Later Roman Egypt: Society, Religion, Economy and Administration 0860788997, 9780860788997

Citation preview

Also in the Variorum Collected Studies Series ALFREDO MORDECHAI RABELLO The Jews in the Roman Empire: Legal Problems, from Herod to Justinian HAROLD LIVERMORE Essays on Iberian History and Literature, from the Roman Empire to the Renaissance DAVID JACOBY Byzantium, Latin Romania and the Mediterranean HENRY INNES MACADAM Geography, Urbanisation and Settlement Patterns in the Roman Near East ROGER FRENCH Ancients and Moderns in the Medical Sciences: From Hippocrates to Harvey DIMITRI GUTAS Greek Philosophers in the Arabic Tradition BERNARD S. BACHRACH Warfare and Military Organization in Pre-Crusade Europe CLIFFORD DAVIDSON History, Religion, and Violence: Cultural Contexts for Medieval and Renaissance English Drama PETER LINEHAN The Processes of Politics and the Rule of Law: Studies on the Iberian Kingdoms and Papal Rome in the Middle Ages W.H.C. FREND Orthodoxy, Paganism and Dissent in the Early Christian Centuries SIR LAURENCE KIRWAN, edited by T. HAGG, L. TOROK and D.A. WELSBY Studies on the History of Late Antique and Christian Nubia JEAN-MICHEL SPIESER Urban and Religious Spaces in Late Antiquity and Early Byzantium SEBASTIAN BROCK From Ephrem to Romanos: Interactions between Syriac and Greek in Late Antiquity

VARIORUM COLLECTED STUDIES SERIES

Later Roman Egypt: Society, Religion, Economy and Administration

Roger S. Bagnall

Roger S. Bagnall

Later Roman Egypt: Society, Religion, Economy and Administration

O Routledge

S^^ Taylor & Francis Group LONDON AND NEW YORK

First published 2003 by Ashgate Publishing Published 2016 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX 14 4RN 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business This edition copyright © 2003 by Roger S. Bagnall. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval s ystem, without permission in writing from the publishers. Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-Publication Data Bagnall, RogerS. Later Roman Egypt: Society, Religion, Economy and Administration. - (Variorum Collected Studies Series: CS758) 1. Egypt - Social Conditions. 2. Egypt - Economic Conditions 332 B.C.-640 A.D. 3. Egypt - Politics and Government - 30 B.C.-640 A.D. 4. Egypt - Religion. I. Title 932'.022 US Library of Congress Cataloging-in-Publication Data Bagnall, RogerS. Later Roman Egypt: Society, Religion, Economy and Administration / Roger S. Bagnall p. cm. -(VariorumCollected Studies Series: CS758) Includes bibliographical references and index. 1. Egypt - History - 30 B.C.- 640 A.D. I. Title. II. Collected Studies: CS758. DT93 .B35 2003 932-dc21

DOI: 10.4324/9781003556732 ISBN 13: 978-0-86078-899-7 (hbk)

VARIORUM COLLECTED STUDIES SERIES CS758

2002074537

CONTENTS Preface Acknowledgments

ix-x xi

SOCIETY I

Slavery and Society in Late Roman Egypt Law, Politics and Society in the Ancient Mediterranean World, ed. B. Halpern and D. Hobson. Sheffield, 1993

220-240

II

Women, Law, and Social Realities in Late Antiquity: A Review Article Bulletin of the American Society ofPapyrologists, 32. Atlanta, 1995

III

Missing Females in Roman Egypt Scripta Classica Israelica, 16. Jerusalem, 1997

IV

Church, State and Divorce in Late Roman Egypt in Florilegium Columbianum: Essays in Honor of Paul Oskar Kristeller. New York, 1987

41-61

V

Official and Private Violence in Roman Egypt Bulletin of the American Society ofPapyrologists, 26. Atlanta, 1989

201-216

VI

The Population of Theadelphia in the Fourth Century Bulletin de la Societe d'Archeologie Copte, 24. Le Caire, 1982

Vn

An Owner of Literary Papyri Classical Philology, 87. Chicago, 1992

65-86

121-138

35-57 137-140

vi

CONTENTS

RELIGION AND SOCIETY Vm

Religious Conversion and Onomastic Change in Early Byzantine Egypt Bulletin of the American Society ofPapyrologists, 19. Atlanta, 1982

DC

Conversion and Onomastics: A Reply Zeitschriftfur Papyrologie undEpigraphik, 69. Bonn, 1987

X

Combat ou vide: Christianisme et paganisme dans PEgypte romaine tardive Ktema, 13. Strasbourg, 1988 [1992]

XI

Charite's Christianity Bulletin of the American Society ofPapyrologists, 32. Atlanta, 199 5

105-124

243-250

285-296 37^0

ECONOMY XII

Landholding in Late Roman Egypt: The Distribution of Wealth Journal of Roman Studies, 82. London, 1992

XIII

An Arsinoite Metropolitan Landowning Family of the Fourth Century Papyrologica Lupiensia, 2. Lesce,1993 [Papiri Documentari Greci]

XTV

Military Officers as Landowners in Fourth Century Egypt Chiron, 22. Munchen, 1992

47-54

XV

Price in 'Sales on Delivery' Greek Roman and Byzantine Studies, 18. Durham, North Carolina, 1977

85-96

XVI

The Camel, the Wagon, and the Donkey in Later Roman Egypt Bulletin of the American Society ofPapyrologists, 22. Atlanta, 1985

128-149

97-101

1-6

CONTENTS

vii

ADMINISTRATION AND TAXATION XVII Agricultural Productivity and Taxation in Later Roman Egypt

289-308

XVm ROxy. XVI1905, SB V 7756, and Fourth Century Taxation

185-195

Transactions of the American Philologcial Association, 115. New York, 1985

Zeitschriftfur Papyrologie undEpigraphik, 37. Bonn, 1980

XIX

Bullion Purchases and Landholding in the Fourth Century Chronique d'Egypte, 52. Bruxelles, 1977

XX

The Taxes of Toka Tyche, 6. Wien, 1991

XXI

The Periodicity and Collection of the Chrysargyron Tyche, 7. Wien, 1992

XXII Count Ausonius Tyche, 7. men, 1992 XXIII Property-holdings of Liturgists in Fourth-century Karanis Bulletin of the American Society ofPapyrologists, 15. Atlanta, 1978

XXIV The Number and Term of the Dekaprotoi Aegyptus, 58. Milano, 1978

322-336 37-43 15-17 9-15

9-16

160-167

Addenda

1-6

Index

1-2

This volume contains xii + 318 pages

PUBLISHER'S NOTE The articles in this volume, as in all others in the Collected Studies Series, have not been given a new, continuous pagination. In order to avoid confusion, and to facilitate their use where these same studies have been referred to elsewhere, the original pagination has been maintained wherever possible. Each article has been given a Roman numeral in order of appearance, as listed in the Contents. This number is repeated on each page and quoted in the index entries.

PREFACE In this volume are collected two dozen of my articles, almost entirely devoted to the period from the third to the fifth century of our era, although a couple of items spanning a broader period are included. Most of them were written during the long gestation period of Egypt in Late Antiquity (Princeton 1993) and treat in more depth topics discussed in that book, but some (XV, XIX, XXIII and XXIV, most notably) are older and a few (II, III and XI) are more recent. I have excluded work on chronological matters, virtually all of it written jointly with Klaas Worp, because we are now engaged on a greatly expanded second edition of our Chronological Systems of Byzantine Egypt (Zutphen 1978), to appear we hope in 2003, in which that material will be regrouped and synthesized. I have left aside also those essays dealing with earlier historical periods and a number of articles still in press at present. Some of these may eventually form the basis for another volume of this sort. The material for the present volume has been grouped under four headings. The first includes studies of various social phenomena, ranging from violence to book ownership, but with a particular focus on the condition of women and the continuing importance - in my view, at least of slavery in late antiquity. The two articles in this section that postdate Egypt in Late Antiquity are the result of external stimuli: in the one case, the opportunity to review Joelle Beaucamp's very important work on women in Byzantium (II), in the other (III) my collaboration with Bruce Frier on The Demography of Roman Egypt (Cambridge 1994). The second section is entirely devoted to questions concerning the Christianization of Egypt. My approaches to this topic, in these articles and in the relevant chapter of Egypt in Late Antiquity, have given rise to a fair amount of controversy. A few references are provided in the Addenda for the reader who wants to pursue the matter further. I should confess that I have not found any reason to give up my convictions that the levers of power in Egyptian society were effectively in Christian hands and the majority of the population at least nominally Christian within a generation after Constantine. I plan elsewhere to come back to the methodological issues involved in this discussion.

x

PREFACE

The last two sections are on the whole devoted to matters of the Egyptian economy, both private and public, with the last section particularly concerned with taxation, tax collection officials, and compulsory public services. Many of these are, as has on occasion been noted, quantitative in their approach. In most cases this is a matter of simple arithmetic, although XII requires one to come to terms with the Gini coefficient (itself hardly a matter of higher mathematics). The addenda printed at the end of the volume refer the reader in some cases to discussions of these articles, particularly where controversy has arisen; in others to more recently published evidence or bibliography concerning the subjects treated. I have not for the most part added references to the sections of Egypt in Late Antiquity where these topics are discussed, as these will be easily found. The index to the volume omits topics readily discovered from the Table of Contents but attempts to pick up subjects that may not be so self-evident. A small number of entries is added for documents discussed more extensively than simple citation or quotation. The appearance of this volume coincides with the twenty-fifth anniversary of my first article written together with Klaas Worp. Readers might hardly guess the importance of that partnership on the basis of the present volume, because of the exclusion of collaborative work from it. But Klaas has over the years discussed with me many of the issues at stake in these papers and has seen and commented on many of the pieces gathered here before their publication. The dedication of the volume to him is a small acknowledgment of the central role that our common work has played in my scholarly life over the last quarter-century. New York October, 2002

ROGER BAGNALL

ACKNOWLEDGMENTS Grateful acknowledgment is given to the following for permission to reproduce the articles in this volume: Sheffield Academic Press (I); Bulletin of the American Society of Papyrologists (II, V, VIII, XI, XVI, XXIII); Scripta Classica Israelica (III); Italica Press, Inc. (IV); Societe d'Archeologie Copte (VI); The University of Chicago Press (VII); Zeitschrift fur Papyrologie und Epigraphik (IX, XVIII); KTEMA (X); Society for the Promotion of Roman Studies (XII); PapyrologicaLupiensia (XIII); Kommission fur Alte Geschichte und Epigraphik (XIV); Greek Roman and Byzantine Studies (XV); The Johns Hopkins University Press (XVII); Chronique d'Egypte (XIX); TYCHE (XX, XXI, XXII); Universita Cattolica del Sacro Cuore (XXIV).

I

SLAVERY AND SOCIETY IN LATE ROMAN EGYPT

1. Introduction Slavery in Egypt may seem a singularly unpromising choice of subject if one hopes to find anything both new and true. Not only do we have a well-balanced synthesis by Iza Biezunska-Malowist (1974-77), but a host of specialized studies of the material have brought us to the point that Ramsay MacMullen can write, in an article on late Roman slavery, 'Egypt would be almost too richly documented to be treated in a summary paragraph or two, were it not that the numbers and employment of the servile element in the population have been so carefully evaluated already' (MacMullen 1987: 364). Biezunska-Malowist's select bibliography for the Roman period alone includes over 160 items, and Jean Straus, writing four years after the 1977 publication of BiezunskaMalowist's Roman volume, added another 40 or so (Straus 1981). If one is to add to this pile yet another treatment of slavery, some excuse is needed. And one lies ready to hand. Biezunska-Malowist stopped with the late third century, though she used fourth-century documents occasionally in her treatment of individual questions. And there is no comprehensive treatment of slavery in the Byzantine period at all. On the other hand, MacMullen found no difficulty in discerning a scholarly consensus on the subject.1 Two examples from the longstanding consensus will suffice. In Johnson and West's standard work on the Byzantine economy of Egypt (Johnson and West 1949: 132-35) we are told that most slaves were household slaves:

1. Given the extensive bibliographic resources in the works of BiezunskaMalowist, Straus and MacMullen cited, I have kept my references to the secondary literature to a minimum.

DOI: 10.4324/9781003556732-1

I 221

Slavery and Society in Late Roman Egypt Egypt has never been a slave-owning country as that term is generally used or in the sense that Italy was in the first century. The dense population of Egypt has always provided an abundance of laborers for every need. Wages have been so low, owing to this superabundance of labor, that there has never been a need for slave labor or any additional profit to be obtained from its use.

E.R. Hardy, in his work on large estates, describes the situation in the Apionic documents of the sixth century (Hardy 1931: 112): Slaves do not seem to have been very common, although there were still some in great households. Curiously enough several of our references to them show them carrying on businesses of their own—a slave of Flavius Apion renting a house at Oxyrhynchus [PS/709], a slave of Cyril, tribune of Arsinoe, borrowing money from another member of the household [BGU 725], or a slave of the patrician Athanasius in the Thebaid holding a government position [P. Cair. Masp. 67166]. The Hermopolis estate owned slaves who received food and clothes, but the purposes for which they were used are not stated [P. Bad. 95].

In short, the main elements of MacMullen's summary of Egypt would probably find few dissenters—'slaves are perceptibly fewer'. 'As in earlier centuries, slaves are employed as domestic servants but are little or never attested in manufacture and agriculture. When we find large estates in the early or late fourth century with their own bakers or potters, those resident workers are freeborn' (MacMullen 1987: 365). Apart from Egypt, however, MacMullen argues that late antiquity saw less change in the numbers and employment of slaves than has usually been thought, largely on the grounds that they had never been so numerous or economically important in most areas as some scholars have claimed. Now there is a little inconsistency here. On the one hand, we find that slavery supposedly declined in Egypt in the third century and later. On the other, we find little change in the employment of slaves. For the empire as a whole, however, slavery is supposed to have remained fairly constant in both numbers and uses. Indeed, MacMullen goes so far as to suppose that 'the number of slaves may then have remained very nearly on a level, from Augustus to Alaric' (MacMullen 1987: 376). Now it is not impossible that the Egyptian trend was very different from that of the rest of the empire, of course; but we may still be a little uneasy that this should occur precisely when Egypt in so many ways seems to be in a process of becoming more like the rest of the empire, from the introduction of

I 222

councils into Alexandria and the metropoleis of the nomes by Septimius Severus, through Diocletian's unification of the currency and the administrative changes of the early Dominate which made the empire more uniform, to the increased imperial prominence of the Egyptian municipal aristocracy in the fourth and fifth centuries. There is a still more compelling reason for uneasiness with this synthesis. Numbers and functions of slaves are not totally independent variables. It is difficult to imagine that a decline of, say, a half in the supply of slaves would not radically alter their availability to various classes of society and the uses to which they would put them. Moreover, no reason for the supposed decline in numbers has been offered. It seems to me worthwhile, therefore, to ask on how secure a basis rests the current synthesis. To that I will add an attempt to look again at the structural place of slaves in Egyptian society in late antiquity: not merely what jobs they held, but what role they played in the total fabric of society and economy. This, surely, is what we would really like to know about this institution which occupies such a prominent place in ancient and modern works on Roman law and society. 2. The Numbers Game MacMullen's assertion that slaves were 'perceptibly fewer' from the third century on appears to be based on evidence, since he has just noted that the papyri sometimes preserve sufficient evidence for us to know just how many slaves there were in a particular time and place. What it is in fact based on 1 is an important article of 1973 (Fikhman 1973), in which the Soviet historian of late antiquity I.F. Fikhman drew attention to the radical decline in attestations of slaves in fourth-century and later documents from Oxyrhynchus. With characteristic carefulness, he showed that this drop in evidence is not the result of a smaller overall number of documents. For example, the Roman period up to AD 284 had yielded about 1700 total documents from Oxyrhynchus, of which 117, or about 6.9 per cent, mentioned slaves by name. For the period from the late third century through the seventh century, on the other hand, about 1130 documents name slaves only 60 times, about 1. Apart from a reference to an assertion of Westermann and a brief remark of Marie Drew-Bear (1984: 326), based on one Hermopolite papyrus in which a welloff councillor of that city left 'only' six slaves in his will.

I 223

Slavery and Society in Late Roman Egypt

5.3 per cent. The decline is actually steeper than that, however, for most of the late texts seem to refer to a handful of slaves holding important positions in the Apionic estates. For the period from the late third century through the fifth century, the figure is only 2.2 per cent, less than a third of that for the Roman period. Fikhman shows clearly that this is not some Oxyrhynchite peculiarity, and he points out that the virtual disappearance of manumissions and freedmen from the documentation confirms the picture already drawn. He concludes that the diminution in the number of attestations of slaves is not simply a consequence of deficiencies in our documentation, but a reflection of the real situation. A closer look at the instances of sales of slaves might seem to give strong confirmation to Fikhman's view. There are ten known from the first century, 16 from the second, 23 from the third and only five from the fourth. 1 And of the third-century sales, only three of 23 come from the last quarter. It would be easy to conclude that there were simply not very many slaves being sold and, with Fikhman, that there were simply fewer slaves around, and fewer still involved in production, than in earlier centuries. And yet we ought to be suspicious. One telling detail is that there are only three Ptolemaic slave sales in the Egyptian papyri, one of them from outside Egypt. Are we to suppose that the number of slaves in three centuries of Ptolemaic rule was only 6 per cent of that in three centuries of Roman rule? This can hardly be the case. Slavery was not uncommon in Ptolemaic Egypt, but it generated a documentation which rarely included contracts of sale. The coming of the Romans produces a dramatic rise in such contracts, a rise which peaks in the third century, then declines, so that there are no such contracts from the second half of the fourth century (apart from one concluded outside Egypt but preserved in it), and very few from later centuries.2 Here we have a reductio ad absurdum. This item of evidence supports Fikhman's conclusion so strongly as to throw it into doubt. We 1. See for references to published slave sales P. Koln IV 187, introd., supplemented most recently by the fourth-century sale P. Koln V 232 (the editor's proposed date of AD 314 is certainly wrong; good possibilities are AD 330, 332, and 337), with references to texts published more recently. 2. For the one contract from AD 350-400, see BGU I 316 (Askalon, AD 359). There is an extraordinarily verbose sixth-century example in Archiv 3 (AD 1906): 414-24; and for a late declaration of freedom, Wenger 1922.

I 224

cannot rationally maintain, faced with the Ptolemaic evidence, that preserved contracts of sale for slaves are a useful index of the presence of slaves in a society. Are we any better off with the other documentation? Our evidence, abundant though it is, has some insidious characteristics. Almost all of it, for one thing, comes from the viewpoint of the propertied classes of the metropoleis of the nomes, telling us little about village life. Most of the documents from the fourth century come from Oxyrhynchus or Hermopolis, fewer from Antinoopolis, Panopolis or Arsinoe, and very few indeed from other cities. These reflect the preoccupations of those who were literate in Greek or could pay scribes who were; transactions petty and large in money or commodities, leases, loans, deliveries, payments; legal complaints and records; the voluminous documentation of tax collection on behalf of the imperial authorities and the management of the finances and services of the towns; private letters. A papyrological voice may be heard objecting that this description ignores the archives of Arsinoite villages, principally Karanis and Theadelphia. That is only partly true. These archives are rich in documents emanating from the same apparatus and processes at work in the cities, by which the villages sent their taxes in to the cities, by which urban residents lent money or grain to villagers at high interest, and by which villagers tried to move the machinery of the cities to do something on their behalf. In other words, village archives are largely the result of demands and relationships originating from the cities. How much would be left of the archive of Aurelius Isidores if his liturgical work, tax payments, leases and loans were removed? Where in this highly archival1 body of documentation do we expect slaves to turn up? In point of fact, they appear mainly in documents concerned with inheritances, in petitions and court proceedings, and in letters, apart of course from slave sales and manumissions. Slaves did not generally borrow money, lease land, submit petitions, have responsibility for tax collection, pay taxes, buy and sell land, hold official positions, and so forth. It is, of course, possible that some— even many—of the people who appear in various occupations are 1. See Bagnall and Worp 1980, 1982 for the character of the documentation of this period. A very high percentage of our Arsinoite texts, for example, come from two large archives plus the Karanis ostraka. The Oxyrhynchite material is very heavily drawn from certain types of official documents. The Hermopolite sources are dominated by a handful of related family archives.

I 225

Slavery and Society in Late Roman Egypt

slaves without our being able to detect the fact. If a landowner writes to his phrontistes, his steward, ordering him to deliver ten jars of wine, we do not know the legal status of the steward. It is not irrelevant to the nature of the interchange, perhaps, but it need not be mentioned in an informal note from one to the other. Slaves, in general, tended not to produce documentation except when they were purchased, sold, manumitted, bequeathed, or reported in a census; when they paid taxes or performed compulsory duties, or when they got into some sort of trouble. In a time when census declarations did not include slaves, one born into a family who lived with it throughout life would not generate much paperwork. To return to Fikhman's numbers, if we subtract slave sales (whose usefulness for this purpose I have challenged above) from the figures, 4.3 per cent of the pre-Diocletianic Oxyrhynchite papyri mention slaves while only 1.6 per cent of those from Diocletian through the fifth century do. The Roman papyri include seven nursing contracts and related texts (a type virtually extinct after Diocletian),1 four census declarations (practically absent after the third century), ten manumissions and related texts (only three known after AD 2842), nine wills (very scarce after AD 3003), six epikrisis documents, three death notices (not found after AD 300 [see Brashear 1977: 3-5]), three marriage documents (a genre lacking between Roman times and the sixth century4). In all, about 62 per cent of the documents Fikhman lists for the Roman period represent genres which are scarce after Diocletian. Subtracting these and the sales, the remaining quarter amount to only 2.6 per cent of the Oxyrhynchite papyri; the comparable figure for the period after AD 284 is 1.9 per cent. When we remember that the latter figure covers a total of twelve documents, the difference may no longer impress us. Another approach casts still more doubt on the enterprise. Sales of land from the Roman period totalled 115 when Montevecchi compiled her lists a half-century ago, those from after AD 284, only 35 1. One instance after AD 258, out of 40 known: M. Manca Masciadri and O. Montevecchi, / contratti di baliatico (C.Pap.Gr. 1) (Milan 1984). 2. For example, P. Oxy. IX 1205 (AD 291), P. Mich. VH462 (fourth cent.), P. Edmondstone (see BASP 15 [1978]: 235-36; 355). 3. See Bagnall 1986: 1-2: only 16 of 131 known wills are after AD 300. 4. Except for P. Ross. Georg. Ill 28, a remarkable piece which I cannot discuss here.

