge or the Ford Bridge assemblages. Second, if Tite and Freestone are correct, and the Henderson Collection material was used to produce a predominantly silver coinage, then fusion of pellet with mould would have been less of an issue than would have been the case were the pellets largely copper. This may mean that it would have been possible to place proportionately more metal in each mould hole. Since twenty (57%) of the 35 holes in the assemblage from which it is possible to obtain a result have volumes greater than 125mm3, and 15 holes (43% of the total) have volumes less than 125mm3, it would seem possible that the assemblage could have been used for making two modules of pellet, the larger of which could not have been greater in volume than an Icenian silver unit, and the smaller of which would have been less than 2/5 that size. However, while it might be theoretically possible, this cannot be considered likely, given the difficulty of judging by eye the volume of holes with very similar diameters.

The largest fragment average hole volume is more than four times the size of the smallest, and it is hard to reconcile this glaring disparity with the idea of a standardized, or of a regulated, coinage. Yet even so, this is still a relatively tight, cohesive, homogenous assemblage – comparable variation in the Ford Bridge assemblage is more than 650mm3 in each of the two observable hole groups, more than 1300mm3 overall.

Smallest hole base diameter

6

If this interpretation of the data is accepted, and the Henderson Collection material is to be seen as a single group, and given that, although there is no lower limit on the size of pellet that can be cast in a mould hole, there is most certainly an upper limit, it may actually be possible to make some tentative statements about pellet module in this assemblage.

Although the relatively small diameter of the holes in this assemblage means that in absolute terms the quantities involved are far smaller than were considered in Chapter 2, §9, Figure 4.14 below demonstrates that in relative terms this variation is very considerable indeed.

8.0mm

4

Figure 4.15: Average hole volume plotted against average hole base diameter.

10 Control of hole volume

Largest hole base diameter

2

In view of the fact that, with its restricted hole size range and unusual homogeneity in other parameters, the Henderson Collection coin mould represents nearideal conditions for the drawing of conclusions about

Figure 4.14: Hole size variation in the Henderson Collection tabulated.

41

Making a Mint pellet module, the conclusion to be drawn from the very rough estimates deduced above is that only the vaguest idea of the pellet module cast in a given assemblage can be obtained from the coin mould itself. The only way that more precise figures can be obtained is with the discovery of intact trapped pellets. Origin

Hend. Coll.

Exp.

Base Dia.

1.00

2.25

Top Dia.

1.17

2.18

Depth

1.66

2.17

An explanation based on the data already recovered might be that, in view of the very high proportion of ‘Straight Section’ edge profiles, which would seem to have been produced by cutting rather than moulding, a method of tray manufacture which has about it something of the ad hoc, and in view of the relatively restricted range of hole size in the assemblage, that the Henderson Collection coin mould – in contrast to most of the assemblages so far examined – was made in response to an unexpected requirement for coin. Given that no fragment with a ‘Straight Section’ edge profile has a trace of chalk wash, it is possible that the fragments which do have chalk wash traces represent spare trays curated from previous episodes of coin manufacture, and used on this occasion to ‘make up the numbers’. Although this may seem a rather tenuous chain of reasoning, additional indirect evidence will be adduced below which would seem to support it.

Figure 4.16: Average intra-fragment total variation in three hole parameters in the Henderson Collection compared with experimentally generated data.

11 Calcium carbonate traces Only two certain instances, and one possible instance, of chalk wash were noted in the Henderson Collection, a meagre 4.8% of the total number of fragments, almost ten times less common than in either the Puckeridge or the Ford Bridge material.

12 Proportions of used and unused mould fragments Of the 62 individually recorded fragments, only one shows signs of vesiculation, and two of vitrification. As noted in §10 above, 21 fragments show signs of reddening, but this in itself cannot be taken as confirmation of use.

There are two possible reasons for this.

Craddock and Tite24 failed to note the two examples of vitrification25 when they examined the assemblage in 1981, and instead attribute use on the basis of surface blackening. Observation of freshly recovered coin mould from other Braughing deposits has shown, however, that this blackening is more likely to be a taphonomic phenomenon resulting from the deposition of mould fragments in close contact with charcoal, than a sign that the material has been heated to the very high temperature necessary to achieve pellet fusion.