I 226

(Montevecchi 1943: 12-19). Sales of building property were 105 for the Roman period, 34 later (Montevecchi 1941: 94-98). If these figures are computed as percentages of the total Oxyrhynchite documents as enumerated by Fikhman, sales of land are 6.8 per cent of the Roman total, those of buildings 6.2 per cent. For the Byzantine period, on the other hand, the comparable figures are 3.1 and 3.0 per cent. Now, more sales have been published since, and the figure is based on all provenances, not just Oxyrhynchus. But I do not think that matters for our purposes. What is clear is that documents recording sales of all sorts decline as a proportionate part of the documentation after AD 284. This investigation is a bit crude and could certainly be refined. But I think it suffices to show that the numerous and profound changes in the nature of the surviving documentation for the second half of the third century are sufficient to account for the entirety of the decline of our documentation of slaves in the papyri. To what extent those changes in documentation reflect simply the archives we happen to have for the fourth century, or rather embody more profound alterations in the whole structure by which social relations and status were recorded, I cannot go into here. At all events, I believe the numbers game must be abandoned for this subject. There is no reliable basis for any claims about the relative abundance of slaves in Roman and Byzantine Egypt. We have little information about how people became slaves in this period. A few were acquired abroad;1 most were no doubt born slaves, and there are explicit attestations of such origin.2 A slave of a military officer describes himself as threptos, exposed and rescued, brought up by the owner (P. Abinn. 36). It was also possible to fall into slavery through debt. For example, one letter complains that the addressee has taken the writer's money in order to get him released but has failed to do so. 'Now then [he says], do not neglect this, master, for God's sake; for you have already given my children as securities to the money-lender on account of the gold' (P. Herm. 7). In another 1. BGU 1316 (Askalon, AD 359) is the best known; but the power of attorney in P. Oxy. XXXVI 2771 (AD 323), whereby a husband in Cos is authorized to sell a slave on behalf of his wife, presumably travelled with the slave to Oxyrhynchos as part of the title papers. 2. For example P. Lips. 26 specifies that two of the four slaves are house-born; the origin of the other two is not specified.

I 227

Slavery and Society in Late Roman Egypt

letter, the writer recounts (to his wife) his seizure in Alexandria by a creditor; since he has nothing to pay him with, he asks her to put their little paidion Artemidoros in hypothec and send him the proceeds (P. Amh. II 144 [fifth century, no provenance]). One supposes that the luckless Artemidoros is already a slave and not their child, but the seizure of the writer himself suggests that the insolvent could wind up in indefinite bondage.1 Roman society, as we know, contained plenty of people whose exact status may have been unclear to those around them, as well as those who moved from free to slave and slave to free. We have two graphic illustrations of the difficulties which could ensue. A papyrus from Hermopolis reports proceedings about the status of one Patricius, evidently because his owner had sold him and the purchaser wanted to register the sale. The slave himself is asked if he is free or slave; he answers that he is a slave. He then answers questions about where he was bought, from whom, who his mother was, what her status was (slave), whether he has siblings (one, a slave); and on the basis of all this the registrars agree to register the sale, but at the purchaser's risk if the facts turn out to be wrong (P. Herm. 18 [323p?]). The reverse case is found in a petition to the prefect, in which the complainant alleges that his wife and children were carried off by another couple (one member of which seems to be a prytanis), on the grounds that they were their slaves. The petitioner claims that his wife is of free descent and married to him, and that her brothers are free; he claims that her parents were also free. The case was probably not quite so clear, we may think, but it is striking just how much one might have to prove simply in order not to be enslaved.2 3. Rural Slavery In saying that 'in the second and third centuries, the work of slaves did not play an important role in Egyptian agriculture', BiezunskaMalowist was not denying its existence, for she goes on to say that 'it is essential to point out its existence, even if only in a minimal amount, in the Roman period' (Biezunska-Malowist 1977: 83). She proceeded to enumerate a number of items of evidence, concluding 1. Another example of creditors seizing children occurs in P. Lond. VI1915 (c. AD 330-340). 2. P. Grenf. II 78 (Great Oasis, 307, cf. Bagnall and Worp 1979: 31).

I 228

that it is impossible to evaluate their significance. This seems to me unnecessarily pessimistic, even if statistically true. Let us look at what there is. An Oxyrhynchus papyrus (P. Oxy. XIV 1638 [282]) preserves a division of the inheritance from one Psenamounis, who lived in a village in the western toparchy of the Oxyrhynchite nome, among the members of his two families by different wives. The estate was divided among seven people; the older family of two got a house, 3i arouras of grainland, and 7 share of four slaves; the remaining five got the rest of the property and a ? share of the slaves. Of the slaves, one is apparently male, one undescribed, one an adult female aged 25 and one a child, her daughter aged ten. There is no sign of any actual assignment of slaves, merely division of ownership of the entire group. A family of moderate means thus owned four slaves. Unfortunately, the total amount of land owned by Psenamounis is not mentioned, but the older family's allotment of 3i arouras suggests a very modest total estate, perhaps no more than ten or fifteen arouras. From the well-known archive of Aurelius Isidoros from Karanis, there is evidence that the wife of Aurelius Isidoros's brother Heras, one Taesis, owned slaves. Her paternal inheritance, shared with her sister Kyrillous, included 61 sheep, 40 goats, one grinding mill, six talents of silver, two artabas of wheat, and two slaves (P. Cair. Isid. 64 [c. AD 298]). It also included land. Each sister seems to have inherited about ten or eleven arouras of grainbearing land, for a paternal total of about 22 arouras; this is a very middle-range holding for a villager of moderate means.1 About the slave holdings of the central figure of the archive, Aurelius Isidoros himself, we know nothing; since his and his siblings' landholdings far exceeded those of Taesis and Kyrillous, it seems hard to suppose that they did not include slaves. The one census declaration by Isidoros preserved (P. Cair. Isid. 8 [AD 309]), states that the household consists only of himself and his son. Similarly, in the census declaration of Sakaon, the central figure of the Theadelphian archives of this period, there are eight free persons and no slaves mentioned (P. Sakaon I [AD 310]). Sakaon was moderately prosperous by the standards of his village, but the village was in very rocky shape. Unfortunately, we do not know whether 1. The evidence for landholdings is found in or deduced from P. Cair. Isid. 6 and 9, using the methods of analysis from Bagnall 1977: 330-31 n. 1. For the landholdings of village liturgists, who are the more prosperous part of village society, see Bagnall 1978: 16; 22 arouras is a very decent holding.

I 229

Slavery and Society in Late Roman Egypt

slaves were required to be listed. Sakaon's return contains eight men and no women. Was this really such a bachelor establishment, even though four of the eight men were over twenty? It seems very unlikely. It appears, in other words, as if these declarations of persons simply include only free males. It is hard to test this generalization; these are the only two declarations of persons known for the entire period from AD 275-450. Yet another small household with a few slaves is attested in a Hermopolite document (P. Lips. 26) of the early fourth century, in which two brothers divide up four slaves in their inheritance. Two are farmers, one a weaver and one a donkey driver. It seems likely that these slaves were part of the rural estate of the two brothers. Another Isidores papyrus (P. Cair. hid. 141) mentions the role of the female slave of another villager in carrying out a criminal raid. In sum, the character of the evidence for slaves belonging to village families of moderate means suggests that ownership of a small number of slaves—one to four—was not remarkable. The economic importance of slavery in such households was perhaps not so marginal as has been thought by almost everyone. On a small family farm, say 10 to 25 arouras, the presence of even one able-bodied adult agricultural worker alongside the family is not a trivial advantage. A farmer without a number of sons or other suitable male relatives would find it hard to work his farm himself without help. The effects of such rural slaves, therefore, though made up of very small units, may have been pervasive and decisive in the structure of the working of the land. That is not to say it was universal. The poorer families, with barely enough land to support themselves, would not have been able to afford a slave. The average slave represented a capital of perhaps 20 to 30 artabas of wheat; to own two such might be likened to having a year's income in human capital. And those metropolitan, absentee landowners who had small amounts of land or had it widely scattered normally leased it out to villagers; slaves probably had no role in the working of this land, either.1 We are, then, far from any statistical appreciation of the phenomenon—we do not even know what portion of independent 1. See the distribution of ownership in the Hermopolite land registers, as analyzed by Bowman 1985, with table on 158-59. 62% of the landowners listed (in Hermopolis and Antinoopolis combined) held fewer than 20 arouras in the countryside.

I 230

landowners had more than ten arouras, after all—but in qualitative terms the importance to this class of increasing by perhaps 50 or 100 per cent the available labor force was significant. Since this is the class which provided the villages with their rotating public officials of all kinds, it needed some means of freeing up time for these uncompensated duties. That precisely these men owned slaves can hardly be surprising. In terms of the economy and structure of the villages, then, the role of even a small number of slaves may well have been critical.1 There is still the question of slavery on larger estates to be considered. Large estates in the fourth century are of all will-o'-the-wisps the most elusive. As Bowman's study of the Hermopolite land registers has shown, the larger the landholdings of an individual, the greater the likelihood that they are spread over the nome rather than consolidated (Bowman 1985: 154-55). Great landowners certainly had a substantial number of staff, with stewards in charge, and there is abundant evidence for a constant flow of orders, information and goods between estates and the master or mistress in the city. But we normally know only the function, not the legal status, of the employees involved in these interchanges.2 The one item of useful evidence for slavery in the management of large agglomerations is the long account of income and expenditures in wheat and various other commodities coming from an estate centered on Hermonthis and dated to the first four months of AD 338 (P. Lips. 97).3 Payments of wheat for salary, or at least subsistence, are made on a monthly or bimonthly basis to various groups of workers, whose numbers vary from month to month. Prominent are organitai, opera1. It is true that some tasks in the agricultural year fell at times when official duties were probably light; but at harvest time the pressure may have been considerable. 2. There are occasional indications, apart from P. Lips. 97, to be discussed below, of rations paid to slaves; see P. Haitn. Ill 68 (AD 402) and BGU I 34; cf. P. Charlie 36 for a discussion of date and purpose, along with re-edition of a small part of it. SPP XX 106, an account of payments by or farpaidaria for contributions to the vestis militaris for the 14th indiction, seems to agree with this view also; it lists paidaria who are butcher, brickmaker and pastry-cook. The date seems likely to be AD 355-56 or later. 3. P. Lond. I 125 (p. 192), which is to be dated to July, AD 336, seems to pertain to the same places and probably the same entity. It is interesting that the Leipzig roll has an extensive text of Psalms 30-55 on the back, while the London one has a magical text.

I 231

Slavery and Society in Late Roman Egypt

tors of irrigation machinery, of whom 21 are paid in Pharmouthi; staff for animal care (cowherds, shepherds, donkey rearers), who number 18; transport workers (wagoneers, donkey drivers, camel drivers), totalling 11 at the peak; and gardeners, 14. But there are also some more undifferentiated groups: 20 ergatai, laborers; four boethoi, assistants; and 15 paidaria, slaves. Occasionally some specific information is given about one of these: one ergates is identified as a hypoboukolos, sub-cowherd, and one slave is a breadmaker, another a weaver (tarsikarios). Most have no particular occupational identity. The identification of the paidaria as slaves seems to me certain. Not only does the term always seem to mean this in the papyri of this period, but none of the people in question, with one exception, appears anywhere else in the long account in another capacity. Moreover, that one exception, with the nice slave name of Philokyrios, master-loving,1 appears in another place with his name adjacent to that of a doule, a female slave. It would be easier to interpret this information if we knew to whom belonged the substantial properties connected with this account. The amounts are not small: wheat expenditures total almost 2200 artabas in the four months; annually that would come to 6600 artabas, the income perhaps on 1500-2000 arouras, in line with the largest holdings found in the Hermopolite registers. The editor, Mitteis, argued that the estates could not belong to an individual; he preferred to think of them as held by a temple. For reasons I cannot go into in detail here, this view seems to me impossible, and Mitteis's arguments against individual ownership are devoid of substance. 2 If this is correct, what 1. Hardly found otherwise in the papyri, cf. NB and Onomasticon. Solin 1982: II, 752, lists a fair number at Rome, none of them demonstrably free-born. 2. Mitteis pointed out that the addressees of the account were apo epitropon, which he thought lent an official cast to things. But the term means that they are retired, and the Hermopolite registers are sprinkled with such terms, which simply identify the persons by their status. Nor does the appearance of a couple of priests, hiereis, among the people with whom business is transacted (one of them as agent for the owners, it seems) mean much. Priests owned or leased land and engaged in normal economic transactions like other people. The term ousia (v. 12) does not help much, either, as it may have a completely nontechnical meaning here. The term phoros can mean either rent or taxes and is thus unhelpful. An official government account would not be organized in this fashion, and there is no evidence for temples having large properties of this sort in the fourth century. In fact, it is well known that temple estates were confiscated by Augustus three and a half centuries earlier.

I 232

conclusions can be drawn about slavery in this context? The landholdings in question are distributed over several villages, and indeed in this pre-harvest season we have little evidence for the full extent of the property. Rather than think of an 'estate', we should perhaps think of the 'great house' which owned numerous properties and employed these people. But the conditions under which it employed them are obscure. The one obvious conclusion is that slaves coexisted with free labor in the pool of undifferentiated staff; those perhaps who filled multiple roles or who belonged to specialties whose members were not numerous enough to warrant their own heading in the account. The main groups of livestock and agricultural workers, however, were not slaves but free. Slaves make up about 15 per cent of the payroll in this particular establishment. I must emphasize that we cannot even be sure that these slaves were located in the countryside. The account could well have been drawn up at the urban headquarters of the great house, not at any of the rural locations mentioned, and the slaves could mainly have been located in the same place. The predominantly rural character of the free employees may speak against this possibility, but it does not quite exclude it. In sum, rather than think about large estates and the role of slave labor on them, we will do better to think about great houses and their integrated operations, city and country. We are in no position to discern the pattern of these in general terms, but so far as slavery goes, we can see a significant slave component to the staff, but not any demonstrable role in agriculture or livestock raising themselves. 4. Slaves in Urban Households Fikhman is undoubtedly right that little of the evidence for urban slaves shows any trace of their employment in craft production. The tarsikarios in the Hermonthis account, if he is in the city, is an exception. We do have one instance in which slaves—three Phoenician women—left behind for safekeeping by their owner with a brother were put out to work with an innkeeper until they could be taken upriver to Panopolis.1 But they were doing it in Alexandria, after all, Overall, the accounts impress one as highly similar to other accounts of wealthy landowners in Roman Egypt, not to official accounting. 1. P. XV Congr. 22 (Panopolis, earlier part of fourth century), a part of the archives of Ammon. Cf. also SB XIV 11929 for Harpocration's slaves.

I 233

Slavery and Society in Late Roman Egypt

not Panopolis. On the occupational front, it may also be worth mentioning that we have a bit of evidence for freedmen's work in a Karanis papyrus of the early fourth century, in which a freedman agrees to work in a workshop belonging to another person for a period of five months at a specified salary, part of which he receives in advance. We do not, unfortunately, know what kind of workshop.1 What we would call economic activities, however, play a small role in our documentation. Instead, the slaves we find are almost all household slaves or personal assistants for their master's business dealings. It is therefore hardly surprising that most of our evidence for urban slaves is connected with owners belonging to the upper stratum of society: members of the bouleutic class, the military, and the imperial civil service. The use of slaves in households for domestic purposes, with their number in proportion to the grandeur of the household, is so universal a fact of life in the Roman world as to need no comment.2 The slave as confidential personal agent of the master in business dealings is also a common phenomenon;3 it has been observed that the master's complete control over the slave and ability to torture him made slaves far more suitable than free persons for such positions. Even though production was hardly affected at all by slave labor, then, the importance of slave assistance for the ability of a small elite to manage business, civic, and military affairs should not be underrated.4 The urban elite was as dependent as the village elite on the use of slaves as support for their own activities. It is, however, at the social level that the effects of slavery are more determinative for the character of life. First, we do not get a very 1. P. Mich. IX 574. The salary is specified as 500 per day, with the unit being lost. But the advance payment total is given in myriads, which must be denarii. The editor has argued that they are drachmas, on the completely specious grounds that *a daily wage of 500 drachmas is a reasonable figure*. Reasonable when? A monthly wage of 10 talents would be perfectly 'reasonable' around AD 325, when we find 8 talents per month paid to a boethos in SB XIV 11592. 2. Some examples from the papyri: P. Ross. Georg. Ill 9 = Naldini 77 (house servants in Memphis receiving goods); P. Stras. IV 296 (Hermopolis); P. Oxy. VI 903 (both members of a couple own slaves in the house). 3. Some examples from the papyri: P. Oxy. XLIII 3146 (AD 347); pais is surely 'slave' here; a business transaction; P. Oxy. XLIX 3480 (c. 360-390; the staff of a tax collector); P. Abinn. 36 (general agent of a. praepositus). 4. Even so minor a testimony as the slave of a praepositus in O. Bodl. 2152 helps show such usage.

I 234

favorable impression of master-slave relations. Slaves ran away when they could,1 even those working for important people in what were probably responsible posts, like Magnus, a slave of an officialis in the office of the prefect of Egypt, who fled to Hermopolis. His owner gave another officialis an authorization to arrest Magnus and bring him back.2 An Oxyrhynchite, who had won many prizes and high rank as an athlete and had consequently been awarded Athenian citizenship, had a slave run away to Alexandria; he authorizes his representative to imprison the slave, accuse him, beat him, pursue those harboring him, and bring him back (P. Oxy. XIV 1643, 298). The correspondence of a nome strategos in the Panopolis rolls in the Chester Beatty collection includes an acknowledgment of instructions from the magisier rei privatae to send up to him an absconding slave (P. Panop. Beatty 1.149 [Panopolis, 298]). A letter from a slave of Abinnaeus to his master remarks, 'I am again your slave and don't secede from you as I did before' (P. Abinn. 36). If slaves did not run away, they tried to get their freedom. We have few manumissions and few freedmen attested, as I have noted already, but the nature of the human relationships involved is made clear by an Oxyrhynchus papyrus from the early fourth century. A female petitioner alleges that some slaves came to her and to her brother Eustochios by inheritance from their parents, and that they owned them jointly and equally. Without effecting any written division of ownership, they divided them in practice (evidently two for each).3 The two slaves of her brother have persuaded him to manumit them, along with their child, but without the petitioner's consent. She objects and asks that the slaves be prevented from escaping the bonds of slavery.4 Legally speaking she was probably correct, but it is obvious that her brother behaved as if there had been a division of the property, leaving him free to do as he pleased with his slaves. Another Oxyrhynchite petition concerns a jointly owned slave (the petitioner's sister is the other owner) who, since the death of their parents, has 1. See for the general problem Daube 1952. 2. P. Oxy. XII 1423 (mid to late fourth cent., according to the editor). 3. Where there was only one slave in an estate, presumably some sort of timesharing arrangement was necessary; but our evidence mostly concerns families with multiple slaves, where a de facto division seems to have occurred. 4. PSI V 452. Biezunska-Malowist 1977: 125 n. 78, thinks that there were originally four slaves, two of whom had a child after the division.

I 235

Slavery and Society in Late Roman Egypt

paid them an apophora\ but now he refuses to pay it. It appears, despite the petitioner's terminology, that the slave had been manumitted by the father in his will, and that apophora is his due to them as patrons of a freedman. The owner's attitude—and perhaps the reality of being a freedman—is made clear enough by her request that the former slave be compelled to pay the apophora and remain in their service (P. Oxy. Hels. 26 [AD 296]). Those who remained in slavery and did not manage to run away or be manumitted seem often not to have behaved as their owners would like. The Roman law of slavery takes cognizance of theft by slaves, regarding the tendency as a vice which could be warranted against, and we find confirmation in a complaint by a member of the council of Hermopolis that his slave has been 'kidnapped'—corrupted, it sounds like—by another person. Last night, we are told, the slave opened the door of the house while he was asleep and stole some of his goods. He caught the slave with the goods in the house of the other man. He asks to recover his property; curiously enough, he does not ask specifically for any punishment for the corrupter (P. Stras. IV 296 [AD 326]). In an affidavit of a wife against her husband, the ultimate insult on his part in household matters is that he trusts his slaves with his keys but will not let his wife have them (P. Oxy. VI 903). This affidavit, moreover, gives a dismal picture of the state of the whole household. 'He shut up his own slaves and mine with my foster daughters and his superintendent and son for seven whole days in his cellars, having insulted his slaves and my slave Zoe and half killed them with blows, and he applied fire to my foster-daughters, having stripped them quite naked, which is contrary to the laws.' Later on, he ordered his wife to send her slave Anilla away. Slaves, in short, could not be trusted. Moreover, slaves talked back to their owners, and to other free persons. The writer of a rather obscurely written Christian letter complains to the recipient that he has been unable to bear the constant hybreis, outrageous remarks, of the addressee's oiketes Agathos.1 A petition preserves the complaint of one Sarapion about the female slave of Melas the silversmith; the slave had come to his house the day before and inflicted hybreis on his wife and virgin daughter in a way contrary to the laws and to their station in life. He requests that she be 1. P. Select. 18 = Naldini 81 (provenance unknown; Bingen, Cd'E 41 [1966] 191 suggests the Hermopolite; IV century).

I 236

punished for what she shouted at them.1 Ulpian tells us that the praetor's edict expressly provides the grounds for these complaints: 'One who is said to have loudly shouted at someone contrary to sound morals (adversus bonos mores) or one through whose efforts such shouting is effected contrary to sound morals, against him I will give an action'. 2 That particular remark has to do with the shouting of crowds, but individual acts are also punished if said for the sake of insulting someone. The jurists quoted in the Digest make it quite clear that the relative status of the parties involved was part of what defined how contumelious such remarks were, and the petitioners I have mentioned make it clear that their station in life makes such insults particularly reprehensible. That a slave's insults of a free person were punishable hardly needed to be said, indeed; the jurists were more concerned to deal with the less obvious cases, as where a slave insults another slave—and thus is taken to have insulted the other slave's master (D. 47.18.1). With this matter we have passed from relations of slaves with their own masters to those of slaves with other free persons. It appears in general that though slaves provided much of life's comforts to the slave-owning class, they were also the cause of much of its trouble and even its danger.3 Apart from insults and theft from their masters, slaves appear to have been sources of violence for others. This is true most obviously when they beat someone on their master's behalf, as someone complains happened to him at the hands of a tax collector and his slaves (P. Oxy. XLIX 3480). But we also find complaints such as that to the police magistrates at Oxyrhynchus by a man whose wife was attacked in their house during the evening hours by one Tapesis and the latter's slave Victoria; it seems (the text is fragmentary) that a theft of gold was also involved. He asks that a midwife be sent to check his wife, who was thus evidently pregnant, and that the culprits provide guarantees in the event anything happens to her (P. Oxy. LI 3620 [AD 326]). The prize document of this sort, however, is P. Lips. 40, a 1. P. Lond. Ill 983 (p. 229; no provenance or date). The discussion makes it clear that hybreis here (and elsewhere) means verbal, rather than physical, attacks. 2. D. 47.15.2, from ad edictum 77. The entirety of Book 47 of the Digest is devoted to contumelious action. 3. Conversely, of course, one might use slaves to protect oneself from such dangers.

I 237

Slavery and Society in Late Roman Egypt

Hermopolite report of proceedings before the governor of the Thebaid from the late fourth century or early fifth century. The accuser, one Philammon, reports that his son had been attacked, beaten to within an inch of his life, and robbed of a substantial sum in public funds which he was carrying (10-12 gold solidi), by a party of four slaves. Philammon says that one of them held the victim's hands, one knocked him to the ground, and the other two beat and robbed him. Only two slaves are in court, Acholius and a young boy. Acholius claims that the supposed victim actually attacked them; that there were only two of them until a third joined them after the incident was over. Acholius is tortured but sticks to his story. His master, Sergius, was and is out of town. A neighbor who came out, with his own slaves, when he heard the noise of the scuffle, says that there were two or three of them, thus incidentally supporting the story of the defendants more than that of the plaintiff, though he seems to imply that the slaves were the aggressors. Since the neighbor happens to be the highest-ranking official in town, the logistes, his testimony carries some weight. All of these accounts convey an impression of town life as involving for the upper class significant risks, not merely of verbal abuse but also of theft by their own slaves who were suborned by others, of assaults by free persons assisted by slaves, and of physical violence in the streets at the hands of unsupervised slaves. Like all crime reports, this one no doubt gives such incidents more prominence than they occupy in normal daily life, but their contribution to the tone of urban existence is nonetheless unmistakable. The society of later Roman Egypt was hardly unique, in antiquity or otherwise, in having a high concentration of wealth and a large degree of dependency of part of the population upon a rich elite. It has often been claimed that this concentration was growing, but so far the evidence stubbornly resists all attempts to make it show that it was (cf. Bowman 1985: 155). It was not only the very top elite in society which relied on a social structure in which some were relatively free and others relatively not. Plenty of farmers of modest means had a few slaves, as we have seen. The wealthy of the towns had more slaves and perhaps depended on them less for their economic survival. In fact, they had dependent tenant farmers of ostensibly free status to work their lands and provide them with their income in rent. Slavery is only one aspect of this pervasive set of relationships of power. If it

I 238

was declining in importance in the face of other such relationships in the fourth century—and I am not yet persuaded that this was the case—it had lost nothing of its character, either as a part of the rural and urban economy, or as a necessity for the public lives of the elites, or as a kind of human relationship.1 MacMullen denies that the Roman world was a 'slave-owning society', on the grounds that slaves played a very small role in agricultural production. But that standard is excessively narrow in focus. Having slaves set the tone of existence; and the human face of slavery in the fourth-century papyri is on the whole an ugly one.2

1. This paper has perforce looked at slavery from the point of view of the slave in society—an owner's point of view, for the most part. Bradley 1984 offers an attempt to look at the mechanisms of control from the slave's point of view. 2. An earlier version of this paper was presented to the Israel Society for the Promotion of Classical Studies, meeting at Bar-Han University in May, 1986.1 am grateful to Ranon Katzoff for the invitation to speak on that occasion.

I 239

BIBLIOGRAPHY Bagnall, R. S. 1977

'Bullion Purchases and Landholding in the Fourth Century', Cd'E 52: 322-36. 1978 'Property-holdings of Liturgists in Fourth-Century Karanis', BASF 15: 9-16. 1986 Two Byzantine Legal Papyri in a Private Collection', in Studies in Roman Law in Memory of A. Arthur Schiller (CSCT, 13; Leiden: Brill): 1-9. Bagnall, R.S., and K.A. Worp 1980 'Papyrus Documentation in Egypt from Constantine to Justinian', in R. Pintaudi (ed.), Miscellanea Papyrologica (Pap.Flor.,7; Florence: Gonnelli): 105-16. 1982 'Papyrus Documentation in the Period of Diocletian and Constantine', BES 4: 25-33. Biezunska-Malowist, I. 1974-77 L'esclavage dans I'Egypte greco-romaine, (2 vols.; Wroclaw: Polska Akademia Nauk). Bowman, A. K. 1985 'Landholding in the Hermopolite Nome in the Fourth Century AD', JRS 75: 137-63. Bradley, K.R. 1984 Slaves and Masters in the Roman Empire (Collection Latomus, 185; Brussels: Latomus). Brashear, W. 1977 'P.Sorb.lnv. 2358 and the New Statistics on Death Certificates', BASF 14: 1-10. Daube, D. 1952 'Slave-Catching', Juridical Review 64: 12-28. Drew-Bear, M. 1984 'Les conseillers municipaux des me*tropoles au Hie siecle apres J.-C.', Cd'E 59: 315-32. Fikhman, I.F. 1973 'Sklaven und Sklavenarbeit im SpatrOmischen Oxyrhynchus', Jahrbuch fiir Wirtschaftsgeschichte II; 149-206. Hardy, E.R. 1931 The Large Estates of Byzantine Egypt (New York: Columbia University Press). Johnson, A.C., and L.C. West 1949 Byzantine Egypt: Economic Studies (Princeton University Studies in Papyrology, 6; Princeton: Princeton University Press). MacMullen, R. 1987 'Late Roman Slavery', Historia 36: 359-82. Montevecchi, O. 1941 'Ricerche di sociologia nei document! dell'Egitto greco-romano III: I contratti di compravendita', Aeg 21: 93-151.

I 240 1943 Solin, H. 1982 Straus, J.A. 1981 Wenger, L. 1922

'Ricerche di sociologia nei document! dell'Egitto greco-romano III: I contratti di compravendita', Aeg 23: 11-89. Die Griechische Eigennamen in Rom: Ein Namenbuch (3 vols.; Berlin: de Gruyter). 'Remarques sur 1'esclavage dans 1'Egypte romaine', Anagennesis 1: 125-28. 'Ein christliches Freiheitszeugnis in den a'gyptischen Papyri', in Beitrage zur Geschichte des christlichen Altertums und der Byzantinischen Literatur: Festgabe Albert Ehrhard (Bonn: Kurt Schroeder Verlag): 451-78.

II Women, Law, and Social Realities in Late Antiquity: A Review Article Joelle Beaucamp, Le statut de la femme a Byzance (4e-7e siecle). Travaux et Memoires du Centre de Recherche d'Histoire et Civilisation de Byzance, College de France, Monographies. I. Le droit imperial', Paris 1990 (Monographies 5). 1 + 374 pp. II. Les pratiques sociales, Paris 1992 (Monographies 6). xxxii + 493 pp. The sheer bulk of this work, some 950 pages, will probably make of it a book more consulted than read. Its origin in a thesis, moreover, is visible throughout in the detailed level at which every question is pursued, regardless of the originality of the resulting conclusions. This, too, may daunt readers. That would be regrettable, because the work amply repays the labor of a careful reading. For one thing, this is a major contribution to a vital subject, the condition of women in late antiquity. For another, it offers abundant material and reflection on the degree to which the period from Diocletian to Heraclius differs from the three preceding centuries. And, not the least of its attractions for papyrologists, it provides an unprecedentedly detailed investigation of the value of the papyri for interpreting late Roman society as a whole, bringing them into confrontation with the legal, literary, and ecclesiastical sources.1 The papyri emerge from this analysis as a generally reliable source for the social practices of the eastern Mediterranean world of late antiquity and probably representative of most of the early Byzantine empire. The very detailed summary of the contents of these two volumes that occupies the first two sections of this review is thus a tribute both to the importance of the book and, unfortunately, to the likelihood that many users will read only portions of it. It also provides the basis for critical comments in the third section of the review. There I shall look at a number of conceptual and evidential problems that, to my mind, the author did not sufficiently consider. 1. Imperial Law The stage is set in the Introduction, which cites a variety of opinions about the situation of women in late antiquity, variously crediting Christianity for its improvement or blaming it for its deterioration, according to an author's views and biases. These opinions have, in Beaucamp's !But not art; visual or archaeological information is nowhere brought to bear.

DOI: 10.4324/9781003556732-2

II 66

view, generally been based far too much on the literary sources; she sees the papyri and the imperial laws as far more reliable guides, with all the rest so much material for verification and fleshing out of the basic picture. The Preamble, occupying §§ 2-4, introduces the discourse about women characteristic of late imperial constitutions. Women are portrayed as weak, passive, easily deceived, morally semideveloped, and inferior to men. With Justinian—who generally assigns a much higher value to women than his predecessors—the vocabulary changes, with the emperors using women's weakness as a basis for protection rather than an excuse for legal incapacities. Despite their low level of moral development, heavy obligations are laid on them, especially in the way of modesty and chastity. Chapter 1 treats incapacities and protections. In § 5 women's exclusion from public life, especially offices, is detailed, and in § 6 the corresponding limitations on their judicial activity are set out: no litigation for others, no representation of others, limited rights to accuse others of crimes (not even the husband of adultery), severe limits on the ability to act as witnesses. In § 7 we see that the restrictions on women's ability to serve as curators or tutors are eroded in the late fourth century and later, at least to the extent that widows are allowed to serve as tutors for their children. A detailed discussion of the prohibitions of the SC Velleianum on "intercessions" occupies § 8. Women were restricted from contracting most obligations for the benefit of a third party and given a defense against suits involving such obligations. The sources of the fourth and fifth centuries are silent on the matter, but Justinian revived and reinforced these protections. Ignorance of the law is the subject of § 9. The Digest has some provisions (especially in matters of procedure, where imperitia was expected), but no coherently developed doctrine. The fourth-fifth century constitutions restrict this protection, showing that it cannot be entirely a compilers' invention. In § 10 prohibition against forced marriages and protection of widows are treated. Most of these are old doctrines that show minimal development in late antiquity. Chapter 2, discussing morality, is the weightiest of the volume. Unlike the protections, the concern with morality is largely a fourth century and later innovation. The first issue is raptus (§ 11), a tremendous preoccupation of late antique legislation, which differs from earlier enactments mainly in its focus on morality, i.e., on repression of immorality (by both women and men). Classical legislation, in contrast, had focused on prevention of violence against women. The offense is constantly classed with murder and adultery, and raptus of virgins is at the top of the list. CTh 9.24.1 eliminates defenses and extenuating circumstances and adds threats against the parents if they connive in the abduction. The woman is also punished, and it is clear that protection of parental choice of marriage partner is the central issue. This is, in Beaucamp's view, repressive and misogynistic.

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

67

Later alterations affect mainly the penalties. Justinian extends the rules to slaves and thus turns raptus from forcible marriage into a crime potentially affecting all women, even those who cannot marry. But he does put all of the responsibility on the raptor (even if he was already the woman's fiance), not on the woman. Before the discussion turns to adultery, two less central issues are handled, laws governing prostitutes and actresses (§ 12) and corruption of chastity (§ 13). The most notable development here is Justinian's elimination of the constitutions that compel actresses to remain in their profession and their daughters to follow them. Adultery, as befits its importance in the codes and to Beaucamp's thesis, gets more than thirty pages. All extramarital sex is criminal for an honorable woman, but not for the man except when with an honorable woman. The fundamental and universally acknowledged inequality of the treatment of the sexes in this regard is oddly characterized as "peu soulignee." The Digest made it difficult to accuse a married woman directly; her husband had to divorce her first. Justinian in fact reversed the order, requiring the accusation of adultery to come first. Constantine eliminated the right of accusation for nonkin entirely. This is clearly a shift in favor of preserving marriage. But Constantine extends the covered classes to include fiancees and concubines. Obligatory prosecution (the classical doctrine) was undermined by its applicability only to inflagrante cases and by the end of public prosecution (an important development). The husband's rights are put ahead of the father's. Justinian's "clemency" is seen also in the penalties in the Novels, which are less severe than the Constantinian ones, and in his provision allowing a husband to pardon his guilty wife. But that is not all. Divorce for misconduct occupies § 15. The motives allowed for a legal divorce in the late imperial legislation are different for the sexes, something Beau camp describes as another aspect of late antique moral repression. Since such divergences in the social acceptability of behavior date back many hundreds of years, it is hard to see anything special about it. Similarly, the behavior of unmarried women (§ 16) is viewed differently from that of men. The condemnation of irregular durable unions occupies § 17: the union of a free woman and the slave of another and concubinage. Despite moralizing, it is clear that the first of these was aimed at protecting owners' property rights. The emperors viewed concubinage as not-marriage, hence illicit. But stable unions were gradually treated more favorably. The laws focus on the ability of natural children to inherit and of concubines to receive gifts. Justinian gradually removed most of the restrictions where there were no legitimate heirs, a change Beau camp calls "spectacular. " Much of the legislation described above affected only matronae, i.e., matres familiarum, women of honorable status. Ulpian had defined the term

II 68

as determined by boni mores, not by marital status, but it hardly seems that young women who had never been married could be included. Excluded from condemnation were non-honorable women (§ 18): actresses, procuresses, prostitutes, tavern girls, and alumnae of those occupations. On the other hand, they were also subjected to other discriminatory measures regarding status and property. Justin I relaxed these rules for actresses and their daughters, and Justinian, in Nov. 117.6 (542) swept away all restrictions. Here if anywhere is legislation driven by an emperor's personal circumstances. From what precedes, Beaucamp concludes that overall moral demands were higher on women than on men—hardly a discovery. But she discerns two evolutions: (1) the penalties for misconduct were highest under Constantius, then diminish; (2) inequality was gradually reduced from the fifth century on, though not drastically. The last two sections of this chapter deal with proprieties: obligatory mourning (§ 19) and divorce and remarriage (§ 20). Mourning was classically limited to women and involved a wait of 10 months for remarriage. The original motive is contested, but may have involved both of the contenders (avoidance of any confusion about paternity, general sense of decorum). Late antique legislation emphasized the moral nature of the wait and became more severe; Theodosius I required a year. The provisions appear aimed at protecting the children of the first marriage. For some reason Beaucamp thinks it all "purely moral." Women also bore a greater moral burden in remarriage, and a delay was imposed. Once again the fourth and fifth century emperors look misogynistic. But it is again clear that the concern of the legislators was mainly for the property of the children, especially that coming from their father. There is a greater degree of equalization in the fifth and sixth century, especially with Justinian, as men become subject to the same moral repression. Chapter 3 considers the limits of women's autonomy and of their power of initiative generally. Beaucamp sets out to show that three key characteristics were accentuated in the late empire: (1) women had little influence on their own lives; (2) women occupied a dependent position; and (3) women's power over their children grew. The first two point to increased passivity, the third to the reverse. The first section (§ 21) is about a woman's role in deciding the course of her life. Marriage was a far more central event for women than for men. Classical law required the consent of both parties and of their fathers if the couple were in their fathers' power. But girls were expected to consent to their fathers' wishes unless there was a good reason not to. Of course, a father's role could also be passive, the girl's active, but that was less common. The same rules apply to engagements as to marriages. Specific cases suggest that women had less autonomy than the general rules would indicate. The enactments of the fourth to sixth centuries freed women to marry at 25 without being disinherited, even without their father's consent, if the

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

69

parents had not by that time married them off. On the other hand, emancipated women under 25 still needed their father's consent. The distinction between minority and majority thus effectively replaces being in power. Where the father was dead, the mother and thepropinqui, sometimes with a magistrate's intervention, decided. The woman's consent remained formal and passive. The reverse seems to be true with divorce; there is no sign that a father could end his daughter's marriage without her consent. Nor do the laws allow disinheritance of a daughter who stays married against her father's wishes. And no one can prevent a woman from divorcing (within the legal bounds). Entry into religious life was the woman's choice, so also leaving it. In § 22 we get a description of "la femme-reflet," as Beaucamp puts it. Marriage was de facto a relationship of great disparity; the woman was "the passive reflection of her husband," who had the dominant role. There are some signs that his intention was primary in distinguishing marriage from other types of relationships (this seems unavoidable, as there is no equivalent of concubinage with the roles reversed). The mother's connection with her children was natural, the man's dependent on his explicit statement. Women acquired their husbands' status and were thus assimilated to them. So too with privileges and exemptions, such as from taxes (this is found in the Hellenistic period already). Beaucamp claims that the late empire accentuated women's passivity. Some detailed investigations occupy the following sections. In § 23 social ascent by marriage is discussed. Classical Roman law was opposed to very radical disparities in the ranks of spouses, which would allow women to rise dramatically. Justinian eroded, then demolished, this barrier. For example, classical law allowed manumission of a slave woman and then marrying her only if the status gap was not too great; the Novels end the prohibitions on unequal rank, which Constantine had still upheld, apparently as part of his general legislation preventing the offspring of irregular unions from usurping rights. Justinian abolished Constantine's law in 542, but he strengthened the woman's dependence on the man in such unions, especially through restrictions on freedwomen trying to divorce their patrons. In § 24 attention turns to concubines. Concubinage was not repressed or penalized, but neither was it regularized. Constantine tried to prohibit legitimation of such unions and their offspring, starting from the date of his law (but legitimating earlier ones retroactively), but this was eroded in the sixth century. Justinian then allowed the legitimation of children from a concubine even without the partner's marrying her in certain cases. Having multiple concurrent concubines was condemned but not penalized; i.e., the man only got a scolding. A concubine's rights depended entirely on the man's wishes.

II 70

The mother's role is discussed in § 25 (rules derived from classical law) and § 26 (late imperial developments). This role was clearly subordinate. Abortion was viewed as depriving a man of his right to a child and thus moderately punished—but apparently there was no penalty if the man also wanted the abortion. A woman had no power over her children, but was owed pietas. Widows, however, acquired more of a role, especially if not remarried. Their wishes for a tutor for their children were accorded a considerable role by magistrates, and they were obliged to request one. Some tendency to increase the mother's responsibility is visible. In the late empire, mothers acquired the right to be tutors (formalized in 390, origins undatable), and even some rights while the father was alive. Some precautions followed: the woman was required not to remarry and had to renounce the protection of the SC Velleianum and hypothecate her property in order to take on the management of property. Overall, the development of the woman's role is much greater as mother than as daughter or as wife, and widows gain at the expense of the agnatic family. The conclusions are summarized in § 27. In matters of weakness and dependence, continuity dominates, with little change from the earlier empire. Women's legal capacity was limited to defending their own interests. Dependence and subordination in family matters also continue; the lack of major change is unsurprising. Change, by contrast, is visible in the increasing weight of morality. Much here is constant, but much is not. Raptus and procuring are offered as major examples of cases where the burden of law increased. Violations of sexual morality and the proprieties both get more rigorous punishment. Beaucamp recognizes that some of this is just the more moralistic style of discourse of the law of the period, but she does not think that this is enough to explain the evolution. Finally, mothers gain power over their children. Arriving at a general conclusion is difficult. Overall, Beaucamp is forcibly struck by continuity. That is not so surprising; what is problematic is that the two major changes she notes do not form a homogeneous evolution, and she is driven to renounce finding any master explanation for the whole pattern that she sees. She points out also that much of the sexual repression loses force after Constantine, whose legislation and penalties seem basically out of line with those of the other emperors. Justinian, especially, is milder, although his moves toward equalization of what is expected of male and female remain modest. The individual reigns and periods can thus be seen to have their own "tonalities," with Constantine and his sons severer than others. Justinian makes the strongest impression of an individual character: his innovations are all favorable for women, and he even says favorable things about them, unlike the others. How is one to explain this? Beaucamp rejects "contingent" explanations, like Justinian's marriage to Theodora, as being insufficient.

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

71

Christianity, the main explanation generally offered, is equally problematic, especially because one cannot tell if whatever influence it had would have been exercised directly on the process of legislation or rather through the larger role of the east and its gradually Christianized population in the formation of law. The discussion thus must turn to the study of social practices to get a handle on this. Beaucamp advances two main reasons to expect significant divergences between practice and law: (1) The law is of centralizing Roman origin, whereas the population was multiform and diverse; (2) the sources for later law are compilations, and it remains hard, even with intensive study, to know what in them was still valid and what was just a fossil. 2. Social Practices And so we turn to Beaucamp's attempt to understand social realities. The Introduction notes that from the late third to the mid 7th century over 1300 papyri involving women survive. They compose the material of Part 1. Part 2 then treats scattered information from "protobyzantine" sources, which are used to test and complete the information of the papyri. Chapter 1 deals with women's exclusion from public life. In § 28 Beaucamp treats magistrates and other public offices, of which she finds no instances in the third and fourth centuries except for priestesses.2 Things seem to change in the second half of the sixth century, with women attested in civic offices, but these are actually public responsibilities attached to patrimonial holdings, munera patrimonalia, and women exercise no actual authority. It might be pointed out that financial responsibility had always been the main point of such functions, and in any case it is only asserted, not actually proven, that in the sixth century the women mentioned did nothing except accept financial liability. Women also sometimes had inherited obligations deriving from paternal liturgies. No women are known to have served as judges or arbitrators, even before this became illegal. Beaucamp concludes that the papyrological evidence conforms to the laws, and that no evolution is visible. She notes also that all daughters of bouleutai whose husbands are known married other bouleutai. In § 29 other public roles are treated. There is no evidence for women representing men at law, but women do appear on their own behalf in many cases. Similarly, women appear in trial records as witnesses on several occasions, but they never act as witnesses to contracts. § 30 considers specific protections for women mentioned in the legal sources (§§ 8-10), with little result. Widowhood is evoked rhetorically for 2 For the 5th century she notes TroXtreuo/ie'^ in SPP XX 114, not having been able to know Worp's correction in BL 8.467 (1992) to iroXiTiq 'PayieW (read 'PoyimcojO.