The first is that this is a function of the slightly poorer state of preservation of the Henderson Collection pellet mould than the Ford Bridge material: surface finishes are generally not in good condition, and it is entirely possible that much of the evidence for chalk wash has been destroyed following deposition. The second is that if the assemblage was used in the production of a largely silver coinage, then the maintenance of reducing conditions – the chief function of chalk wash – would not have been an absolute imperative. In other words, the infrequent occurrence of calcium carbonate traces could be the result of predepositional factors.

Nonetheless, their observation that use need not be associated with supra-microscopic signs of heating holds good. In particular, the very low incidence of vitrification may well be associated with the poor preservation of surface finishes noted above.

However, to this it might be objected that if chalk wash was not necessary for the production of largely silver pellets, then why do any fragments at all in the Henderson Collection have chalk wash?

There are, however, four more possible reasons for this observable paucity of evidence that the Henderson mould has been heated to the high temperatures required for casting pellets.

Since only a very small proportion of the assemblage was tested for metal residues, then it would not be impossible that some fragments in the assemblage might have been used in the production of a copperrich coinage. However, this hypothesizes the existence of undiscovered information, and does not take account of the fact that 21 fragments exhibit some degree of reddening, which is firm evidence that at some point between tray manufacture and deposition they have been heated in the presence of oxygen.

The first possibility is that the overwhelming majority of fragments in the assemblage has not been used. Against this may be advanced that fact that Craddock and Tite discovered significant traces of silver and copper on at least one fragment which, by their own observation, did not exhibit signs of extreme heating.

24  25 

42

Craddock, P and Tite, M in Partridge, C , 1981: p. 326. Frags. HC/0/41 and HC/0/49

Chapter 4: The Henderson Collection (Braughing) coin mould assemblage The second possibility is that the clay from which the fragments were made was highly refractory, with a high proportion of kaolin in its composition, as with the Levroux mould fragments. However, the clay from which the Henderson mould was made does not seem in any way different from the clay used to make both the Ford Bridge and the Puckeridge assemblages, both of which have yielded numerous examples of vitrification, and of vesiculation, which in the Henderson assemblage is even less common than vitrification.

assemblage (where it is more common than in any other fully-examined assemblage) is only 1.8%, it is not possible to state with any degree of certainty that its absence from the Henderson Collection is significant of anything more than an accident of preservation. However, it should be borne on mind that it could equally result from tray manufacture occurring at a point other than harvest-time.

The third possible reason is that the chemistry necessary for vitrification was not present when the Henderson mould was used. Although it has been surmised that vitrification may be dependent on the presence of alkalis and/or alkaline-earth elements in the clay, or from woodash in the furnace,26 it is hard to imagine how wood-ash in particular could have been excluded from the process.

No examples of inclusions or tempers were noted in the Henderson Collection. This is another clear point of difference between it and the Puckeridge and Ford Bridge assemblages, and is therefore significant, since it confirms that the Henderson material most probably belongs to a separate episode of pellet manufacture, and that therefore it is not unreasonable to conclude that the circumstances of its manufacture could well be different.

15 Inclusions and tempers

The fourth possibility is that temperatures sufficient to induce either vitrification or vesiculation were not reached. This is entirely credible, since the melting point of silver – 961oC in the pure state – can be lowered to 879oC by alloying,27 which would appear to be somewhat cooler than the temperature required to induce vitrification in the Braughing clays.28 The absence of any signs of deformation or heat-induced slumping would seem to add force to this interpretation.

16 Clay caps and luting There are no examples of either clay caps or luting in the Henderson Collection. If it is accepted that the retrieved assemblage is a reasonable reflection of the size of the pre-depositional assemblage, and therefore that the material represents the remains of a relatively small episode of minting, then this absence becomes entirely understandable, since indirect evidence has already been adduced that the number of mould trays was adjusted to fit the desired quantity of coins to be minted,30 and that the episode was carried out as a ‘single commission’.31 Under these circumstances, it seems unlikely that expedients such as clay caps or luting would be required.

The salient conclusion to be drawn from this discussion is that the contention that attribution of use on the basis of supra-microscopic examination alone is not possible29 is correct. 13 Grass marks, chaff marks and matting marks

17 Conclusions

The absence from the Henderson Collection coin mould of a single example of any of these markings can be interpreted in two ways.