II 72

captatio benevolentiae (nothing new there), but not for any specific statusrelated protection. Much the same is true of feminine weakness (the references, none after the fourth century, are collected on p. 46). Intercessions, in the technical sense, figure only in that women appear occasionally as guarantors; these typically are widowed mothers. The problem of intercessions simply does not seem to have been a live issue. For forced marriages and excusing of ignorance of the law there is no evidence. Chapter 2, on morality, begins with the fact that the papyri are short of information on morality. § 31 treats social ethics and the proprieties. There is little evidence for concubines, and none for men taking concubines instead of remarrying. For free women's unions with the slave of another person there is also no evidence. Nor is there much about prostitutes and actresses (Beaucamp sees BGU IV 1024 as a rhetorical exercise). Imperial legislation on these groups is never cited in papyri. The section on slaves and freedwomen is not strong, and Beaucamp unfortunately follows Fikhman's views that slavery was in decline. She is wrongly skeptical that women would commonly escape slavery.3 The proprieties are also not well represented: there is no reference to any obligation of widowhood, for example. All known dissolutions of marriage permit immediate remarriage; their only interest is in the dissolution of property links. Practices thus oppose or ignore imperial legislation. A list of attestations of female remarriage is given (p. 67), although this status would in any case rarely be documented. There is, Beaucamp thinks, no sign that children were an obstacle, and no sign that social practice followed either the imperial or the Christian disapproval of remarriage.4 In § 32 the difficult subject of sexual morality is treated. Only a few papyri deal with rape or abduction. On the whole, it seems that marriage, not prosecution, followed abduction—just what the emperors tried to prevent. Imperial concern to avoid the imprisonment of women (to avoid compromising their chastity) seems to have been ignored, even by the authorities: there are a number of documented cases of women being detained, mainly for financial reasons. On the repression of misconduct 3 Against Fikhman, see my "Slavery and Society in Late Roman Egypt," Law, Politics and Society in the Ancient Mediterranean World, ed. B. Halpern and D. Hobson (Sheffield 1993) 220-40. For virtually universal manumission (later for women than for men, however), see R. S. Bagnall and B. W. Frier, The Demography of Roman Egypt (Cambridge 1994). 4 The evidence of Roman census returns, however, suggests that women actually remarried after divorce or widowhood at a much lower rate than men did. See Demography of Roman Egypt (above, n. 3) 126-27, 153-55. The dramatically different patterns for men and women found in Roman Egypt are paralleled by those in fifteenth-century Tuscany: see David Herlihy and Christiane Klapisch-Zuber, Tuscans and their Families. A Study of the Florentine Catasto of 1427 (New Haven 1985) 213 fig. 7.3 for a graphic representation. (They seem to suggest on 217 that the non-remarriage of widows was a matter of choice, but the Catasto provides no warrant for such an inference about motives.)

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

73

there is little clear information. Even the few references to punishing misconduct do not seem to connect at all with imperial law. Divorce for misconduct is handled in § 33. Only two marriage contracts relate moral obligations to anticipated causes of divorce, and those inconclusively. The partners' duties were, unsurprisingly, asymmetrical. In some marriage agreements, the husband's only duty was support, while the wife's were largely moral; in others, the husband had more detailed obligations spelled out, some of them moral, while the woman's moral duties were treated briefly. Acts of divorce, on the other hand, with one exception, are all formally a matter of mutual consent. The Trovrjpbs bal^v is generally blamed.5 Only P.Oxy. I 129 preserves a unilateral divorce. A consideration of P.Oxy. L 3581 leads to the conclusion that imperial enactments on process (how to divorce a spouse) fared better than those on substance (why you can divorce a spouse). Beaucamp rejects as bad method the use of NovTheod 12 to date this papyrus after 439, mainly because of the lacunose character of our knowledge of legislation and the underlying uncertainty about its effectiveness. There is no proof of a direct linkage of the terms of marriage or divorce to the imperial laws, nor any evidence that in the fourth century unilateral divorces suffered any legal consequences except any agreed on contractually. Constantine, in sum, leaves no echo in the papyri. People seem to know and "use" imperial law, but not to observe it. Overall, the evolution of morals seems less marked in practice than in law; put another way, marriage in Egypt is still dominated (as it had been for many centuries) by provisions of private law, not by imperial criminal enactments. Chapter 3 deals with the location within the family of power over women's lives. In first place comes the ability to conclude a marriage (§ 34). The fifteen surviving contracts related to marriage are varied in nature and preservation. The groom is a party to all but one (in P.Cair.Masp. I 67006 verso it is his father); the woman is a party in six, her father in about the same number, her mother in two. Many of these are purely property settlements, but even they show the father as the key figure. Since the parents usually supplied the bride's dowry, this is not surprising. Beaucamp concludes that for a parthenos, her parents normally arranged a marriage, but that women concluded their own marriages if orphans or formerly married. The contracts do not, alas, answer the controversy over whether it is consent or ekdosis that makes a marriage. Narratives about marriages offer similar problems. The vocabulary connected to the act treats women as passive, but whether (sK)bibu>iu is descriptive or juridical is hard to say. Interestingly, the persons mentioned most often as intervening are mothers, with the father in second place. Others (uncles, older brothers) show up 5

At p. 90 n. 40 there is a long note on this topic, good on the jealous charactaer of the demon but missing altogether its connection to the evil eye.

II 74

only occasionally. Beaucamp argues that "to give" has a strong sense and refers to the person who actually concluded the marriage. She remains unsure about the ability of a father to give a daughter in marriage without her agreement. None of this seems to reflect much change from the Roman period. In § 35 women's subordinate role in the household is explored. The marriage contracts of the sixth century show clearly an asymmetry of duties, with the wife required to obey her husband; this is a matter of custom, not of law. Wives were usually of the same origo as their husbands; residence and domicile questions thus play little role in the papyri. Relying on documentation in Annexe V, Beaucamp tries to see if a wife's status followed her husband's; overall this seems to be so, although by the last third of the sixth century women of the senatorial aristocracy tend to inherit their rank to some degree. There is an interesting discussion of the matrona stolata, tending to reject Holtheide's thesis that this term always refers to the wives of equites. Scholarship on divorce (§ 36) has been dominated by the question of a father's ability to end a daughter's marriage. The surviving divorce documents do not help very much, since all except P.Oxy. 129 are dialyseis by mutual consent, in which marital obligations are liquidated. They are always symmetrical and carried out by the spouses; the woman acts autonomously.6 In P.Oxy. 129 there is no evidence of the daughter's views. Beaucamp concludes that probably a father could end a marriage without the daughter's consent, but there is absolutely no evidence in this document to support this view. Narratives, as in petitions, can help flesh out the formalism of legal acts, but they are mostly very partial, in both senses of the word. Against one known case of a wife's abandoning her husband, there are three cases of a husband's expelling his wife, plus some abandoned women. Beaucamp notes that women had a need to litigate to get their property back at the end of a marriage, and men did not. For assessing the role of the family, P.Oxy. LIV 3770 and P.Sakaon 38 are the key texts; the latter especially points to a father's role in reality much greater than the authorities—bound to enforce the imperial laws, where a woman's consent is central—would accept. From all of this, Beaucamp concludes that feminine passivity was normal. Thus legal acts and social realities operated at different levels. In this, too, she sees a continuity with earlier centuries. The relative powers of the father and mother in a family are treated in § 37 (the situation with both living) and § 38 (after the father's death). For adoption the three known documents show couples acting together when both are alive, but women acting alone when widowed. Beaucamp argues 6

P.Ness. Ill 33 alone stands out; but it is not certainly a divorce settlement and is, Beaucamp argues, probably a patrimonial post-marriage settlement, not a divorce.

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

75

that widows' power was derivative and that women could not, unlike men, adopt alone. Egyptian adoption remained unconnected to Roman practices, not being a matter of patria potestas but rather of creating rights of succession. A key question is whether widows simply raise orphans or have a real tutela. P.Sakaon 37 offers an interesting example, in which a woman is KrjdsarpLot, best taken (Beaucamp argues) in a general sense (one of the children is between 14 and 25, and thus would have a curator, not tutor). But there are no later examples of the term, despite the evolution of imperial law. An analysis of instances of tutors or curators who are not the mother shows no case in which the mother is demonstrably living; there was thus a preference for the mother if she was living. But the father could dispose otherwise by testament. Widowed mothers acting as tutors had real control over the children, including the ability to give them in adoption or marriage, or to put them out to service or apprenticeship; this was also true in Roman period. With the property, similarly, they took part in settling the succession from the husband and acted for the children. Beaucamp argues that women normally took on curatela/tutela responsibilities only when they were widows, that their power derived from the mother's role and father's death and did not need any official designation. If anything, she thinks, Egyptian practice was ahead of the laws. Once again, widows appear to have been far ahead of girls and married women in their ability to act independently. Chapter 4 is a remarkable and satisfying study of the transformation of the guardianship (or tutela) of women into the husband's power over the wife, something of a tour-de-force of papyrological investigation. Early studies of the tutela mulierum, Beaucamp believes, tended to see it too much in Roman terms. The Greek papyri, of course, use the term Kvpioq, rather than (like Latin) use the same term as that used for tutors of minors. In § 39 the formulas connected with tutela are examined in detail. Strikingly, for the 350 years of this period, only four documents (scattered in time) use the phrase yisra. Kvpiov, which is thus virtually extinct, while more than 60 use phrases mentioning absence of a kyrios. These thus seem to be operating on different planes. In all four positive cases, the kyrios is the husband. This continues the normal Hellenistic and Roman practice in Egypt—in contrast to Roman law, where the tutor normally was not the husband. Women acting without a kyrios appear in 66 cases from 284 to the 7th century. Beaucamp has sharp criticisms of Sijpesteijn's articles on this subject for failing to make essential distinctions. She identifies 13 formulas (listed, with examples, pp. 199-202). They show chronological and geographical variation, with actual mention of the ius liberorum dropping out of use after 389. They are thus not at all a homogeneous body of material and need much more analysis to answer the many questions raised: What does the post-389 formula mean? Is the absence of a tutor a

II 76

permanent right? Does a woman using it lack a husband or act without one that she has? Beaucamp argues that the formulas refer to a continuing state and specifically denote widowhood. There is not a single instance in which a woman using this formula is known to have had a husband. The usage of the terms for widowhood (xripot, xripsvovaoi) in documents is similar in character to that of the "without kyrios" phrases. Among the conclusions derived from this investigation are the following: (1) Only husbands were tutors; (2) from the second half of the fourth century on, if not before, husband and tutor were the same; (3) widows escaped tutela in the sixth century, mothers in the third-fourth century. But these then lead to new questions: (1) When did the change occur? (2) what about divorcees and the unmarried? (3) was this a matter of the husband's control or of a wife's incapacity at law? Overall, Beaucamp thinks that the freedom of widows from tutela dates to the mid fourth century; there were thus dual escape routes from tutela for a time. Some fourth century documents have both "without a kyrios" and a husband present. This is a sign, Beaucamp shows, that even ca 300 incapacity was not the issue any more; the husband's power was. That is, a woman could have the juridical capacity to transact business independently but still need to have the husband involved. Increasingly, the husband was seen less as a tutor and more as a husband. The argument is pursued in § 40, where Beaucamp undertakes to look systematically at documents showing tutela and husbands' appearances to see how they resemble one another or differ. This is, she notes, a difficult process with methodological pitfalls that she describes with care. It is clear that the ius liberorum is not used systematically. This may be partly the reflection of the type of document, but it is easy enough to show that this is an insufficient explanation.7 Rather, it looks as if the nature of the transaction, not its form, is critical. The central problem is the alienation of patrimonial goods. A detailed study of transaction types shows that the husband is virtually always present if a significant impact on the couple's property is at stake. In § 41 the argument is continued through a comprehensive analysis of all legal acts involving either a husband or a tutor. This shows that references (positive or negative) to tutela by married women or widows are uniformly present only in those acts where mention of the husband is normal. Elsewhere, it is variable. Thus all the references to tutela essentially refer to marital control, throughout the Byzantine period. There is no evidence for the autonomous existence of tutela in the Byzantine period; it is always an expression of marital control. The final stage of this investigation (§ 42) covers the presence of third parties in documents; these are mainly persons with a property interest (especially one of succession) in the goods being alienated. Beaucamp then 7

Beaucamp shows also that no geographical or chronological explanation is adequate.

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

77

looks more closely at the term owfiorco^, which she shows is not a technical juridical term (instances are listed in Annexe VII). Mostly it is not a form of tutela (thus connected with juridical incapacity) but an expression of marital power. The conclusions reached are then in § 43 connected to the development of imperial law. The tutela mulierum disappears in law, too, but the moment is unknown; it could have been by abolition or by disuse. The last mention is in Frag. Vat. 325 of 293 or 294. As the husband-tutor is already a third century development in Egypt, this can hardly be a case of Egypt's diverging from the empire in the fourth century. Even if the husband-tutor is not found in imperial law, he is consistent with a late antique tendency to favor the husband's involvement in the wife's patrimonial affairs. Once again, then, the legislators favor the reinforcement of the nuclear family cell, privileging the common patrimony of the couple. Beaucamp believes that the changes are simply a legitimation of longstanding social practices, not a big ideological change. But this brings her to move to the nonEgyptian evidence. The "protobyzantine" world, which occupies Part II, is approached through an enormous range of source material, each genre of which needs to be considered on its own terms. The different types of material are not easily comparable to others. The inscriptions yield little for this subject, and overall Beaucamp finds in these sources few facts and many representations: attitudes about virtues and duties. The discussion proceeds in the usual systematic fashion. Chapter 1 is thus about seeing whether the pattern is one of incapacities or of exclusions (§ 44). Just as in the papyri, the only public function of women is as priestesses. Nothing corresponding to the fiscal burdens of the papyri turns up in other sources, and only a few references to patrimonial burdens. Women, especially widows, did use the courts, but almost entirely to defend their own or their children's interests. The protections for women found in legislation are missing in other sources as they are in the papyri. There are many references to widows as needing protection and as potential victims, but as in the papyri it is a matter of captatio benevolentiae, not of specific rights to be protected. Similarly, there are references to women's weakness, often using the story of Eve. Beaucamp suggests that the absence of this theme after the fourth century in the papyri is probably chance. In § 45 the norms of the church are brought into the discussion. Women's participation in religious rites is routinely taken as a sign of heresy. Their exclusion from the priesthood was taken for granted, based both on biblical examples and on arguments based on the subordination of women. Similarly, women were forbidden to teach in public, but not in private; their teaching girls was normal. None of this, certainly, is an innovation of the period. Family powers are the subject of chapter 2. The question of who decided about a girl's marrying (or not marrying) is brought back for a

II 78

reprise in § 46. All of the abundant information on the arranging of first marriages (where the woman is called kore or parthenos) indicates the woman's passivity, which becomes more visible as the social status rises. The evidence supports the idea that early marriage was the norm, but some anecdotes refer to engagements, and these are not easily disentangled. Paternal power was paramount, even though both parents were sometimes mentioned. Widows, however, acquire more independence. The religious life was in principle a woman's own choice, but it was often hard to realize if she was faced with her family's hostility. On the other hand, a family might force a monastic vocation to avoid coming up with a dowry. In divorce the situation is similar to that in the papyri: parents (and mainly the father) play a role in social reality, but the state concentrates officially on the woman's wishes. In § 47 the wife's dependent status is explored. A wife's status and fate are universally represented in the early Byzantine sources as depending on those of the husband. But the quality and quantity of real evidence are not great, and hard questions on the margins are unanswerable. Not surprisingly, the overall picture is one of endogamy within social strata. Social climbing by women is viewed positively in most sources. Husbands were expected to treat wives kindly and instruct them; one is strongly reminded of Xenophon's Oikonomikos (not mentioned by Beaucamp). But this stress on the husband's duties suggests that in reality there was a high incidence of domestic violence, especially routine "correction" of the wife by the husband. Nothing in this sphere need to be assigned to Christian influence, as little of it is new in late antiquity. The literature is particularly clear on the husband's domination of the wife. But Beaucamp cites Julian's criticism of the Antiochenes (Misop. 27, 35; Letter 84) for allowing their wives too much autonomy. One might well conclude (but Beaucamp does not) from this that the commonplace male vision of obedient wives was as far from reality as many other moral norms. Various questions about mothers' roles are treated in § 48. The frequent references to mother love and widowhood are not very helpful in discerning the degree of maternal authority. Church and society were both hostile to abortion. Fertility was, after all, usually essential to a wife's status and even to the survival of the marriage, and reproduction was in all ancient thought the principal, even the only, purpose of marriage. Christian thought developed a strong view of male/female equality, but this did not lead the church to adopt any different legal norms from those of the surrounding society. Unlike the papyri, the literary and ecclesiastical sources are very interested in moral prescriptions. In § 49 the social norms and legislative strictures are compared. There is, just as in the papyri, little sign of any application of the imperial constitutions; what consequences there are for seduc-

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

79

tion and similar misdeeds are nonjudicial. The sources seem to suggest widespread knowledge of the stated penalties for adultery and similar crimes, but concrete cases of actual application are scarce. Continuity is visible in the fact that the importance of moral obligations varies with social status. Peasant women, unlike their wealthier urban counterparts, were not secluded. Literature accepts the killing of adulterers by the husband and expects respectable women to be invisible. There is praise for women married only once, and also for widows who do not remarry; but if there are no children, there is no blame for a remarriage. "Sous ces deux angles, la legislation nouvelle apparait done en harmonie avec 1'ensemble de la societe protobyzantine. Ou, plus exactement, avec les principes qu'elle met en avant. Car la pratique du remariage, dans la mesure ou elle peut etre saisie, suggere des conclusions quelque peu differentes" (p. 350). Remarriage and divorce were known in all social circles, and in general Beaucamp regards the papyri as representing typical behavior (352). A comparison of imperial and ecclesiastical norms (§ 50) finds both convergences (about actresses and other low estates, and about nuns and on deaconesses), and divergences (about abduction, where the church is less severe and more inclined to accommodate social realities, especially favoring the marriage of the parties; and about adultery and infidelity, where the church's penalties are severe but equal and there is hostility to any publicity that might harm the parties). Church texts show two tendencies, one (in authors) ethical and equalizing, opposed to both legislation and the prevailing social norms, the other (in canons) more nearly reflecting civil conceptions in the penalties set for some misdeeds. As the church was more hostile to divorce than the state, so was it also to remarriage. The conclusions (§51) give a final look to some of the key questions that occupy this book. Beaucamp thinks that the "society of the papyri" is despite some difficulties the most readily describable zone of late antique society. There are both areas of stability and areas of change. In the former, she cites women's absence from the public sphere and the modest degree of change in the family throughout the period. For the latter, the most salient feature is the growing role of widows in controlling their children's fates.8 Social practices elsewhere seem to be in agreement with the papyri. They change much less than the legislation, which gradually seems to be taking account of practices in some respects. The legislation about morality, however, does not conform to this pattern. This, Beaucamp argues, is because of the influence of the church. But she admits that there is absolutely no evidence for such influence except the convergence of ideas in the two. That is, the argument is not even post hoc 8

This is indeed an area in which the change from the earlier empire seems to be genuine. See Susan Treggiari, Roman Marriage. lusti Coniuges from the Time of Cicero to the Time ofUlpian (Oxford 1991) 501.

II 80

ergo proper hoc. The two streams run side by side. Legislation certainly saw much more change than did reality. Beaucamp attributes this difference to the much swifter Christianization of the state than of society. In the end, she refuses to see all of this as either progress or regress for the status of women, at least not in any simple fashion, and she thinks that the role of Christianity is generally overestimated. The work concludes with two key questions. First, the relationship of law and society, certainly a complex situation. Some elements of law, mainly those affecting official behavior and public life, are clearly effective. In some others, application is partially effective; yet others are known but not applied. In still others, the authorities try to enforce the law but are ignored. Where there is silence, then, one must be prudent and not assume that any one of these patterns offers a simple answer. Second, the individuality of Egypt. There are large areas where the evidence from elsewhere confirms the picture in the papyri. In some others, there is no way of confirming or disproving the wider applicability of the Egyptian evidence. What discernible differences there are may reflect only the state of the documentation. Overall, any such distinctiveness of Egypt seems to be minor, and there is not one specific question in which it can be proven. Seven appendices set out in more detail the evidence for some of the key subjects treated in this volume: I: Proposals for readings, restorations, datings, and identifications in the papyri. II: Women of curial family. Ill: Women engaged in judicial process or arbitration. IV: Requests emanating from women. V: Titles and dignities of women: matrona stolata, Xa/xTrpordrT/, £j>6o£orc*T77, fjisyaXoirpsirsaTaTrj, i\\ovoTpia, virctnaooi, TotTpudot. VI: Contracts to which a woman is a party. VII: HvvsaTuq and related terms. Each volume has a full index of sources, and there is an index of subjects (including Greek and Latin words) to both volumes at the end of volume 2. 3. Critical Reflections This is obviously a work of remarkable range and power. Even the bare sketch of the subjects treated and the main conclusions given above should give some sense of the vast reading of ancient sources and modern bibliography that Beaucamp has carried out. This is a work of great learning. It will not end discussion of the subject she chose, but it will obviate any foreseeable need to repeat the work of collecting and analyzing in detail most of the evidence. The treatment of individual texts seems to me usually accurate, sound, and dispassionate, despite Beaucamp's clearly stated and strongly worded opinions about the subject. She is almost always well

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

81

informed about critical literature treating the papyri and other texts she discusses. Many of her conclusions are original, and all readers of this book will find their own approach to the subjects it treats transformed. Many books, no matter how long in the making, seem destined to appear just when another year or two would have allowed the author to take into account important contributions to the subject. The most obvious instance here is Susan Treggiari's Roman Marriage (Oxford 1991), which appeared in the year between Beaucamp's two volumes. It might have spared Beaucamp some of the detailed analysis of the classical background of late antique legislation. Anyone who reads both books cannot help noting the large areas of convergence, strengthening the sense of continuity between classical and late antique Roman norms and behavior. Because Treggiari treats legal and literary materials together, she is much less repetitious than Beaucamp and often more realistic in discussion of the legal questions. Treggiari does, however, pay scant attention to documents and almost none to the papyri; her focus on the Roman elite is far more pronounced. Gillian Clark's Women in Late Antiquity: Pagan and Christian Lifestyles (Oxford 1993), on the other hand, was able to profit from volume 1 of Beaucamp's book and several of her articles, but not from volume 2. Clark's short textbook is certainly written for a different audience, and it will be accessible to students in a way Beaucamp cannot be. Unfortunately, it shows almost no interest in the documentary evidence that Beaucamp exploits so brilliantly and effectively. Despite the many merits of this book, there are many aspects of the author's approach and conclusions that I find problematic. First and foremost are the main components of the subject, as stated in the title. There are only occasional nods to the difficulties posed by "women" as a unitary category (a problem with Clark's book, too). The feminism of the author's approach seems to date from a more youthful period of women's studies, when gender alone appeared to offer a magic key. By now it is widely recognized that—at the very least—differences of social and economic standing fragment women into groups with divergent experiences. Our sources are enormously more rewarding for the top stratum than for lower, and this is true of the late antique papyri as much as it is of the literary sources. To be sure, the papyri do not limit us to the tiny aristocratic crust of Constantinopolitan society, nor even to its counterpart in other major cities. But they are overwhelmingly urban and emanate with relatively few exceptions from the propertied elements in the upriver cities of Egypt.9 Their convergence with the literary sources, which Beaucamp well documents, is important and striking, but it is very much bound to status and class. 9

See my remarks on this subject in Egypt in Late Antiquity (Princeton 1993) 4-6.

II 82

Just a single example will suffice to show the kind of difficulty at stake, but it is an example central to the conclusions of this book. Widowhood keeps emerging in Beau camp's discussion as a much more attractive status than marriage—attractive, that is, judged by the standards of independence of action. To the women of wealthy families who are the source of the documentation used in this book, there may be something to this idea. But to the poor and destitute woman, the loss of a working husband was certainly a disaster, leading in some cases to unattractive unemployment, even to prostituting herself or her daughter.10 Independent control of property was not meaningful to those who had none. It is hard not to see a certain upper-class fear of and fascination with widows in the picture Beaucamp has given us; the quotidian reality for most people was very different. "Byzantine" is equally problematic if not equally important. Despite our long usage of the term in papyrology, it is difficult to defend as a term for the period from Diocletian through the fourth century, wherever one may wish to distinguish periods.11 Perhaps the most important drawback of a Byzantine conception of this book is that it allows the author to exclude all western evidence. If the evidential record in the East were ampler one might not regret this omission; it can certainly be argued that it allows an unbiased investigation of the East and its eventual comparison with the West. And who would wish to see another 500 pages added to the bulk before us? But when we are told that the epigraphic record for this period is meager—which is true for the Greek East—how can the vision of the massive volumes of the Inscriptiones Christianae Urbis Romae help but rise before our eyes? There are other important conceptual and theoretical difficulties. To start with volume 1, Beaucamp is well aware that the normative legal sources may not reflect the actual state of affairs, and she explicitly seeks to confront them with other evidence throughout volume 2. The results of the discussions there often suggest (as the summary above shows) that these norms were by no means observed. But the discussion in volume 1 constantly slips back into a mode suggesting that the legal sources do display the social realities of late antiquity. Most strikingly, Beaucamp never really comes to terms with the view—to use A. H. M. Jones's formulation—that the laws' "chief evidential value is to prove that the abuses which they were intended to remove were known to the central government."12 From that l °Cf. Herlihy and Klapisch-Zuber (above, p. 80, n. 4) 113 for a similar phenomenon at Florence in the fifteenth century. ^See on this A. Giardina, "Egitto bizantino o tardo antico? Problem! della terminologia e della periodizazione," Egitto e storia antica (Bologna 1989) 89-103. 12 The Later Roman Empire 284-602. A Social, Economic, and Administrative Survey (Oxford 1964)1, viii.

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

83

point of view, legislation is a guide to what was happening that the government felt it had to prohibit, either because of its visibility or because of its widespread character. If anything, then, the laws would suggest that the opposite of what they ordain was commonplace, and that what they order reflects an elite code of behavior. They may therefore be seen as corresponding to Julian's criticism of the Antiochenes for allowing their wives too much autonomy: an elite male view of how things ought to be, but not a reflection of unruly real life. The organization of volume 1 is of considerable interest and concern in its own right. Beaucamp rejects the usual legal framework of subjects and questions, deriving ultimately from Gaius's Institutes, in favor of her own tripartite approach (incapacities and protections, morality, limits of autonomy). That structure has its merits, but she does not defend it on any a priori grounds, nor does she adduce reasons to think that it corresponds to any ancient structure of thought. I shall suggest below that it may actually lead her to separate things that belong together and thus reach doubtful conclusions.13 In volume 1, this problem of organization becomes important in thinking about the emperors' motives. Beaucamp does not always distinguish clearly between the effects and the intentions of legislation. Since effects are not demonstrated in volume 1, all the discussion can hope to do is to tell us about intentions (that is, the ideology underlying the enactments). That is valuable in itself, but it matters a great deal whether emperors enacted rules affecting women because they were specifically concerned about women or because they were interested in other subjects. In the second case, decisions or views about women would have to be seen simply as byproducts of a concern with larger legal questions or social structures. Beaucamp's general hypotheses in volume 1, however, depend in part on the assumption that women per se actually occupied an important space in the emperors' thinking. This may not be true. It is, in fact, possible to take all of her findings in volume 1 and from them construct a social ideology and long-term development that affects women but is not principally about them. Beaucamp recognizes this development but does not ask whether it might explain much of what she observes. For example, at various points (lastly the conclusions to volume 2) she notes the importance for the status of women of the decline of the agnatic family. We may suppose that the legislative changes of late antiq13 One anomalous deficiency in Beaucamp's treatment of the legal sources needs to be noted here. She consistently cites the Digest as if it (and the classical jurists) were a unitary source, troubled only by questions of the compilers' abridgements and interpolations. But the classical jurists were characterized by numerous controversies, and treating them in this way does violence to their thought. Beaucamp hardly ever even indicates from which jurist a particular passage from the Digest comes, a practice for which I can see no conceivable defense. Treggiari is often preferable on matters of classical law discussed by both authors.

II 84

uity were responding to the well-known, long-term shift of Roman society from an agnatic family structure to one based on the marital union and the family created by it. This shift may (as Beaucamp seems to recognize) derive in part from an increasing impact on legislation by the social norms of the eastern empire. But there are signs of it in the classical western sources, too. It makes the survival of marriages much more important than it had been before. The liberality of the classical Roman law of marriage and divorce is inseparable from a legal structure that largely insulated the transmission of patrimonial property to male heirs from the vicissitudes of marriages.14 As property came to be conceived of as the goods of the couple,15 as de facto happened already under the earlier empire,16 this link became unstable. It is hardly surprising that a whole panoply of measures (some "repressive", some not) aimed at regulating marriages and stabilizing them came into being. These included preserving parental choice in the formation of marriages (hence the hostility to abduction), keeping partners together and reducing divorce, enabling the mother to take care of children after the father's death, and many of the other rules and practices described in both volumes. In the long run this shift is of enormous importance. Whether or not it is positive for women is difficult to say. Overall, I think it was; the classical "liberalism" of Roman law worked in large part because women did not really matter. Now they did. But there was a price to pay for mattering. Constantine was ferocious about it, Justinian almost pastoral; but the underlying concerns were the same. The structure imposed on the legal evidence continues to organize volume 2. There is a certain logic in arranging the material from the documents along the same lines as that from the legal sources, but it creates a predisposition to ask the same questions where different ones might be more interesting. Volume 2, in short, seems to me far too bound to legal interests to be able to support the broad conclusions that it reaches. The persuasiveness of the conclusions of volume 2, then, is much less than it might seem at first reading. The questions asked are prefabricated, the material from earlier periods is not brought into the discussion, and the entire approach is much less conscious of the limits of the evidence than 14 Cf. the rhapsodic treatment by F. Schulz, Classical Roman Law (Oxford 1951) 115121 (118: "Free marriage had no immediate effects on property"), 132-36. Even the brief and sober remarks of W. W. Buckland, A Textbook of Roman Law3 (rev. by P. Stein, Cambridge 1963) 106-07 note that "Apart from issue, the effects of marriage were few, in law, a result of the Roman conception of liberum matrimonium." 15 See now Beaucamp's important article, "L'Egypte byzantine: biens des parents, biens du couple?" Eherecht und Fami liengut in Antike und Mittelalter, ed. D. Simon, Schriften des Historischen Kollegs, Kolloquien 22 (Munich 1992) 61-76. 16 See Treggiari, Roman Marriage, chapter 11.

II WOMEN, LAW, AND SOCIAL REALITIES IN LATE ANTIQUITY

85

one would like. Various remedies would have been possible. One might, for example, ask systematically how far what is found in the documents could have been applicable to various other sectors of the female population. One might simply define as precisely as possible which women are being discussed and argue that the others cannot be studied meaningfully with the available evidence. Or one might look for comparative evidence, from other periods of Egyptian history or from other societies, in part to allow other questions to emerge, in part to suggest hypotheses for testing. Throughout the reading of Beau camp's two volumes, and especially in the second, I was struck by the sense that parallels for many of the social situations she describes could readily be found in Ptolemaic Egypt. And it is here that a troubling set of paradoxes and difficulties remains. If Egyptian practices in the domain of male-female relationships and family structure in the Roman period change only modestly in late antiquity, and if the practices of late antique Egypt resemble closely those of the rest of the East—both claims supported by Beaucamp's investigation—it should follow that these practices in Roman Egypt also resembled those of the rest of the Roman East. The next step is to point out that Roman Egypt is in this respect generally considered to bear a marked resemblance to Ptolemaic Egypt, and that Ptolemaic Egypt is usually supposed to be strongly marked by the encounter and partial transformation of Greek marriage customs with and by distinctively Egyptian patterns of behavior. At this point we face several possible directions, all difficult. Egyptian patterns could be argued to be much less distinctive than the ancients thought. Perhaps much more of the Hellenistic and Roman East resembled what the Ptolemaic papyri indicate. Or perhaps the Greek style of behavior seen in the evidence from classical Athens may have been less typical of the Greek world than we imagine. Or our constructs of "Greek" and "Egyptian" may themselves be flimsy deceptions, usable as they stand only if we ignore differences of class. These are not, of course, mutually exclusive possibilities. I rather imagine, too, that large parts of the picture could be matched in Mediterranean cultures of later periods.17 This larger context is entirely lacking from Beaucamp's treatment of the evidence. It is, yet again, hard to reproach the author of a work as long, as exhaustive, and as learned as this for not having extended the investigation still further. And yet I do not see how to draw any meaningful significance from this mass of evidence and analysis without a wider context. In particular, the large and haunting questions with which Beaucamp begins18 look as unresolved as ever at the end. 17 For one example see the citation of Herlihy and Klapisch-Zuber (above, p. 80, n. 4) on remarriage after death of a spouse. 18 These, to be sure, have problems of their own. Beaucamp seems to have accepted the widespread notion that change is what defines history, and set up change as the central problem of the book. This may in itself distort the investigation.

II 86

It is, however, a tribute to what she has done that the work of synthesis can, for the period of more than three centuries that this book treats, build on such a solid base.

Ill Missing Females in Roman Egypt

On a late summer day in the year 362 of our era, in a village in one of the remotest corners of the Roman Empire, three Egyptians executed a contract for the sale of a slave girl.1 The agreement was drawn up in the village of Kellis, today Ismant el-Kharab, in the Dakleh Oasis — then called the Mothite Nome after its capital, Mothis — of the western desert of Upper Egypt. This village lay more than 300 km from the nearest parts of the Nile Valley, itself already a distant and exotic locality to Roman tastes.2 The contract is, in terms of law and form, nothing remarkable, although it is the latest such sale of a slave from Egypt to include a price. The participants — a married couple as sellers, a village carpenter as purchaser — look equally unremarkable. But in this transaction is encapsulated, I believe, an entire pattern of behavior that seems likely to have been typical for Egypt as a whole and probably much of the Roman Empire. Most of the elements of this pattern have been noted by students of ancient history, but the way in which they fit together into a pattern has generally been either ignored or denied. Neither this single contract nor the other evidence I shall cite can fairly be said to prove the reality of the pattern beyond all doubt, but I hope to offer persuasive arguments for its reality. First, let us examine the details of this sale. The sellers were Psais son of Pekysis and his wife Tatoup, officially from Kellis but, they say, actually living in an epoikion, hamlet, the name of which is damaged beyond readability. The purchaser, Tithoes son of Petesis, is a resident of Kellis. Their names are all Egyptian, Psais coming from the god of fortune Sha'i and Tithoes from the local god Tutu.3 Tithoes, a carpenter, lived in part of a large and elaborate house in 1

2

3

P.Kellis I 8, August-September, 362. I am indebted to Klaas Worp and John Whitehorne for knowledge of this text in advance of its publication. This article draws heavily on the work by Bruce Frier and me cited in note 5, and I must acknowlege particularly my debt to my co-author's demographic expertise. Equally, however, he is not to be held responsible for the speculations in this article which go beyond the account given in our book. For the excavations at Kellis, in my view the most interesting and important excavation of a site from Roman Egypt since the Michigan work at Karanis, see the introduction to P.Kellis I and the excavation reports cited there. Tatoup is in form also a theophoric name, but I do not know its derivation.

DOI: 10.4324/9781003556732-3

Ill 122

MISSING FEMALES IN ROMAN EGYPT

Kellis which has been excavated, and several other texts from the excavation concern him; the sellers lived, as I have mentioned, in a nearby hamlet but used Kellis as their market center and are otherwise unknown to us. At least one of Tithoes' children bore a distinctively Christian name, suggesting that Tithoes himself was probably Christian. Whether he was born into that faith or converted, however, we do not know. The sellers, who execute the contract, are stated to be illiterate, and the contract is written in a competent documentary hand by one Timotheos son of Harpokration, a former magistrate — presumably of Mothis, the local nome's capital city. The slave is described carefully as 'the slave girl belonging to us, picked up from the ground (xct|iaLp€Toc, an otherwise unattested Greek word),4 nursed by me the aforementioned woman with my own milk.' She is not, however, given a name in the document, and the comparatively low price of two solidi suggested to the editor that she must have been 'little more than a toddler at the time of the sale.' Presumably, however, she was old enough that the purchasers thought she had a reasonable chance of surviving; two solidi, although a low price for a slave, was enough to support a small family for a year, and one would not want to risk it on a baby still likely to suffer from the enormous infant mortality that this society experienced.5 The sequence of events indicated by the sale contract is important. A baby girl was exposed, left in the open, either in Kellis itself or in the small hamlet where the couple lived. We cannot tell which of these from the information given. They took her up and brought her to the point at which she is sold, perhaps a couple of years later,6 with the wife in the couple nursing the baby herself. The child was then sold to a relatively prosperous artisan in the local center for what would have been quite a lot of money to most Egyptian villagers — a reasonable return, they may have thought, for their personal involvement and perhaps their investment in earlier foundlings who had not survived infancy. Although juridically a village and not the metropolis of the nome, Kellis should probably be seen more as a small urban center than as a rural village. It had large houses, the residences of important landowners, a culture that included Greek and Coptic literary texts (both classical and Manichaean), and monumental architecture — not the characteristics of ordinary Egyptian villages in this period.7 4

5 6 7

One usually finds dvaiperoc or KOTrpiaipcroc; cf. I. Biezunska-Maiowist, 'Die expositio von Kindern als Quelle des Skavenbeschaffung im griechisch-romischen Agypten', Jahrbuchfur Wirtschaftsgeschichte 1971/11, 129-33. A third dead by the age of a year, roughly. See R. S. Bagnall and B.W. Frier, The Demography of Roman Egypt, Cambridge 1994, 32-6, 151-3. Perhaps at not much more than a year, even. See below, Appendix I, for a discussion of the economics of the business. See on this point R.S. Bagnall, Egypt in Late Antiquity, Princeton 1993, 310-9.

Ill 123

A building complex most of which is yet to be excavated is apparently even larger than what has been brought to light so far. As I remarked earlier, some of the phenomena we see here have long been described as normal for the Roman Empire. A recent article by William Harris on 'Child-Exposure in the Roman Empire' has argued that exposure was relatively common, that more girl babies than boy babies probably were exposed, and that exposure was an important source of slaves.8 Harris is, however, reluctant to see any overall pattern in which such exposure has a significant effect on the demographic structure of society, particularly on the sex ratio. It is precisely such a pattern that I am going to try to elucidate. Strictly speaking, I shall be talking about evidence from Egypt, and one could argue — as has often been claimed — that this province was peculiar. But this claim has suffered significant blows in recent years and to my somewhat biased view no longer seems even intellectually respectable.9 For most demographic purposes, I think it fair to say that 'the basic demographic attributes of Roman Egypt are, at the least, thoroughly at home in the Mediterranean; they tend to recur in historical Mediterranean populations with considerable regularity. Nor is there any strong a priori reason why most of these attributes should be regarded as unique to Egypt among Roman provinces.'10 The attributes in question have been discovered by an analysis of the information provided by the census returns from Roman Egypt, from which we know some 1100 persons. This body of information is not without its difficulties, and I must sketch these briefly. The declarations are very unevenly distributed in space and time, coming predominantly from the second and early third century and from the Arsinoite Nome, today's Fayum province. These non8

9

10

JRS 84, 1994, 1-22; see 3-11 for the commonness of exposure, 4-5 and 11 for the prevalence of girls among those exposed, and 18-19 for exposure as source of slaves. Cf. also Harris's earlier note on the possibility of widespread exposure in CQ n.s. 32, 1982, 114-6. The 1994 article has extensive bibliography of other discussions of the subject, which I do not repeat here; for the papyri, however, see the remarks of I. Biezunska-Matowist, 'Les enfants-esclaves a la lumiere des papyrus,' Hommages a M. Renard II (= Collection Latomus 102), Brussels 1969, 91-6 and her article in Jahrbuchfur Wirtschaftsgeschichte 1971/11, 129-33, concentrating on the profitability of the raising of children as slaves (on which see Appendix I, below). See, for example, Dominic Rathbone, The Ancient Economy and Graeco-Roman Egypt', Egitto e storia antica daW Ellenismo all'eta araba. Bilancio di un confronto, ed. Lucia Criscuolo and G. Geraci, Bologna 1989, 159-76; A.K. Bowman and Dominic Rathbone, 'Cities and Administration in Roman Egypt,' JRS 82, 1992,107-27. Bagnall and Frier (n. 5), 173. T. Parkin, in his view of this book (BMCR 6, 1995, 88-98 at 94) quite reasonably views this statement as unproven, but he does not argue that it is wrong.

Ill 124

MISSING FEMALES IN ROMAN EGYPT

random characteristics, however, do not seem to have important consequences for the analysis. The sample is also unduly biased toward returns originating in cities, but this can be corrected by a simple weighting of numbers. More serious is a clear tendency for males to vanish in the years just before and around the age at which one became liable for capitation taxes, either by simple failure to register them or by their migration from their home villages to the more anonymous cities. We shall have to return to this distortion. Overall, though, the single greatest deficiency of our data base is simply that it is not large enough. For the most part, conclusions drawn from the entire population have proven to be fairly robust, but those drawn from subsets diminish in reliability very sharply as one moves to smaller numbers.11 My concern here is not with any effect that a general tendency to expose infants might have had on the demography of Egypt. If infants were exposed without regard for sex, and equal proportions of boys and girls died as a result, the only effect would have been to change slightly the shape of the fertility curve; exposure would in this case have been only a postnatal form of birth control. This actually does not seem to have been the case, however, for the fertility curve derivable from the census declarations shows the distinctive shape of a population in which no such control is exercised.12 The Egyptians used other means to prevent overpopulation, particularly breastfeeding and the failure of women to remarry after being widowed or divorced.13 My interest, rather, lies in what results differential exposure might have had: If it is true — as the commonplaces of the literary sources would suggest — that more girls than boys were exposed, what results would follow for the sex ratio in the free and slave populations, and can those results be identified in the data we have? And would the results even be visible against the backdrop of extremely high infant mortality? The first problem we must acknowledge is that a small differential probably could not be identified in our data. It should manifest itself in a higher than natural sex ratio for the free population: the sex ratio is generally expressed as 11

12

13

The problems that we faced with the Egyptian data are in many respects paralleled by difficulties in the study of the Florentine catasto: see David Herlihy and Christiane Klapisch-Zuber, Tuscans and their Families. A Study of the Florentine Catasto of 1427, New Haven 1985, chapters 5 and 6. But the size of their data set is enormously greater than ours. This has been shown in detail by B.W. Frier, 'Natural Fertility and Family Limitation in Roman Marriage', CP 89, 1994, 318-33. Where birth control (including postnatal) is used to control family size, the fertility curve drops off much more sharply than in populations where no such controls are in use, because these measures are usually adopted after families have reached the desired size, not in the early years of marriage. For a discussion of overpopulation as the major threat to the stability of ancient populations, see Frier, CAH XI, forthcoming.