Although the Henderson Collection assemblage of coin mould is neither exceptionally large, nor possessed of any startlingly novel features, it is nonetheless exceptionally important for our understanding of the social, political and economic contexts within which minting took place in the Late Iron Age and Roman periods, both at Braughing, and more widely across Britain. It is clearly unfortunate that, in common with so many assemblages, this material comes without any certain temporal context.

The first is that the mould trays from which the assemblage derives were not laid on grass, chaff or matting while they were drying. The second is that this absence is a reflection of the slightly poorer state of preservation of the Henderson material. While it is not possible to resolve this with absolute certainty, it would seem that the latter explanation is probably to be preferred. 14 Grain casts

30  See above, Cap. 4, §3, where it is suggested that the presence in very small numbers of two of the three edge profile types represents the inclusion in the assemblage of trays left over from previous episodes of minting; and also Cap. 4, §10, where it is surmised that the presence of a few fragments exhibiting traces of calcium carbonate (which would have been unnecessary in the production of predominately silver pellets) also suggests the inclusion in the assemblage of ‘leftover’ coin mould from previous episodes of minting. 31  See above, Cap. 4, §5: if it is accepted that incised guidelines are in some way indicators of ownership, their absence from the Henderson Collection material could be interpreted as a sign that only one authority was involved in the commissioning of the coin manufactured in this episode.

No grain casts were observed on any fragment in the Henderson Collection. However, given that the rate of incidence of this phenomenon in the Ford Bridge Tournaire, J and Henderson, J in Tournaire et al., 1982: p. 432. Kitco Metals Inc. website. Inferred from Tite, M and Freestone, I: 1983. They seem to suggest that vitrification occurred in the Braughing sample they tested at around 950oC. 29  See above, Cap. 3, §13 26  27  28 

43

Making a Mint Central to this importance is its location. Its findsite, although not absolutely certain, is within 500 metres of the findspots of all the other coin mould assemblages from the Braughing/Puckeridge complex, excepting (perhaps) only the Puckeridge assemblage. This gives comparison between it and the other Braughing material particular force.

As pointed out, these are the lowest possible figures to account for the retrieved assemblage – there is good reason to assume that the pre-depositional total of trays was almost certainly higher: the undeniable 25% shortfall in edge fragments demonstrates that a substantial fraction of the original assemblage was not deposited with the retrieved portion.

The presence of chalk wash on the Henderson material is a very strong indication that this assemblage is firmly within the Braughing/Verulamium pelletmaking tradition. However, if it is accepted that the Ford Bridge and Puckeridge assemblages are to be regarded as ‘typical’,32 then the clear differences at every level of observation between these two very large bodies of material and the Henderson Collection must be significant of different circumstances surrounding the episode of minting represented by the Henderson material.

To paraphrase Pete Seeger, where have all the corners gone? It was suggested in Section i. of this chapter that the Henderson material might represent the ‘antonym’ of a structured deposit, and that what we have is what remained after the ‘significant’ bits had been removed. This theory entails the existence of structured assemblages of coin mould, but before we can look for such assemblages amongst the various deposits of coin mould so far retrieved, we should perhaps look to what seems to be missing from the Henderson Collection.

The first significant difference must surely be the lack in the Henderson Collection of both edge and corner fragments. If this assemblage had approximated the percentages noted for the two larger assemblages,33 then one might have expected to see between 7 and 11 corners (as opposed to the actual figure of 3), and some 27 edge fragments (compared with the actual figure of 20). It should be noted that these projections assume a similar proportion of Verulamium and Puckeridge form trays to the two larger assemblages, whereas all the evidence (admittedly far from conclusive) is that only a single tray form, the five-cornered Verulamium form, was present in the Henderson material, and that therefore this assemblage would actually have had a 10% greater number of corners.

First, it would seem likely that such deposits would tend to be at the lower end of the size spectrum for coin mould assemblages, since certain parts only of the parent tray seem to have been selected for removal.35 The second plausible feature one might hypothesize for a structured deposit of coin mould can be drawn from the type of mould fragment that appears to have been preferentially removed from the Henderson material (and, to a lesser extent, the Ford Bridge and the Puckeridge assemblages). It would seem reasonable to suggest that a structured deposit of coin mould might have a statistically unusual makeup – an unusually high proportion of corners and/or edge fragments would be a good intimation that a particular assemblage might be structured.

It is possible to look at this from another angle: there are 232 holes and traces of holes in the Henderson material. If, as the evidence suggests, the original trays were of the 7 x 7 + 1 Verulamium form, the smallest number of trays forming the pre-depositional assemblage would have been 5, which would make a minimum total of corners (before deposition) of 25. If a subrectangular 6 x 10 hole ‘Saintes’ form is assumed for this assemblage, the smallest number of trays required would be 4, making a total of 16 corners; if the maximum 7 x 11 hole form posited by Elsdon34for the largest Old Sleaford fragment, the minimum number of trays required would again be 4, and the total of corners 16.

If an assemblage is both small in size and unusual in composition,36 we may suspect that its deposition was not casual, and that in consequence - where these are known with any certainty - both the circumstances of deposition and the composition of the deposit (what else, besides coin mould, was deposited?) should be noted and considered with care and close attention: it is from these details that one might hope to glean insights into the symbolic thinking of the smiths, the moneyed classes, and the wider society within which they lived. However, it is not merely the possible existence and characteristics of structured deposits of coin mould that can be deduced from the Henderson Collection. In

See below, Cap. 5 and Cap. 6: there are sufficient differences between these two assemblages to enable their differentiation, and sufficient similarities both in composition and in minute detail to permit the conclusion that they are very closely related in time as well as space. 33  See below, Cap. 5, §1 and Cap. 6 §1, where it is demonstrated that both of these assemblages are themselves substantially deficient in corners. 34  Elsdon, S, 1997: p. 54. 32 

Although it must be borne in mind that an impression of the preferential removal of corners could arise simply because a shortage of corners in an assemblage can be demonstrated with some certainty, while an absence of middle fragments can not. 36  Simon West (pers. comm, 2010) has already suggested that at least some of the smaller deposits of coin mould might well derive from much larger pre-depositional assemblages. 35 

44

Chapter 4: The Henderson Collection (Braughing) coin mould assemblage more than one section above, reference has been made to the homogeneity of the assemblage, in terms of the evidence of methods of tray manufacture derived from the typology of edge profiles; in terms of the restricted diameter range and depth of the holes when compared with several other assemblages; in terms of the restricted range of variation in the thickness of the tray fragments. In all of these measurable aspects it has been demonstrated that the Henderson Collection conforms to a single pattern, and therefore it seems at least possible that there is a single explanation underlying this. Might we be looking at a single episode of minting in which only one denomination of coin was being made (although, of course, we can never be certain of the denomination)? Assuming that this hypothesis is correct, we can derive useful guidelines from the parameters of this assemblage, the most obvious of which is an evidentially-derived figure for a single module diameter range. Certainly, the thread on which this depends is tenuous, but this is still a significant improvement on the purely arbitrary – and it ties in well with the range of inherent diameter variability observed in experimental tray manufacture. The 4mm base diameter range of the Henderson Collection coin mould can therefore help to evaluate the module diameter ranges postulated by Elsdon for the Old Sleaford material. We might also use this 4mm range as a yardstick against which to measure other assemblages of coin mould. Nonetheless, it should be remembered that the best that can be achieved using any diameter-based evaluation of module is a demonstration that an assemblage could have been used to make more than one denomination of coin – and that to prove possibility is a very long way from proving actuality.

45

Chapter 5

The Ford Bridge (Braughing) assemblage greater incidence of heat-induced ‘bloating’3 in the latter material.

1 General observations In the opinion of Jonathan Hunn, director of the Ford Bridge excavation, and Isobel Thompson of the Hertfordshire County Council Historic Environment Unit, the Ford Bridge coin mould must have been redeposited in antiquity, since the context from which it was retrieved seemed far more likely to be Roman than Late Iron Age.1 While this still remains an explanation entirely consistent with the material evidence, there is another possibility to be entertained – that this is a mainly post-conquest deposit arising from post-conquest activity, and that this activity included the making of coin pellets. This will be aired more fully both at the end of this chapter, and in Chapter 12, §B.1.c.

It is markedly thicker than the all-sample average: this accounts for the greater weight:hole ratio for this assemblage when compared with the Old Sleaford material. The standard deviation is also somewhat greater than the all-sample average, suggesting that trays habitually – and perhaps designedly – varied in thickness. The assemblage averages for Thickness 1, Thickness 2 and Thickness 3 confirm this tendency for the trays from which the retrieved fragments derive to be thicker in the middle than at the edges. This is not precisely characterized in the raw data, but the conclusion is nonetheless firm, because the Thickness 1 and Thickness 2 measurements comprehend the bulk of the data obtained from fragment edges.