Ill 125

the number of males per 100 females in the population, and in most modern populations it is around 105 at birth and declines subsequently toward 100 as the higher mortality rate of males affects it. We do find a sex ratio in the overall Egyptian data diverging sharply from this norm, namely a startlingly high 120.4.14 But two caveats are needed. First, that ratio comes from a body of material in which the nome capitals are overrepresented compared to the villages. These metropoleis, as they are called, show a much higher sex ratio than the villages — where the ratio is only 88.2, more females than males — do, and when we correct for this bias, the ratio drops to a less dramatic but still high 111.4.15 Second, the ratio is not consistent through the period for which we have declarations — the relatively few first-century declarations greatly underrepresent women. Nor does it remain constant through the lifespan — the reported sex ratio drops steadily as boys approach the age at which they become liable to taxation and only recovers thereafter. Its pattern contains enough swings to appear chaotic at first glance, but it shows that the male edge grows with increasing speed as the population ages. In all likelihood, the largest component of the early distortion results from the deliberate concealment of young village males approaching the age of taxation. Part — offsetting the preceding — also probably comes from the underreporting of very young females, especially in the cities. Part, too, comes from the tendency of random fluctuations to be greater in small samples of the evidence than they are in reality.16 At all events, it is very probable that the sex ratio in the free population was rather higher than natural levels, but the data are subject to enough 'uncertainties and biases' that a skeptical observer may not feel a great deal of confidence in our assessment of the sex ratio.17 A more detailed analysis of a limited body of the best-preserved declarations helps to clarify matters. For this purpose, only those returns with relatively young parents — 35 or under — with their children were used.18 In this way one gets some sense of what those families which were still pretty much intact, from which children had not yet moved out, looked like; only free persons are taken into account. The metropolitan families had more than twice as many sons as daughters; those from the village reported 34 daughters to 19 sons. Both of these inspire caution, even suspicion, for both underreporting of daughters (in the 14 15 16

17 18

Bagnall and Frier, Demography (n. 5), 93. Florence also had a much higher sex ratio than the Tuscan countryside and even than other Tuscan cities. Bagnall and Frier, Demography (n. 5), 103 shows the male age structure in the census declarations compared with that visible in tax lists from Fayum villages; the tax lists show less violent swings while displaying much the same shape graph. If we had more data, these swings would gradually diminish. So Parkin (n. 10). See Bagnall and Frier, Demography (n. 5), 152-3 for this discussion.

Ill 126

MISSING FEMALES IN ROMAN EGYPT

metropoleis) and concealment of sons (in the villages) seem very probable. They do, however, survive enlargement of the data to include all free persons with preserved ages: in the villages, females under 15 outnumber males 57 to 42, but in the metropoleis males outnumber females 62 to 33. We note, 'the lopsided juvenile sex ratio in villages is probably not significant, since it is implausible that village parents practiced active postnatal sexual discrimination in favor of daughters; concealment of sons is the more obvious explanation. By contrast, the metropolitan sex ratio for juveniles cannot be brought into balance even if large allowance is made for underreporting of very young girls.' Despite all of the deficiencies of our data, then, it seems very likely that a certain number of free female children in the metropoleis were eliminated. There are at least three significant methods by which this may have occurred, and all three may have been operative. First, female babies may have been exposed and died. Second, they may have been exposed and turned into slaves, like the object of the sale from Kellis. Third, girls may have been treated less well and fed less generously, resulting in an even higher mortality rate in infancy and early childhood than that from which boys suffered. The last of these, if it was operative, probably had most of its effects before the age of five, for the gap between boys and girls does not widen after that age. But we are unlikely to be able to tell the difference between the effects of exposure followed by death and death resulting from poor treatment at an early age; the robustness of the data from year to year in the population under five is simply inadequate. If the entire shortage of young females in the metropoleis was caused by death, either through exposure or through maltreatment, it would in principle have no impact on the sex ratio in the slave population. One might even expect to find that slave-born girls were also differentially eliminated, in which case the slave population would have a sex ratio like that of the free population, with males overrepresented.19 If part of the explanation for the missing females is exposure followed by enslavement, however, it ought to be the case that the slave population would be more female than could have resulted from births to slave women, as these should have yielded roughly equal numbers of males and females. Such a pattern offemaleness is in fact present. Our population sample in the census declarations includes a total of 118 slaves, or about 11 percent of the population. The declarations from the cities, which are about half of the total, include about 60 percent of all slaves. But, because village households are on average larger than metropolitan ones, slaves are 13.4 percent of the urban population and only 8.5 percent of that of the villages. Given the significant concentration of wealth in the nome capitals, this is not a particularly surprising state of things. Overall, slaves will have made up just 19

This seems to be the view of Harris, 'Child-exposure' (n. 8), 6, although he does not believe there is any evidence for a high sex ratio in the free population. Cf. the next note.

Ill 127

about exactly 10 percent of the total weighted population. This is in line with most estimates of this figure by other scholars. When we look at the sex ratio among slaves, however, the picture becomes more interesting. Men and women occur in roughly equal numbers in the servile population of the metropoleis, but in the villages women are overwhelmingly predominant, 36 women but only 6 men. It is possible that these numbers are somewhat unduly influenced by a few households, but it seems unlikely that the pattern would disappear with more evidence. Overall, women slaves outnumber men by two to one, 68 to 34. If we remember that the villages are greatly underrepresented in the data and correct the raw figures accordingly, the weighted population looks somewhat different, with femaleness enhanced. The total female to male ratio is now about 2.5:1 (104 to 40). But the great preponderance of women is changed only in degree. Why are there more women slaves? We cannot leap immediately to the conclusion that exposure is responsible. One immediately evident answer, indeed, is that there are no male slaves older than 32, and only one older than 29, whereas women continue to appear into their forties. It is impossible to escape the conclusion that women were often retained as slaves as long as they appeared to be fertile, whereas social expectations and economic decisions led to the manumission of men by about age 30. (That does not mean that there were not some male slaves over 30, only that they must not have been numerous.)20 In our weighted population, using only slaves for whom ages are preserved, there would be 69 women to 29 men; eliminating those over the age of 30, there would be 48 women to 28 men. The differential pattern of manumission, therefore, accounts for a bit over half of the surplus of female slaves visible in the raw numbers. But even after we restrict our inquiry to slaves aged 30 and under, women still outnumber men by better than three to two, and by nearly three to one in the villages.21 Although 20

21

There is not enough documentation of the ages of freedmen in the papyri to provide confirmation of this view. I. Biezunska-Maiowist, L'esclavage dans lEgypte grecoromaine II, Wroclaw 1977, 146 notes that despite the lack of specific ages preserved, freedmen generally seem relatively young and certainly active, a sign that they were manumitted when a considerable part of their working life lay ahead of them. She also points out (p. 145) that some 60 percent of attested freedmen are male, although the statistical value of this figure may be limited not only by the quantity of data but by the fact that men's activities are more likely to have generated documentation. Harris (n. 8), 6 remarks that 'an unbalanced sex ratio probably did prevail in the population of slaves, and one of the mechanisms by which this was brought about was perhaps the selective exposure of girls who were born to slave mothers.' He thus apparently supposes that the sex ratio among slaves was skewed toward masculinity, the reverse of what the data show. This view is presumably at the root of

Ill 128

MISSING FEMALES IN ROMAN EGYPT

births to slave women were probably the largest source of slaves in Roman Egypt, then, there must have been some other source to have yielded the remainder of the young female slaves. Before we go on to pursue the implications of this statement, we ought to ask how vulnerable it is to attack. Its main weakness is simply the rather small size of the population under study and the very serious possibility that the results are not statistically reliable. Until such time as new evidence allows a considerably larger number of households to be described, particularly in villages, direct improvement of the situation is impossible. But the essential point is the numerical preponderance of females in the slave population. If this could be confirmed from other sources, the trustworthiness of this essential link in the argument would be greatly strengthened. Sales of slaves are one such source of information. The tabulation of these sales (for the period up to the end of the third century) in Hans-Joachim Drexhage's recent book on prices22 shows 58 female and 41 male slaves. If only those with preserved ages are counted, the gap is 41 to 22. For slaves under 30, females outnumber males 37 to 15.23 (Curiously enough, more sales of males over 30 than of women are found, but the numbers are small and probably reflect some circumstances we cannot determine.) Not too much reliance should be put on the exact ratios; if these are open to doubt in the case of the census declarations, they are that much more doubtful here and will no doubt change as more data are published. But the central tendency of the data is the same and is not in doubt. All of the femalermale ratios in both sets of data fall in the range from 3:2 to 5:2. Such numbers are consonant with the general conclusion reached by modern scholars that slaves of working age in Egypt were owned primarily for personal use, not for economic exploitation: they were, in other words, items of consumption and not of investment.24 If, then, we need a source of young female slaves other than birth to a slave mother, it is hard to see what this source can have been except the exposure of free female infants unwanted as members of their birth families but valued by those who rescued them from death as potential salable assets, exactly what we

22 23

24

his failure to see some of the connections for which I am arguing here, which otherwise fit rather well with his views. Preise, Mieten/Pachten, Kosten und Lohne im romischen Agypten, SanktKatharinen 1991,271-9. Drexhage (n. 22), 254. The mutability of such numbers may be seen by looking back at the statement of I. Biezunska-Matowist (n. 20), 145 that men and women occur with roughly even frequency in sales; that statement rested on a much older list published by O. Montevecchi. See generally I. Biezunska-Matowist (n. 20), 73-108; some slaves were employed in agriculture, few in artisanry.

Ill 129

see in the Kellis sale.25 And there is other evidence that the Kellis sale is not an isolated phenomenon. The most salient is the corpus of contracts for hire of a wet-nurse, which have been collected and studied by Mariadele Manca Masciadri and Orsolina Montevecchi.26 Of the 31 infants put out for paid nursing of whom we know the legal status, 23 were slaves, or three-quarters. For 19 of those 23 slaves, we are given information about the way in which they became slaves: 12 of 19 were picked up from exposure, and only 7 are described as something else, mainly engona, i.e., born in the household. And they are predominantly female. Slaves also account for a high percentage of infants put out to nursing in contracts that have not survived but are listed in summary form in the registers of the record office of Tebtunis.27 There is no point in pushing the exact percentages here, but the evidence is certainly consonant with our hypothesis that exposure of girl babies was a major source of slaves and, specifically, of a disproportionately female slave population. Assessing the impact of enough exposed girls to account for these surplus female slaves is tricky, and given the problems in our data base an attempt at great precision is probably not particularly valuable. A rough calculation, however, might go as follows: The total weighted population of females under age 30 in the declarations would have been 326; of these, 48, or 14.7 percent would be slaves. Arriving at a comparable figure for males is very difficult because of what appears to be the concealment of village males, to which I have already referred. If for the sake of hypothesis we imagine that the sex ratio for the under30 population in villages was really 100, we would obtain a restored weighted population of 359, of whom only 28 were slaves, or 7.8 percent. Applying that percentage to the female population would have yielded a slave population of about 26, rather than the actual 48. This result then indicates that 22 additional females in this sample population had been enslaved beyond the number that one would expect. Once again, this figure leaves out of account any undifferentiated impact of exposure. For all we know, a considerable part of the male slaves may come from exposure as 25

26 27

One might of course hypothesize large-scale importation into Egypt of slave girls from elsewhere, but this is most unlikely: (1) It would have had to begin with infancy to produce the pattern we see; (2) there is no evidence for such a trade in the documentation; and (3) such a hypothesis only displaces the problem from Egypt to some other place. The other possibility is sales of infants by their own parents, which in a sense would be another means of accomplishing the same general transfer of females from the free population to the slave that exposure produced. Just such a pattern, at least in times of economic stress, has been suggested by M. Manca Masciadri and O. Montevecchi, / contratti di baliatico (= Corpus Papyrorwn Graecarum I), Milan 1984. / contratti di baliatico (= Corpus Papyrorum Graecarum I), Milan 1984. See below, Appendix II, on the problems connected with the interpretation of these registers.

Ill 130

MISSING FEMALES IN ROMAN EGYPT

well, but that is invisible to us, and I am considering here only that portion of the female slave population in excess of the male. Those 22 females are 6.7 percent of the total females in the under-30 bracket. If that represented the percentage differentially exposed, the result would be — and this is both assuming an inherent move toward 100 and leaving out of account all other sources for skewing of the sex ratio — a sex ratio for the free population of about 107. If, on the other hand, we assume that preferential treatment of male children cancelled the normal drift of the sex ratio from the 105 found at birth down towards 100, a sex ratio among the free of about 113 would result. At this point we may recall that a weighted sex ratio of 111.4 for the free population is actually found in the declarations. This number does not, however, add back in some village males to compensate for underreporting, as we have just done. When this is done, the result should be about 115. That would be compatible with an overall ratio for the population somewhere in the range of 108, a figure that we arrived at on other grounds as a reasonable approximation.28 The reader may at this point feel that I have pushed the numbers entirely too far. But I think that one further point about the Egyptian situation is worth making briefly. This is that female slaves in the villages are also on average younger than those in the cities (mean age of 18 in the village vs. 22 in the city). This fact also is perfectly compatible with the hypothesis that villagers made a practice of taking in exposed infants from the metropoleis, reared them, and eventually sold some of them back to urban residents.29 This hypothesis would, of course, help to explain why the missing young females are a facet of the metropolitan figures but not of the village numbers. In this way the cycle would have been completed. Metropolitans would be rid of unwanted daughters, shift the risk of rearing them as slaves to villagers for whom the costs were low, and then buy them back once they had passed infant mortality, were no longer recognizable, and might even be starting to be useful as servants. There are obvious difficulties remaining. In the absence of more than approximate sex ratios, we cannot tell if some allowance needs to be made for exposed babies who died rather than being enslaved. We cannot be certain if some of the missing males in the villages are actually in the metropoleis, registered 28 29

Bagnall and Frier, Demography (n. 5), 95; cf. 108 for reluctance to argue strongly for anything except a range of 100 to 110. The number of wet-nursing contracts for slaves collected from exposure is not large enough to verify or disprove any hypothesis about the ownership of the slaves being reared in this fashion. If one supposed that villagers operating on their own account usually did their own nursing (as in the Kellis contract) and metropolitan entrepreneurs usually hired nurses, one would at least find some support in CPGr. I 15 and 24; but in most of the relevant contracts some element of the needed data is lacking.

Ill 131 apart from their birth families. The examination of young families, however, with their surplus of males over females, suggests that the high metropolitan sex ratio exists independently of migration and cannot entirely be explained by it. At this point it seems useful to try to provide a wider perspective for the patterns that I believe can be seen, even if imperfectly, in the Egyptian data. The first such context is that of Ptolemaic Egypt. We have a significant body of census-related accounts from the third-century B.C. Arsinoite Nome, which are currently being edited or reedited by Willy Clarysse and Dorothy Thompson; I am indebted to them for the use of some of their data before publication.30 The figures we have are from a papyrus dating between 253 and 230 and are almost complete for adults, omitting only a small number of soldiers. They do not, however, include children, an important omission from our point of view. At least one may note that they concern the same region from which come most of our census declarations in the Roman period. The families of members of the army, both military settlers and those on active duty, show a sex ratio of 115.8; given the masculine character of military service, this is perhaps not surprising. The civilian population, however, shows a sex ratio of 90.9. This is remarkably low, but it will not escape notice that it bears a certain resemblance to the village figures for the Roman period. It is of course possible to argue that concealment of males is again at work in the Ptolemaic data, but Ptolemaic capitation taxes, although bewilderingly numerous, were far less burdensome in amount than Roman taxation, and the argument is thus less persuasive. Also interesting is the fact that within the civilian population there is a distinction between those persons legally designated as Hellenes and those classified as Egyptians. These categories are ones of legal status, and many — perhaps most — of those legally listed as Hellenes were of varieties at which most classical Greeks would have turned up their noses: Macedonians, Thracians, Paeonians, Jews, and so on.31 Still, their sex ratio is 97.5, while that of the Egyptians is 88: identical to the sex ratio in village returns of the Roman period. But matters are more complex. We also have a number of census lists giving not aggregate numbers but the members of individual households. When these are classified and counted up, they yield a sex ratio of 105. To be sure, the numbers involved are much smaller than the aggregate nome figures in the account cited earlier. But they are still puzzling. I do not think that the lists of names include children under 14, either, so a much different sex ratio under the age of 14 cannot be the explanation. All in all, however, the Ptolemaic evidence makes me wonder if our belief that the Roman village sex ratio is to be ex30 31

To appear as Counting the People in Collectanea Hellenistica (Leuven). On this matter see my article on The People of the Roman Fayum', in Portraits and Masks: Burial Customs in Roman Egypt, ed. M.L. Bierbrier (London, British Museum, forthcoming in 1997).

Ill 132

MISSING FEMALES IN ROMAN EGYPT

plained by concealment of teenaged males is entirely justified. Even the 'Hellenes' have a much lower sex ratio than seems probable for the Roman period. It should be remembered that nearly all of the population in the Ptolemaic period lived in villages; the metropolis of the nome had not yet become a significant city. It is therefore at least possible that, even with some allowance for underrepresentation of males, the figures for Egyptians in the Ptolemaic period and for villagers in the Roman period together point to an underlying demographic reality very different from what the population of the Roman period as a totality experienced. The data from the Florentine catasto of 1427 provide another interesting point of comparison. The sex ratio graph by age is remarkably similar to that for Roman Egypt as far as the mid-forties, after which they diverge very considerably.32 Very high (120 and above) ratios for ages under 10 drop toward equality as the age of tax liability approaches; the ratio then rises to a new peak in the mid-twenties, only to drop off again into the middle thirties (this may be only a blip), after which it again rises. In the case of the Florentine population, the incidence of plague needs to be taken into account, but still the similarity is striking. So too are the explanations offered, including nonreporting of those who died young and devaluation of female infants.33 And Tuscans also had incentives to conceal young men reaching the age of liability for the head tax, or at least to make sure that girls — whose gender had hardly mattered until now — were reported as such.34 Despite all of these distortions in reporting, however, the authors of the principal study of the Tuscan data conclude that 'it is impossible to believe that the recording of women would have been uniformly poor at nearly every level of life. Rather, it appears incontrovertible that the Tuscan population was marked by a true deficit of females. Social factors of some sort must have deprived girls of their normally better chances of survival.'35 They go on to canvass particularly infanticide and abandonment, the latter of which led to being a foundling, left to be raised by charitable institutions. Girls made up 70 percent of the foundling population.36 We may be reminded that this is about the same percentage that females formed of the slave population of Roman Egypt, according to the census declarations (72.2, to be precise, in our weighted population). Still more striking is the fact that 'wealthy Florentines (and Tuscans) declared in their households substantially fewer women than did the poor, and ur32 33 34 35 36

Herlihy and Klapisch-Zuber (n. 11), 133, 143: Egypt shows a sharp rise in the sex ratio, Tuscany a much gentler one. Herlihy and Klapisch-Zuber (n. 11), 135-44. Herlihy and Klapisch-Zuber (n. 11), 138. Herlihy and Klapisch-Zuber (n. 11), 144. Herlihy and Klapisch-Zuber (n. 11), 145.

Ill 133 ban families also showed fewer females than do homes in the countryside.'37 As to the latter, the difference (112 urban vs. 109 rural) was not as sharp as in Roman Egypt; but for Florence itself the ratio was 117.6 vs. 108.9, and there was considerable variation from town to town and district to district.38 In part this difference reflects migration, a factor we have supposed played some role in Roman Egypt as well, but in part it was a matter of differential nutrition, abandonment, and infanticide. The Tuscan data certainly remind us just how variable by local circumstances these matters must have been, and they may make it less surprising that we find in the Ptolemaic Fayum a discrepancy between overall numbers and the lists for particular villages. They are also useful in showing that similar patterns in the population figures may have roots in a mixture of similar and different causes. Roman Egypt had no foundling hospitals or nunneries, and young children were not put out in the households of others as free servants. Although slavery existed in Tuscany, it was a negligible factor. Tuscany may, however, have been almost as urbanized as Roman Egypt. What assessment of the situation in Roman Egypt are we to offer, then? All of the data to which I have called attention may point to a particular pattern of treatment of females in the society of Roman Egypt: less care in upbringing than for males, shorter life expectancy, higher rate of exposure as infants, prolonged time in slavery, maximum exploitation as breeders of slaves. This pattern seems particularly connected with the cities, but the villages are directly implicated in it by reciprocal roles in the cycle, acting as collectors of exposed infants, and as rearers and sellers of slaves. We have become accustomed to a picture of Egyptian society as far less male-dominated and less prone to subjugate women than was normal in classical Greece. Egyptian women were, already before the Hellenistic period, able to own property and to marry and divorce by their own decision. I think it is fair to say that the common wisdom among historians of Hellenistic and Roman Egypt is that the Greek immigrants into Egypt in the Ptolemaic period absorbed and appropriated much of the juridical and social freedom ascribed to Egyptian women. It is possible that the low sex ratios we have seen in the village populations of both Ptolemaic and Roman periods reflect an underlying Egyptian regime which even after all allowances for faulty reporting shows a demographically more favorable situation for women than in, let us say, the pre-1960s rural population of China, where excess female mortality in infancy ranged from 5 to 15 percent.39 For the Roman period, the melioristic process is usually assumed to have continued, benefiting from women's ready ownership of property, including 37 38 39

Herlihy and Klapisch-Zuber (n. 11), 151. Herlihy and Klapisch-Zuber (n. 11), 156-7. See Ansley J. Coale and Judith Banister, 'Five Decades of Missing Females in China,' Demography 31, 1994, 459-79.

Ill 134

MISSING FEMALES IN ROMAN EGYPT

land, and freedom of divorce under Roman law. This view has become characteristic of scholarship based on the papyri. Sarah Pomeroy, for example, has reached the 'inevitable conclusion' that 'under Roman rule women gained in economic and legal capacity.'40 How are we to reconcile this picture of economic and legal gains with the darker demographic view sketched here? It would far exceed the scope of a single paper to attempt a general assessment of the situation of women in Roman Egypt. The explanation or explanations of the higher rate of infant exposure of females, and of their upbringing as slaves, may well be complex. But two points in the evidence seem to me to suggest the direction in which we ought to look. The first is that the Ptolemaic situation seems to have been very different; it is only with the Roman period that exposure begins to be a significant phenomenon. The second is that it is mainly an urban phenomenon, remaining foreign to the Egyptian population of the countryside. These two facts suggest that we are looking here at one of the consequences of the more systematic imposition of Graeco-Roman patterns of the organization of social and economic life that we find in Egypt from the reign of Augustus on. Paradoxically, the Ptolemaic period saw relatively little development of characteristic Greek institutions in the Egyptian countryside, above all because the Ptolemies created few Greek cities in Egypt. It is unfortunate that the Ptolemaic records lack individual ages, making it impossible to calculate the age-specific sex ratio, an important tool in reconstructing the total picture. But the fact that our information is also based on nearly complete figures for one nome, amounting to more than 58,000 individuals, in large part outweighs such regrets. These figures probably indicate that women in this Ptolemaic population had slightly higher life expectancies than men. The Roman period, on the other hand, saw from the beginning a development of the metropoleis of the nomes as Greek cities with a privileged class of landowning notables, whom the Romans treated juridically as Egyptians but fiscally as a group apart.41 By the third century this class was ready to become a full-fledged curial order like those in Greek cities throughout the Roman East. In keeping with this pattern, the Roman period in general saw a growth in the hierarchical relationship of city and countryside, and I do not think it is farfetched to see the differential demographic patterns I have tried to describe as a part of that transformation. It is entirely possible — and the Florentine data would suggest very strongly — that the urban elites, followed to a lesser degree by the rest of the urban population, showed significantly different patterns of childrearing; exposure may have been both a strategy of the rich to minimize the 40 41

'Women in Roman Egypt,' ANRWII 10.1, 1988, 708-23 at 723. See Bowman and Rathbone (n. 9); J. Meleze Modrzejewski, 'Entre la cite et le fisc: le statut grec dans 1'Egypte romaine,' Symposion 1982, Valencia 1985, 241-80 = Droit imperial et traditions locales dans I 'Egypte romaine, Aldershot 1990, chapter I.