The Ford Bridge Assemblage comprises 1153 fragments of clay coin mould, ranging in size from 90.38mm to 5mm, and the data from them is listed on 684 pre-printed record cards. There are 89 bulk bags containing more than one fragment, and 595 bags containing single fragments.

Origin Thick. 1 Thick. 2 Thick. 3 Aggregated average

The total number of fragments in bulk bags, 558, represents 48.8% of the total number of fragments, compared with 33.7% in the Puckeridge Assemblage. This disparity could be attributed to poorer preservation at Ford Bridge, but in view of the demonstrably higher standards of excavation at Ford Bridge, it seems more likely that this results from more efficient retrieval.

Ford Bridge 19.64 19.27 20.60 19.84

Study Average 18.03 17.47 18.55 18.10

Figure 5.1: Average thickness measurements from the Ford Bridge assemblage compared with study averages.

As Figure 5.2 below demonstrates, once again the ‘thickness signature’ is peculiar to this assemblage. Since, as noted above, these measurements are much less skewed by heat-induced distortion than the same measurements for the Puckeridge assemblage, we can be reasonably certain that the data we have is an accurate reflection of the original material before use.

There are 595 individually listed fragments in the assemblage. The average size of an individually listed fragment is 27.19mm (Length 1) x 25.15mm (Length 2). Of the 2219 mould holes noted, 2049 are incomplete and 170 are complete.

20 18 16 14 12 10 8 6 4 2 0

In terms of thickness, the Ford Bridge coin mould ranges from just over 13mm to slightly less than 26mm, more than twice the 6mm range seen in the Henderson Collection assemblage, and less than half the 29mm range of the Puckeridge assemblage. It has been surmised2 that disparity in thickness between the Ford Bridge and the Henderson Collection mould fragments may be the result of the use of different tray manufacturing techniques, whereas the very large disparity between the Ford Bridge and Puckeridge assemblages would seem to reflect the

Series1

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 to to to to to to to to to to to to to to to to 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

Figure 5.2: Fragment average thicknesses expressed as percentages.

Thompson, I, 2014: ‘Apart from the possible middle Iron Age sherd, none of it [the pottery assemblage] suggests a broad date range and in fact all of the grog-tempered pottery could be post-conquest, or at least no earlier than the broad middle of the 1st. century. The date of deposit in these contexts, of course, is another matter, as this must be later.’ 2  Chapter 4, §1 1 

This term has been borrowed from Tite and Freestone, 1983. It is used here to denote the third most extreme form of vesiculation. As Tite and Freestone describe, the next stage involves the collapse of the vesiculation, which has been characterized in this work as ‘sagging’ and ‘slumping’. The scale terminates with ‘heated beyond use’.

3 

46

Chapter 5 : The Ford Bridge (Braughing) assemblage Conversely, all edge and corner fragments with holes with a base diameter of 15mm or greater were consonant with a rectangular or sub rectangular tray form. This is also the case in the Puckeridge Assemblage, in which the presence of two larger fragments enabled the demonstration of the 5 x 5 hole Puckeridge tray form. Although the Ford Bridge Assemblage contains no such diagnostic fragments, the very close stylistic parallels between the two assemblages would seem to suggest that – as is surmised of the Puckeridge Assemblage – the larger hole diameters are to be associated with trays of the Puckeridge form. It has been suggested that the 5 x 5 arrangement was adopted because a tray with the usual 7 x 7 + 1 format and holes of 15mm or greater would have been impracticably large and fragile.

Of the 595 individually listed fragments, there are 202 middle fragments; 248 straight edge fragments; no curved edge fragments; 26 90o corners; 35 oblique corners and 13 unquantifiable corners. There are 71 fragments with an unquantifiable position type. The standard of preservation of the retrieved portion of the assemblage is good, better than either the Henderson Collection material or the Turners Hall Farm coin mould, although not as good as the Puckeridge Assemblage. Some fragments retain vitrified surfaces unabraded; many fragments retain traces of chalk wash in holes, fewer retain traces on surfaces. The proportion of very large fragments of coin mould is lower in the Ford Bridge Assemblage than in the Puckeridge Assemblage, and there are far fewer conjoining fragments (three, or 0.26%, in the Ford Bridge material against 58, or 2.12%, in the Puckeridge Assemblage). These disparities could be explained by the limited scope of the Ford Bridge excavation, which dealt only with the edge of the coin mould deposit, where the greatest disturbance might be expected to occur.

The proportion of holes with a base diameter greater than or equal to 15mm to holes with a base diameter of less than or equal to 14.5mm in the Ford Bridge Assemblage is 23 out of 595 individually listed fragments, or 3.9%, compared with a ratio in the Puckeridge Assemblage of 143 out of 1815 individually listed fragments, or 7.9%. Although both figures are low, the fact remains that these larger hole sizes are less than half as common in the Ford Bridge Assemblage than in the Puckeridge mould.

2 Tray forms There is strong evidence for one tray form in the assemblage, and reasonable circumstantial evidence for a second.

The ‘minimum number of trays’ formula is somewhat complicated by the strong possibility4 of the presence in this assemblage of two tray forms, the Verulamium form and the Puckeridge form. Using hole base diameter as the determining criterion for attributing tray form,5 it is possible to suggest that there are no fewer than 2118 holes which probably derive from Verulamium form trays, and 52 holes which probably derive from Puckeridge form trays.

That there are 42 fragments, or 10% of the total number of edge fragments, with the distinctive features of the pediment always associated with the Verulamium tray form suggests strongly that this form is the most common in the assemblage. The fact that no fragment exhibiting such features has holes with a base diameter larger than 14.5mm would seem to suggest that this form is to be associated with smaller diameter holes, as was noted of the Puckeridge Assemblage.

Dividing the first number by 50, and rounding up to the nearest whole number, we arrive at a minimum of 43 trays to accommodate the number of smaller holes in the assemblage. The minimum number of trays is then multiplied by the number of corners possessed by a Verulamium form tray (5) to yield the minimum number of corners implied by the number of smaller holes. In this assemblage, the minimum number of corners from Verulamium form trays would have been 215. Dividing the second number by 25 and rounding up to the nearest whole number, we arrive at a minimum number of three Puckeridge form trays to accommodate the number of larger holes in the assemblage. Multiplying this by the number of corners possessed by a Puckeridge form tray (4), we arrive at a ‘minimum corners’ figure of 12.

Explored in §2., below. Holes ≥15mm are treated as diagnostic of the quadrilateral, 25-hole, Puckeridge form. Holes 0 to 1

9

1 to 2

46

2 to 3

38

3 to 4

20

4 to 5

2

Total

115

Series1

9 to 10 10 to 11

11 to 12

12 to 13

13 to 14

14 to 15

15 to 16

16 to 17

17 to 18

18 to 19

19 to 20

20 to 21

Figure 5.10: Top diameter distribution.

8 Predictable relationship between hole base diameter and pellet module

Figure 5.9: Variability in relationship between top and base hole diameters in the Ford Bridge Assemblage

Although, as has been shown in Chapter 3 §9, it is highly unlikely that there could be a predictable relationship between base diameter and pellet module under the conditions that prevail in the Braughing/Puckeridge assemblages (and in all the coin mould examined using the same methodology as this study), it is nonetheless useful to confirm this by analysis of the variability in base diameter shown by a particular assemblage.

From Figure 5.ix above we can see that the total range of variation in the difference between top and base hole diameter is 5mm, more than twice the size of Reynolds’s proposed hole size groups. Furthermore, 60 of the fragments (more than 52% of the sample population) exhibit difference between top and base diameters either equal to or greater than this 3mm limit.

Furthermore, comparison of the results of this analysis with the results of similar analysis carried out on experimental trays on which the holes were made in wet clay using a single-pronged dibber should make it possible to conclude whether or not the same method of hole-making was used on both samples.