Ill 135 partition of wealth and a response of the poorer residents of the cities to the survival pressures they faced. Greater differentiation between rich and poor was characteristic of the Roman empire in many respects, of course, and it would not be surprising to see it operating as well in the realms of life, reproduction, and death. In any event, we should probably be cautious about assuming that the role of women in the transmission of property that we see in the papyri of the Roman period actually represents their well-being; it is entirely possible, in my view likely, that the Ptolemaic period was more favorable for women — and perhaps for some other less privileged social groups — than was the Roman.42 We historians may still be too prone to our desire to like the people we study, and the people of the Roman papyri, especially the members of the propertied classes who produced most of the papyri we read, are perhaps just a bit too sympathetic to our own middle-class tastes. Lurking in our all-too-imperfect data are signs that stratification of wealth and privilege in Roman Egypt may have been linked to a deterioration of the daily lives of those left behind. To an American, familiar with the differences in such indexes of social welfare as infant mortality between the United States and those countries with a higher rate of income redistribution, this pattern cannot be very surprising. Appendix I: The Economics of Bringing up Exposed Infants Diodorus Siculus famously tells us that bringing up children is so cheap in Egypt that the Egyptians do not expose any of their children. The cost for Tatoup to nurse their foundling was presumably minor — a bit more food for her to eat — and the actual cash invested in the young slave was therefore small. But it is evident from the wet-nursing contracts discussed above that not all slave infants were nursed 'free' by an owner. Manca Masciadri and Montevecchi (/ contratti di baliatico 12) express surprise also at the fact that the engona, children of slave mothers, are not being nursed by their mothers; surely, they suppose, we would not expect so many mothers to have died or be dry. They therefore suspect that the designation might be used to cover some other reality (sale of children by their parents, they think; cf. Appendix II). But it must be remembered that allowing the slave mother to nurse the child might have delayed her becoming pregnant again, and to the extent that one owned female slaves partly for breeding of slave children, such a delay would be undesirable. It may have been more profitable to pay someone else to nurse the child than to sacrifice some of the reproductive potential of the slave. 42

See J. Meleze Modrzejewski, The Jews of Egypt from Ramses II to Emperor Hadrian, Philadelphia 1995, 161-83 on the degradation of the status of the Jews under Roman rule.

Ill 136

MISSING FEMALES IN ROMAN EGYPT

In either case, the question arises: How much did it cost to raise an infant taken up from exposure and turn it into a salable slave? And was it still profitable to take up exposed infants if one could not have them nursed without cash outlay? Opinions have been expressed on the subject (cf. above, n. 8, for I. Biezunska-Maiowist's repeated assertion of profitability), but without even a rudimentary analysis. For present purposes I use figures from the first century A.D. from the chora. In Alexandria, both wages and the price of slaves were higher; but the documentation is also earlier in the Roman period (from the reign of Augustus) than what we have from the Arsinoite and Oxyrhynchite nomes (Tiberius and later), so the difference (Alexandria is something like 40 percent higher) may not be explained entirely by location. The average cost of a hired nurse was 6 drachmas per month; almost all values fall in the range 5-7.43 After age 2, when nursing was no longer required, the cost of support presumably fell; I have assumed that it was half the expense of nursing, which is probably not off the mark by a large amount. If we use the model life table for women (Level 2, West) argued in Demography of Roman Egypt to be the closest representation of Egyptian experience, we find approximately the following cost for a cohort of 100 infants collected from exposure at birth:44 Age 1 2 3 4 5

Survivors 67 62 58 54 50

Year cost 6,012 dr. 4,644 dr. 2,160 dr. 2,016 dr. 1,872 dr.

Cumulative cost 6,012 dr. 10,656 dr. 12,816 dr. 14,832 dr. 16,704 dr.

Unit cost 90 dr. 172 dr. 221 dr. 275 dr. 334 dr.

The cumulative unit cost per survivor to a given age is shown in the righthand column. A large operator would need to exceed this figure to make money. Obviously not many, if any, operators actually put cohorts of a hundred babies out to nurse even over quite a few years; individuals with one or a few nurselings would have had fewer over whom to spread the risk of death and would thus probably have needed a higher premium for risk. After the age of 5, when mortality became a smaller factor, the unit cost rose by the actual cost of support for each child. The number of surviving slave prices for children under the age of 10 is not large, and one should be cautious in using them. But at these ages skills do not 43 44

All data here derive from the information given in the corpus of Manca Masciadri and Montevecchi. It is assumed that infants dying during a given year live for exactly half of that year; that may not have been true in the first year, in which case the cost figure is somewhat too high.

Ill 137 play the part they may have later, and children are thus probably more fungible than adults. At all events, the prices collected in Drexhage, Preise 265, show the following (omitting one outlier): Age 2-3 7 8 9

Price range 300 dr. 500-700 dr. 600-1200 dr. 1600 dr.

It looks as if anyone who could afford to take risks had a fair chance of making a good profit margin on a child whose acquisition cost was zero. We have no evidence what the market value of a neonate was; not high, I would judge, or exposure would have been uncommon; but it might have been as much as 100 dr. Appendix II: Exposure or Sale in the Tebtunis Register? Manca Masciadri and Montevecchi have argued45 that the numerous contracts for nursing infant slaves listed in the registers from Tebtunis, which give summaries of them, are best interpreted not as showing exposed infants being put out to nurse but as witnessing to a practice of sale of neonates by their financially distressed parents. Although I see no way of excluding this possibility, I also see no grounds for this hypothesis. The view that cxvaipeTOi generally were not really exposed infants but children sold by their parents had in fact already been proposed by B. Adams; Manca Masciadri and Montevecchi say Tipotesi e gratuita' (149). So it is; but they go on to offer a virtually identical hypothesis for the Tebtunis registers. Their reasoning is essentially as follows (154-7 of the article cited): It is incredible that a master would pay for a nurse for the child of his own slave, so we may exclude the notion that these slaves (whose origin is not given in the summaries) are engona. But the Egyptians rejected the entire notion of exposure, being firm believers in the value of life. Tebtunis was a largely Egyptian milieu, and its Greek population cannot have produced such a large crop of exposed babies for Egyptians to pick up. Therefore the slaves can be neither engona, nor the exposed children of Egyptians, nor the exposed children of Greeks. The only plausible hypothesis is therefore that they were sold by their own parents as a result of financial distress. Because a large portion of the contracts date from years 6 and 7 of Claudius, which we know to have been hard times in Egypt, the oc45

In summary in the introduction to / contratti, in more detail in Aegyptus 62, 1982, 148-61.

Ill 138

MISSING FEMALES IN ROMAN EGYPT

casion for such distress is known. The mechanism used is likely to have been the civf) e v mcjTci, the sale subject to redemption, where the child becomes the property of the creditor at the end of the loan period if it cannot be paid back. This argument is not solid.46 First, we have already seen that a master might indeed have good reason to prefer to have his slave woman get pregnant with the next slave child rather than nursing, with the partial protection against pregnancy that nursing provides. Indeed, if the master was the biological father of the slave children, he might not wish to have the slave woman observe the sexual abstinence during nursing normally called for in the nursing contracts. Second, the view that there would not have been an adequate supply of exposed infants in Tebtunis takes it for granted that all of those raised in Tebtunis were exposed there. There are no grounds for such a view; they could quite well come from Arsinoe. Once one admits the possibility that urban infants who had been exposed might have been taken to villages to be nursed, indeed, the entire argument about the Egyptians' cultural preferences becomes moot. On the juridical side, the argument also seems to me suspect. The child is already treated as a slave of the hiring party in the nursing contract, if we may judge from our brief summaries; this would not already have been the case unless a sale contract had intervened, and no such sale contract is ever recorded in the register. Nor is it obvious where in the two contracts that are recorded (nursing contract and loan agreement, the latter of which made the wages paid in advance to the nurse more readily recoverable if necessary) a provision for redemption of the child would have been found. Certainly none of the surviving contracts has any such provision. It may well be true that there is a concentration of these contracts in the period of economic distress, but that is no argument for infant sale. It can just as well point to an increased likelihood of abandonment of unwanted infants in the nome metropolis or elsewhere, and to a need for funds that would have made village women readier to take on wet-nursing jobs.

46

It is accepted without discussion by D.W. Hobson, EMC 28 = n.s. 3, 1984, 373-90 at 378-9.

IV

Church, State and Divorce in Late Roman Egypt

A recently published papyrus from Oxyrhynchos1 contains a petition from one Aurelia Attiaina to Flavius Marcellus, tribune and officer in charge of the peace.2 She recounts that A certain Paul, coming from the same city, behaving recklessly carried me off by force and compulsion and cohabited with me in marriage. [She had a female child by him and lived with him in her house. He behaved badly (the context is damaged) and cohabited with another woman. Then] after some time again he beguiled me through priests until I should again take him into our house, agreeing in writing that the marriage was abiding and that if he wished to indulge in the same vile behaviour he would forfeit two ounces of gold, and his father stood surety for him. I took him into our house, and he tried to behave in a way that was worse than his first misdeeds, scorning my orphan state, not only in that he ravaged my house but when soldiers were billeted in my house he robbed them and fled, and I endured insults and punishments to within an inch of my life. So taking care lest I again run such risks on account of him, I sent him through the tabularius a deed of divorce through the tabularius of the city, in accordance with imperial law. Once more behaving recklessly, and having his woman in his house, he brought with him a crowd of lawless men and carried me off and shut me up in his house for not a few days. When I became 1. P.Oxy. L3581. 2. This paper grows out of a portion of a lecture given at Bar-Han University (Israel) and at the Hebrew University of Jerusalem in May, 1986. I am grateful to Ranon Katzoff for the invitation to Bar-Han, which prompted its writing, and to the audiences at both institutions for their comments.

DOI: 10.4324/9781003556732-4

IV pregnant, he abandoned me once more and cohabited with his same so-called wife and now tells me he will stir up malice against me. Wherefore I appeal to my lord's staunchness to order him to appear in court and have exacted from him the two ounces of gold in accordance with his written agreement together with such damages as I suffered on his account and that he should be punished for his outrages against me. Scarcely an edifying tale. The document has, alas, lost its dating formula, and the editor dates it to the fourth or fifth century on the basis of the handwriting.3 As the editor points out, such a date makes the text of interest for the question of the interaction of church and state in the restricting of divorce in late antiquity. That in the fourth to sixth centuries both civil and ecclesiastical authorities sought to discourage divorce has long been known; but the nature of these attempts, their relationship to one another, and their effects in practice have all, I think, been misunderstood. First, the civil restrictions. The Roman law of marriage had for centuries allowed both divorce by mutual consent and unilateral declarations ending marriages, with hardly any restrictions,4 until Constantine introduced new regulations in 331 (CTh 3.16.1): women were forbidden to send notices of repudiation to their husbands on "slender" grounds such as that they were drunkards, or gamblers, or philanderers. Wives could use only homicide, sorcery, or tomb destroying as grounds; otherwise, they would be deprived of property and exiled to an island. A man could claim only adultery, sorcery, or procuring, on penalty otherwise of restoring the wife's dowry and not being able to remarry —and if he did remarry, his former wife could seize his house and the dowry of the new wife. None of the restrictions thus actually invalidated either the divorce or the new marriage. Constantine's legislation lasted three decades. Julian seems, to judge from a passage of 'Ambrosiaster', to have cancelled Constantine's penalties for forbidden repudiation:5 "Ante luliani edictum, mulieres viros suos dimittere nequibant. Accepta autem potestate, coeperunt 3. The reference to priests, presbyteroi, guarantees in any case a date after the church in Egypt came above ground in the second decade of the fourth century, after the edict of toleration of 313. 4. F. Schulz, Classical Roman Law (Oxford, 1951), p. 134, notes that Augustus forbade remarriage by a freedwoman who divorced without his consent a husband who was also her patron. 5. See M. Kaser, Das Romisches Privatrecht, 2 ed. (Munich, 1971), pp. 175 ff. The passage is 'Augustine', Quaestiones de utroque testamento mixtim 115 (PL 35: 2348-49). 42

IV facere quod prius facere non poterant; coeperunt enim quotidie viros suos licenter dimittere." One fragment of this law, dated 363, apparently survives in the Theodosian Code,6 but we do not have its text for the part which concerns us here. 'Ambrosiaster', to be sure, misleads in suggesting that Julian changed a situation existing from time immemorial, rather than merely restoring longstanding Roman custom, but his purpose in this argument did not make exact legal history necessary or, perhaps, desirable. So far as we know, it was not until 421, long after Julian's death, that restrictions were again imposed. In CTh 3.16.2, Honorius imposed a new set of distinctions. Repudiation by the woman without cause led to the loss of marriage gifts and dowry and to her deportation, as well as to denial of the right of remarriage; repudiation for ordinary traits of bad character led to the same financial penalties and denial of remarriage, but not to deportation; and repudiation for serious crimes allowed her to remarry after five years had passed. Husbands' repudiations were divided in the same three classes, but with somewhat different penalties: groundlessly, the same penalties as for the woman who divorced on grounds of bad character (and the repudiated woman would be allowed to remarry after a year); on bad character, he returns the dowry, recovers his gifts, and can remarry after two years; on grounds of criminal behavior, he keeps all of the property and can remarry immediately. Eighteen years later, Theodosius (in NovTheod 12, 10 July 439) ordered that the dissolution of marriage required the sending of a repudium.1 He went on, however, to say, "But in sending a notice of divorce and in investigating the blame for a divorce, it is harsh to exceed the regulations of the ancient laws. Therefore the constitutions shall be abrogated which commanded now the husband, now the wife to be punished by the most severe penalties when a marriage was dissolved, and by this our constitution we decree that the blame for divorce and the punishments for such blame shall be recalled to the 6. CTh 3.13.2 (363); this passage deals with dowry and marital gifts. 7. The reference to imperial law in P.Oxy. 3581 is clearly tied to the notion of repudium: "I sent him through the tabularius a repudium through the tabularius of the city in accordance with imperial law." This reference seems to me to refer most obviously to the requirement of a repudium stated in the legislation of 439 and repeated in 449. (Coles in his note to line 16 cites P.Stras. Ill 142 [a divorce document of 391, discussed below] as evidence for Julian's repeal of Constantine's legislation, thus allowing a fourth-century date for this papyrus. But, as we shall see, the Strasbourg papyrus reflects a divorce by mutual consent, not affected by either Constantine or Julian, and thus is irrelevant to the date of the Oxyrhynchos papyrus.)

43

IV ancient laws and the responses of the jurisprudents." This law of Theodosius followed by little more than a year the appearance of his Code, which included the law of Honorius (issued in the name of Theodosius and Constantius as well as himself) quoted above, which it appears to contradict. Ten years later, however, Theodosius (CJ 5.17.8) returned to a more specific regulation of the causes for divorce which could be cited in a repudium. This time the sexes are treated more equally, for a woman has a wide range of causes allowed, including the husband's adultery, criminality, or violence toward her. Justified repudiation allowed the woman to recover her property and remarry after a year, a man to remarry immediately. Unjustified repudiation subjected the woman to a five-years' wait, the man to forfeiture of the dowry. Anastasius later (CJ 5.19.7, of 497) allowed the woman to remarry after a year if a groundless repudiation was by common consent.8 These meanderings of imperial legislation have received quite differing interpretations from modern scholars. It is neither possible nor useful to cite all of the extensive bibliography here; two examples of imprecision by great scholars will suffice. H.J. Wolff 9 describes Constantine's legislation as a "rigid restriction of justified divorce to a few cases of grave offense," and regards the legislation of 421 and 449 as a continuation of its development, while the novel of 439 is a "shortlived return to classical principles." A.H.M. Jones10 claims that "since divorce under the old legal forms had been rendered so difficult, many couples dissolved their marriage by consent. This was forbidden by Theodosius II in 439, but he at the same time abolished all the penalties for divorce, and went back to the classical law." It is true, of course, that not requiring grounds for termination of the marriage is classical, but the requirement of a formal repudium to accomplish this termination is not. On the other hand, it is surely incorrect to think that this requirement made divorce by mutual consent impossible. Anastasius' constitution, mentioned above, shows that that is not what 8. A.H.M. Jones, The Later Roman Empire, 284-602 (Oxford, 1964), p. 975, (= Jones, LRE) wrongly describes this law as ruling "that if a husband divorced his wife with her consent" she could marry after a year. Anastasius actually does not distinguish the sender of the repudiation, ordering only that it be sent by common consent ("Si constante matrimonio communi consensu tarn mariti quam mulieris repudium sit missum, quo nulla causa continetur"). 9. "Doctrinal Trends in Post-Classical Roman Marriage Law," Zeitschrift der SavignyStiftung fur Rechtsgeschichte, Rom. Abt. 67 (1950), 262. This article, occupying pp. 261-319, has very extensive bibliographical notes. 10. LRE, p. 974.

44

IV

he thought the law meant; and Valentinian's abrogation of the Theodosian novel of 439 (NovVal 35.11, of 452) suggests that that novel's burden for him was its permission for unilateral and unjustified repudiation, not any restrictions put on consensual divorce.11 Indeed, Justinian's legislation in the next century prohibiting divorce by mutual consent (Novel 117.10) says explicitly that it was not previously prohibited. To summarize the course of legislation: apparently Constantine's limits on the possible grounds for unilateral divorce were repealed by Julian some thirty years later, and for six decades there was no imperial legislation known to us on the subject. In 421 similar restrictions were reintroduced; in 439 they were removed in the East in favor of a requirement for written repudiation, a law accepted in the West in 448 ;12 but in 449 Theodosius changed his mind and again reintroduced restrictions (a bit more systematic and equitable this time), a course essentially followed by Valentinian in 452 with his return to the law of 421. Apart from Anastasius's clarification late in the century, the next major change did not come until Justinian tried to prohibit divorce by mutual consent in 542, a ban repealed by Justin in 566 (Novel 140). Many modern works, apart from failing to observe consistently the distinction between unilateral and consensual divorce and that between prohibiting and penalizing a divorce, go on to describe this legislation as the result of Christian influence. "Christian legislation allowed divorce only for certain reasons and instituted penalties in case of contravention," said Schulz.13 "The Christian empire, however, brought restrictions on divorce itself," said Thomas.14 Such statements will not do; the imperial legislation of the fourth and fifth centuries is not "Christian" legislation, and the empire was not a "Christian" empire, no matter how often it is called one. But underlying these remarks is a serious debate about the extent of Christian influence on the imperial legislation on divorce. Positions have ranged from the view expressed by Schulz and Thomas to the opposite extreme, a denial

11. Valentinian's novel orders that "those regulations which were decreed by our sainted father Constantius shall be preserved inviolate." Pharr's footnote: "Not extant, but cf. CTh 3, 16, 1." An error for citing 3.16.2? The latter law includes Constantius among the Augusti, in 421, and is surely what is meant. 12. As part of the general affirmation of Theodosius' laws after the Code, see NovVal 26. 13. Above (n. 3), p. 134. 14. Textbook of Roman Law (Amsterdam, 1976), p. 426. 45

IV of all Christian influence.15 Jones rather enigmatically remarked that "the Christian emperors tightened up the laws of divorce, but not in an entirely Christian sense."16 There is no doubt that the emperors of the fourth and fifth centuries did enact some laws which show a distinctively Christian viewpoint. A random example is the edict of Valentinian, Theodosius, and Arcadius (CTh 9.7.5) in 388 forbidding any marriage of a Jew with a Christian; such a marriage is to be considered adultery, and anybody is allowed to accuse the culprits. Many others could be cited. Nor was the church shy about asking for imperial legislation when it thought it needed; the title on heretics in the Theodosian Code (CTh 16.5) is sufficient evidence.17 And there is even one piece of well-known evidence for a request by a church council (that of Carthage in 407) for imperial legislation to prohibit remarriage while a former spouse is alive;18 it seems to have been ignored. On the other hand, the church showed considerable reluctance to try to impose generally on this world what it proclaimed as the ethical norms of the City of God.19 But these are not sufficient considerations to establish the intent of these laws. For that, two important points need examination. First, do the laws correspond to the position of the church both on divorce and on the relationship of church and state with respect to it? And second, what do the legislators have to say about their reasons? The church took a strong position against divorce right from the start. Paul (1 Co 7:8-11) gave as Christ's prescription that those who were married should not be separated and that a woman should not leave her husband; if she did, she should reconcile with him or stay unmarried; and that a husband should not dismiss his wife. This position rests on the same tradition reported in the Synoptic Gospels 15. A summary can be found in Wolff, pp. 263-69, esp. pp. 267-68 n. 23, who himself thinks that "doubtless the emperors did go a long way toward meeting the ethical demands of the new faith." Cf. the opposing view of J.B. Bury, History of the Later Roman Empire (London, 1889), 2: 416: "The influence of Christianity on the legal conception of the conjugal relation was, as Zacharia remarks, small up to the time of Justinian; and it was the Isaurian Emperors who really introduced a Christian legislation on the subject." Bury points out that it was only Leo, in 740, who instituted punishment for fornication and who forbade divorce, even by mutual consent, except on very narrow grounds (wife's adultery, husband's impotence, life-endangering slander by either, leprosy). 16. LRE, p. 974. 17. Cf. the chapter on "Conversion by Coercion" in R. MacMullen's Christianizing the Roman Empire (New Haven, 1984), pp. 86-101. 18. C. Munier, Concilia Africae, Corp. Christ. 149 (Turnhout, 1974), p. 218: "In qua causa legem imperialem petendam promulgari." 19. See the cogent remarks of S. Mazzarino, Aspetti sociali del quarto secolo, Problemi e ricerche di storia antica 1 (Rome, 1951), pp. 38-41. 46

IV (Mk 10:2-12; Mt 5:31-32, 19:3-12; Lk 16:18), where Jesus denounces divorce as Moses' concession to "hardness of heart" and not in accordance with God's plan for men and women. Paul on his own authority further counsels Christians not to divorce unbelieving spouses, urging them to hope for their conversion (1 Co 7:2-16). Neither the Gospels nor Paul's letters, of course, are systematic treatises; and Paul does not give the church at Corinth any sanctions for disobedience to enforce on those who do not follow his prescriptions. The church fathers down to the fifth century dealt with the subject of divorce on many occasions. Their reasons for forbidding it vary, as does their attitude toward marriage itself (some, for example, admitting affection as well as procreation as grounds for marriage), but they are mostly agreed that while divorce (in the sense of separation) is allowed for adultery by the other spouse, remarriage is not allowed to either party under any circumstance while the former spouse is still living.20 The theologians and councils thus strongly urge the faithful not to take advantage of their civil right to remarry after divorce, even though many (a majority, in fact) of the theologians took as a command Jesus' permission to divorce an adulterous spouse (unclear though the interpretation of that passage is) and considered it mandatory. Ambrose (Commentary on Luke, 8.4) deals directly with the relationship of the church's teaching to civil law, in saying that human law allows divorce, but divine law forbids it. As a sacramental conception of marriage takes form with Augustine, a still stronger theoretical basis is provided for regarding marriage as indissoluble, at least at a spiritual level, while the partners lived. In taking Jesus' radical statements on divorce at face value as ethical prescriptions for believers, the church had already from early days 20. For a careful and exhaustive treatment of the patristic discussions of divorce, see Henri Crouzel, L'eglise primitive face au divorce du premier au cinquieme siecle, Theologie historique 13 (Paris, 1971), on which most of what follows relies. The subject is, of course, enormously controversial. Crouzel deals with many of his critics and those of a different persuasion in a collection of articles, Manage et divorce, celibat et caractere sacerdotaux dans I'eglise ancienne (Turin, 1982). Ample bibliography will be found in both books. B. Lobmann, Zweite Ehe und Ehescheidung bei den Griechen und Lateinern bis zum Ende des 5. Jahrhunderts (Leipzig, 1980), though he lists (p. 47 n. 59) Crouzel's book of 1971, seems hardly to have used him, repeating his examination in detail. He sees a more substantive theological division (eschatological vs. legalistic) between East and West; while he seems right that the West is more legalistic, it is not clear to me that this is the result of theological differences so much as of disciplinary and pastoral ones. The book is overall less satisfactory than Crouzel's. Much of the bibliography on divorce is apologetic and polemical in tone, seeking (especially since Vatican II) to support a particular view of how the Roman Catholic church of today should deal with divorced and remarried members.