Given the range of variation in difference between top and base hole diameter, and since the difference between top and base diameter for the majority of individual fragments is at least as great as a Reynolds 3mm group, it would seem safe to conclude that it is impossible in the Ford Bridge Assemblage to infer top diameter from base diameter or base diameter from top diameter, and that in consequence there is no predictable relationship between the two, and hence no obvious link between top diameter and ‘coin module’

On the basis of the evidence, it would not seem unreasonable to conclude that there are two clear groups of hole diameter in the Ford Bridge Assemblage, that these groups are broadly similar to those observed in the Puckeridge Assemblage, and that they coincide with the two tray forms which seem to be present in both assemblages. Holes with a base diameter less than or equal to 13mm never appear on the same fragment as holes with a base diameter greater than or equal to 16mm – while there is a continuous spectrum of diameters on fragments with holes up to 13mm, there is a clear diastema

The chart below, showing the frequency with which various top diameters occur, provides evidence that top diameter is inherently more variable than base diameter. 15  Reynolds, J in Elsdon, S, 1997; §8.2, p. 56. The hole size groups are 4 – 6mm and 7 – 9mm.

51

Making a Mint at this point. Although, as noted in Section 2 viii, there is no lower limit on the size of pellet that could be cast in a hole of a given diameter, and only an upper limit, and that therefore there can be no certainty as to the size of pellet actually cast in that hole, it can be suggested with some force that, since this diastema coincides with the distribution of the two tray forms that probably comprise the assemblage, the distinction between the two hole diameter groups was both recognized by, and significant to, the makers (and probably the owners as well) of the mould trays in the Ford Bridge Assemblage.

Furthermore, given that the distribution pattern of this group forms a reasonable bell-curve, there is no reason to believe that it is anything other than homogenous. Table 5.7 below, in which base diameters are plotted against individual fragment ID numbers, demonstrates that the composition of the assemblage is homogenous throughout the various contexts in which the coin mould was found. 9 Hole Depth As demonstrated in Chapter 2, §9, very precise control of the volume of a hole is a necessary precondition of the credibility of two theories concerning the purpose and method of use of coin mould,16 that this precision requires in particular close control of the depth of a hole, Series1 and that the larger the diameter of the hole, the more stringently the depth must be controlled.

45 40 35 30 25 20 15 10

There are 453 fragments in the Ford Bridge assemblage with measurable depths, yielding a total of 1126 measurable holes. The shallowest hole is less than 4mm, the deepest more than 13mm. As set out in Chapter 3, §10, fragment average hole depth has been derived from individual hole measurements, and average intrafragment variability derived from individual fragment hole depth standard deviation, and it is these averages which form the basis of the discussion that follows.

5 0 7 to 8 to 9 to 10 to 11 to 12 to 13 to 14 to 15 to 16 to 17 to 8 9 10 11 12 13 14 15 16 17 18

Figure 5.11: Base diameter distribution by percentage.

Experimental work suggests that it is impossible to control the depth of a hole made with a single-pronged dibber to an average accuracy17 of better than ± 0.86mm, which equates, assuming an 11mm diameter pellet, to a weight in silver of approximately ± 0.3g, or very nearly 25% the weight of an Icenian silver unit, and a range of variation of 1.72mm from a target depth of 5mm. There are two conclusions which may be drawn from this: first, that variability of this order could not be acceptable in either coin production or the accurate measurement of metal for alloying; second, that mould fragments exhibiting variability significantly less than this could not have been made using the single-pronged dibber method.

Figure 5.12: Homogenous base diameter distribution. Context 00 forms the first cluster, context 03 the second, and the extended third cluster is formed from contexts 04; 06; 09 and VH.

Table 5.6 above, compiled from data derived from the 196 fragments with measurable base diameters, demonstrates very clearly the existence of the two base diameter groups, and also shows the very broad parameters of the smaller base diameter group, from 7mm to 13mm. While it might be possible to argue – but not to prove - that the larger base-diameter group was intended for the manufacture of a larger coin, such as a Tasciovanus double-unit bronze, it would be very difficult to infer the intention to manufacture any single denomination from a diameter group with a range of variation of 6mm.

The first theory is the Casey/Sellwood hypothesis that the mould was not part of a minting process, but was instead used as a means of ready-reckoning for the production of alloys; the second is the suggestion that metal was introduced into mould holes by pouring in the molten state. Many writers cite this theory only to dismiss it, but it has not proved possible to trace a single wholehearted proponent. Van Arsdell (1989: p. 45) comes close, contenting himself with pointing out the inherent difficulties of the method, rather than ruling it out altogether, while Frere (1983: p. 32), comes closer, declaring the pans evenly balanced. 17  See Appendix I, ‘Some experiments in the manufacture of coin mould’. 16 

52

Chapter 5 : The Ford Bridge (Braughing) assemblage

Range

Count

%