47

IV taken a step it did not take with much of his other radical preaching.21 Perhaps more important, affecting all areas of moral teaching, was the general development of a conception of church discipline which excluded from communion not only heretics but those regarded by the church as grievous sinners. This development did not take place all at once, but it was nearing maturity in the fourth century. It must be emphasized that this legalistic conception of regulating Christian behavior was equally a choice which could have gone differently. And even with that choice made, the formation of canon law was a long process, with no centralization or uniformity for a long time. It is anachronistic, I think, to look to the authors of this period for clear distinctions between behavior which is allowed and that which is only tolerated.22 The authors tell us that certain behavior is contrary to divine law. A few of them, and some of the church councils, tell us that the church imposed penance or more severe penalties on certain behavior. We find in the theologians and councils of the ancient church a considerable variety of opinion about how best to deal with the pastoral issues raised by the failure of many of the faithful to conform to the church's teaching on the issue of divorce. (None of them, however, except the council at Carthage cited above, sought to make the civil law conform to the church's teaching.) Not all of them took the view which ultimately prevailed in the West, that those who remarry after divorce should be excluded from communion. From the third and fourth centuries, the texts which pertain to the actual practice of pastoral clergy (as opposed to the abstract dictates of theologians) show some latitude. Origen cites the practice of some bishops in permitting remarriage to a woman while her husband is alive. He describes this behavior as contrary to scripture, but as intended to avoid greater evils, 21. See Crouzel, L'eglise primitive,^ p. 33, on Jesus' equally uncompromising teaching about oaths: "Or, non seulement 1'Eglise a constamment admis le serment dans la vie publique et privee, mais elle a condamne a plusieurs reprises des heretiques qui en refusaient la liceite et elle a impose des serments aux membres du clerge, reconnaissant par le fait meme qu'elle vit dans un monde ou le "mal," c'est-a-dire le mensonge, existe encore, bien que 1'ideal du Royaume comporte la sincerite absolue. Or elle refuse d'appliquer a la troisieme antithese le meme realisme, supposant ainsi que le bapteme a radicalement supprime la "durete de coeur" de 1'Ancien Testament, et elle impose a tous les separes avec une rigidite juridique absolue, sous peine de refus des sacrements, la continence "en vue du royaume des cieux", comme si c'etait "donne a tous" malgre les empechements psychologiques et les insuffisances spirituelles." 22. On this point I believe that Crouzel is throughout his work too quick to use categories not really applicable at this date; that does not, however, diminish my agreement with his analysis of the authors' views.

48

IV and he thinks that the bishops have not acted unreasonably.23 The Council of Elvira in Spain (probably 306), while maintaining on the whole a strict line, makes an exception in admitting to communion, in the event of infirmity, a woman who remarried after divorcing an adulterous husband.24 The Council of Aries, in 314, ordered that husbands who put away unfaithful wives and were prohibited by the church from remarrying be advised (consilium eis detur) not to remarry while their wives lived; but no sanctions were imposed on them if they did remarry.25 As these are the first two councils from which such canons are known, they are valuable evidence for the state of things in the early fourth-century West. A more complex situation, which shows how much actual practice might vary, is found in the canonical work of Basil of Caesarea.26 Basil's views are found in three letters of 314-315; all of them are based on and envisage specific situations. Basil allows hardly any reason for a woman to leave her husband; if she does, a woman who lives with the deserted husband is not condemned, he says. (The latter's behavior is classified as fornication but excused.) Basil says that the reason for the difference in customs for men and women is not easy to give, but that this is the way local custom goes. The view described by Basil (it is not at all clear that it is his own) is somewhere between the Roman and Pauline views of male/female equality in the matter of adultery, but closer to the Roman; the husband is treated as a fornicator with unmarried women, as an adulterer only with a married one. Even more striking is Basil's distinction in the matter of what behavior a spouse must put up with: a wife is to put up with almost anything, a husband not with adultery. Basil's practice and doctrine are thus different for the sexes, in that an abandoned husband can cohabit without being excluded from communion, whereas an abandoned wife cannot. Most striking of all is his admission that he cannot justify the local Cappadocian practice on the basis of the church's doctrine. The apocryphal canons of Nicaea27 seem to provide that a wife or husband unjustly accused by her or his spouse can repudiate him or her 23. Origen, On Matthew 14.23, cited by Crouzel, L'eglise primitive, pp. 82-83, from Die griechischen christlichen Schriftsteller 10: 340.25. 24. See Crouzel, L'eglise primitive, pp. 116-21 on this text. 25. Crouzel, L'eglise primitive, pp. 121-23; see C. Munier, Concilia Galliae, Corp. Christ. 148 (Turnhout, 1963), p. 11. Crouzel, Manage et divorce, pp. 88-91 and 127-42, discusses this canon, but his attention is focused on one textual crux, the outcome of which does not affect the question at stake here. 26. Crouzel, L'eglise primitive, pp. 133-50. 27. Ibid., pp. 240-43. 49

IV and remarry within the church. These canons come from an Egyptian source, perhaps as early as the start of the fifth century, but the date is not absolutely certain. They do not, however, give permission to divorce an adulterous spouse and remarry: if anyone does this, he or she will be chased from the church. There are also some Armenian canons28 whose date is uncertain; some probably belong to 365. From them it seems that the Armenian church allowed remarriage after delay and penance. The general conclusion to which these texts lead is that neither in the West nor particularly in the East was there uniformity in practice in the fourth and fifth century church in the discipline of communicants who remarried after a divorce. This does not mean that the church accepted and blessed such marriages, but it did in some instances show pastoral indulgence in dealing with the remarried member. We have no way of estimating how widespread exclusion was; the majority of the church councils do not deal with such marital issues at all.29 Nowhere does the church invoke penalties other than public penance or exclusion from communion, nor suggest the application of the restrictions actually known to us from imperial laws, most of which concern property which the majority of communicants did not have. On our second major question, the actual character of the late imperial legislation, two approaches are useful. First, the legal doctrine underlying the legislation. Wolff has shown that "the main pattern of thought which dominated Constantine's and his successors' divorce laws is already apparent in Augustus's well-known ban on the divorce of her husband-patronus by the liberta"^ And after extensive study of the doctrinal foundations of the late imperial legislation, he concludes, divorce is forbidden, but once effected it cannot be brushed aside, since it always remained essentially the actual dissolution of a primarily social relationship. Only —in the lex lulia — some of the normal consequences of a lawful divorce are eliminated, or—in the legislation of Constantine and his successors —the

28. Ibid., pp. 244-46. 29. The Gallic councils hardly touch on the subject at all; the African ones only a little more. Ferrandus' Breviarium Canonum, drawn up in Carthage c. 523-546 on the basis of a wide variety of councils from West and East, includes only two rules (out of 232 canons) dealing with divorce (Concilia Africae, pp. 287-306). 30. Wolff, "Trends," pp. 279-88. 50

IV unlawful divorce brings into play private and criminal law sanctions against him or her who disobeys the law. In short, nothing in the late legislation either specifically adduces or implicitly rests on a doctrine of marriage different from the classical Roman one, so opposed in character to what we have seen of the church's understanding of the relationship. Motives are harder. Wolff, though denying any doctrinal change in marriage law, thinks that "the Christian emperors were motivated by objectives entirely different from those of Augustus."31 Is this true? The emperors are on occasions explicit about their reasons. First, as Theodosius II said in his Novel 12 (439), "the consideration of children demands that the dissolution of marriage must be more difficult." The remark is repeated at the start of CJ 5.17.8 (449). From the fact that all of the marital legislation is largely concerned with the disposition of dowry and marital gifts, we may surmise that what this means is that the consideration of the protection of the property of the children was at stake; the same is true of restrictions on remarriage.32 The children of a first marriage were assumed to be injured by a second one. There was also a need for the safe-guarding of the separateness of the upper classes. We see both these motives at work, for example, in CTh 9.9.1, an edict of Constantine in 329: he prohibits the union of a women with her own slave. Both violators are to receive a capital sentence, and anyone has the right of prosecution. Any existing unions of this sort are to be broken up and the slave exiled; any children are to be deprived of the insignia of rank and their property. It is this last point which gives the game away, and it is brought home by later imperial legislation as well.33 Most people did not have any insignia of rank to be taken from them, after all. Exactly the same motives lay behind Augustus1 legislation on morals and marriage: he wanted senators to marry within their order, keep their marriages intact, be faithful, and produce lots of children. What the less exalted classes did, he could not have cared 31. Ibid., p. 287. He rejects, as too closely tied to the details, the detailed arguments of EJ. Jonkers, Invloed van het Christendom op de romeinsche Wetgeving betreffende het Concubinaat en de Echtscheiding (Wageningen, 1938), who emphasized the continuity of this legislation with earlier imperial enactments and identified in prechristian thought sufficient basis for the ideas which the emperors express. As we will see, however, when one tries to move from the specifics to the "general tendency," as Wolff calls it, one finds oneself in quicksand. 32. Cf. Jonkers, Invloed, p. 190, citing CJ 5.9.5.6. 33. See especially NovAnth 1, where the rule is extended to include freedmen as well as slaves, though existing unions are not broken up. Anthemius is explicit that it is the upper class which is the intended audience. 51

IV less,34 and the same is true of his Christian successors more than three centuries later.35 These prudential considerations for family property and the integrity of the upper classes have nothing in common with the dogmatic basis of the Christian view of marriage described above, any more than the doctrinal basis underlying the actual edicts corresponds to a Christian understanding. And the laws themselves, which limit unilateral divorce to various acceptable grounds, allow divorce by mutual consent until Justinian, and permit remarriage after no lapse of time or one determined by the frivolity of the excuse for divorce, have no common ground with the church's view that remarriage after divorce was inconsonant with the nature of marriage in the first place. It seems, therefore, that neither doctrinal underpinnings, nor stated motives, nor actual results concord with a view that Christian influence is at work in the limits on repudiation. And yet almost all writers on this subject have thought that the Constantinian and Justinianic limits on unilateral divorce proceeded from the influence of Christianity. For the most part, however, these writers have simply affirmed this influence, as if it were self-evident.36 The few discussions to speak to the matter more explicitly are 34. Cf., e.g., R. Syme, The Roman Revolution (Oxford, 1939), pp. 443-46. 35. The same attitudes can be seen in other areas of legislation. For example, social policy did not discourage sex for hire, as long as it involved women of the lower class. A law of Constantine from 326, classified in the Theodosian Code (CTh 9.7) as being on the law of adultery, makes this and much else clear: in cases of accusation, one is to check into the status of the woman. Tavern girls go free: "In consideration of the mean status of the woman who is brought to trial, the accusation shall be excluded and the men who are accused shall go free, since chastity is required only of those women who are held by the bonds of law, but those who because of their mean status in life are not deemed worthy of the consideration of the laws shall be immune from judicial severity." Serving women in drink shops were practically synonymous with prostitutes in antiquity, and it could hardly be clearer that (as three centuries earlier with Augustus) social class was the essential point of morals legislation. One sees no trace of any intrusion into this legislation of the uncompromising teachings of the gospels. 36. The literature is vast. As examples: C. Hohenlohe, Einfluss des Christentums auf das Corpus juris civilis (Vienna, 1937), an uncritical and pious account; J. Gaudemet, La formation du droit seculier et du droit de I'eglise aux IVe et Ve siecles, 2 ed. (Paris, 1979), who describes Christian influence on family legislation as "indeniable, encore que plus limitee qu'on ne 1'a cru parfois"; J. Vogt, "Zur Frage des christlichen Einflusses auf die Gesetzgebung Konstantins des Grossen," Festschrift fur Leopold Wenger, Miinchener Beitrage 35 (Munich, 1945), 2: 118-48 at p. 134: despite the lack of resemblance of Christian and legal views, he insists on a connection (admitting, however, that the punishment of a second marriage is only a matter of protecting the children of the first, not of a Christian view of remarriage); cf. Wolff, "Trends," p. 267 n. 23 for further bibliography, and a more detailed report on the literature concerning Christian influence on late Roman law by L. Wenger inArchivfiir Papyrusforschung 12 (1937), 298-307. 52

IV unsatisfactory, no doubt because motives which find no expression and which cannot be deduced from the doctrine of marriage at work in the laws are hard to demonstrate.37 For example, Wolff cites in support of "the general tendency which, especially with Constantine and again with Justinian...was directed toward strengthening the marital bond in the Christian sense" a work of Biondo Biondi on Justinian, in which Biondi proclaims that Justinian was "the first Christian emperor who promulgated a truly Christian law," referring to Novel 117.10. "Riforma audace che di tanto si allontanava dalla tradizionale concezione pagana del matrimonio di quanto rispondeva alia concezione cristiana," says Biondi.38 Now the Novel in question does not offer any justification for its denial of consensual divorce, any more than for other provisions. Insofar as one sees a consistent theme throught Novel 117, it is (as usual) the protection of the property rights of the children of repudiated marriages. The one overtly "Christian" element is the exception made for those who wish to devote themselves to a chaste religious life;39 but such separation is in general not approved of by the main Christian theologians of the period before Justinian.40 Biondi's affirmation is in fact nothing more than an act of

37. Cf. V. Basanoff, "Les sources chretiennes de la loi de Constantin sur le repudium," Studi in onore di Salvatore Riccobono (Palermo, 1932), 3: 177-99, who bizarrely describes the penalties in the Constantinian law as (p. 194), "la formule de transaction inspiree, selon toute vraisemblance, par le texte du Pasteur, adapte a des conditions nouvelles." It is hard to see what possible connection between Hernias' teaching and the Constantinian legislation can have provoked this remark. Equally odd is his insistence that classical Roman law knew no divorce by consent (p. 193); since the decision of one person was sufficient, surely that of two was also adequate, and classical law required no particular form of divorce. 38. Giustiniano primo, principe e legislatore cattolico, Pubbl. Univ. Catt. del Sacro Cuore, 2 ser., Scienze giuridiche, 48 (Milan, 1936), p. 43. 39. Castitatis concupiscentia, in the deliciously oxymoronic phrase of the apparently contemporary Latin version of the law preserved in the Appendix to the Epitome of lulianus (sophrosynes epithymia in the original Greek); Biondi apparently did not appreciate the phrase, for he substitutes desiderium (from the translation of the editor, Schoell) for concupiscentia in quoting it! One should also note relegation to a monastery as punishment for contravention of these laws by women ae hardly a Christian view of the monastic vocation. 40. See Crouzel, L'eglise primitive, pp. 376-77. Such separation is forbidden by a canon from a Council of Gangra cited in Ferrandus' Breviarium, in Concilia Africae, p. 301, no. 164: "Ut si qua mulier quasi religionis causa virum dimiserit, anathema sit."

53

IV faith in his "catholic" Justinian.41 Now one cannot demonstrate Justinian's state of mind before ordering this legislation, and that a Christian emperor wanted to discourage divorce may be quite right; but to affirm his motives, in the total absence of evidence, is the work of a hagiographer, not an historian.42 We now turn to inquire what the effect in practice of all of this legislation was. An enumeration and description of the surviving documentation for divorce is enlightening. Traditionally, Egyptian law had, like Roman law, allowed divorce either by mutual consent or by repudiation, with no penalties for either party unless some had been established in a contract concerning property.43 In Roman Egypt up to the fourth century, these compatible traditions produced a consistent habit of divorce agreements by mutual consent, in which both parties are freed of liability.44 The first of our fourth-century documents (P.Oxy. XXXVI 2770, of 26 January 304) shows the expected pattern: Herakles and Maria45 agree that they are divorced, that each has his own property, and that they have no claims against each other. There are no children, each is able to remarry at will. The document itself is called tot ifj/

/

^^^^-^^^

0.1

0.2

0.3 0.4 0.5 0.6 0.7 Cumulative Share of Population

0.8

0.8-]

j

0.9

1.0

j

1

Q7

-D

I

I

//

//

I °-3"

0.0

-/I

//^

-D 0.8-

0.1-

s'

.S

0.9-

FIG. 2. HERMOPOLITE METROPOLITANS : DISTRIBUTION OF LAND BY DECILES

°- 6 ~

I °'5~ nj 0.4-

"b

0) 0.3oj

^ 0.2-

o|

n

— . - — . — .^.^.r—..r—i.rn.l 1

2

3

4 5 6 7 Deciles of Population

8

9

M I 1 0

A closer look, however, reveals problems. I have recomputed the Gini index and found that the correct figure is much lower: .532 using all persons listed, .516 limiting ourselves to cases with complete data (the latter figure is probably more reliable, though it leaves out some large holders).18 A distinction is in order, however. The Philadelphia list is composed mainly of 183 villagers, plus a minority (17) of Alexandrian or metropolitan citizens. The index for the Philadelphians is .518 for the cases with complete data, or very close to that for the entire list. That for the city-dwellers, however, is harder to discern. Because of physical damage to the papyrus, the portion with city-dwellers is badly preserved; the index is .438 using all seventeen cases, but only .248 using the fully preserved ones. It is hard to say which is the more realistic, if either; at any rate, the sample is too small for much meaning. These figures are theoretically more readily comparable than the village ones to the Hermopolite index. But there is again an important difference, in that the Hermopolite index reflects total (except around the city) landholdings of a portion of city residents, while the Philadelphia register includes the holdings of all city residents but only in one village. A further complexity, probably more important, is that the Philadelphia list includes only private land (both grainland and orchards); public land is not included, and it is a reasonable bet that this land, leased out, was held in much more equal parcels than the private land, which was presumably acquired by 18 Bowman's own recalculations now agree (letter of 18 March 1991); the source of the original error is obscure to both of us, but it was unique to this calculation, as my

own computations have confirmed all of his other calculations,

XII 132

inheritance and purchase.19 By the fourth century, public land had been privatized, but there is no guarantee that the contents of these categories had not changed in the intervening years.20 Bowman cites a third data set from which he computes a Gini index, 'the only other place in Egypt which permits a comparison', late Ptolemaic Kerkeosiris. There, he finds a lower/? (.374). This population's holdings include both allotments to military settlers and leaseholds of crown land, and the data are virtually complete. It is thus partially comparable to the Philadelphia villagers' index, but could be expected to be lower as a result of the inclusion of royal land leased to peasants. As Bowman says, 'it is hardly possible to generalize from such sparse data and even without the evidence of Kerkeosiris we would be tempted to suppose that the increased scale of private ownership of land in the Roman period had the effect of exaggerating the inequality of ownership' (151-2). IV.

DATA FROM KARANIS

In point of fact, however, there is one other source for the distribution of ownership of village land in the fourth century, the cluster of taxation reports from A.D. 308/9 (though submitted several years later) from Karanis published in The Archive of Aurelius Isidorus. These are not land registers,21 but rather accounts of the assessment and the collection of wheat and barley taxes for the village. For the majority of the taxpayers, we know the amounts in both grains, but for a minority we know only the amount of wheat paid, and for a handful we know only amounts in barley. Now barley taxes were assessed at a fixed amount (.75 artaba per aroura) on both 'private' and 'public' land, 22 and it is thus a fairly simple procedure to calculate the total area of land for which barley taxes were paid where that figure survives. But this does not help with the distribution between private and public land, and of course it does nothing for those taxpayers for whom only a wheat figure survives. Worse still, there is clear evidence in these reports that the collection of barley did not come up to the assessed level, unlike the wheat collection. Some of the barley figures will therefore generate distorted numbers for landholdings. The problems in using these reports are thus not trivial, and it is perhaps for this reason that no one has ever analysed them systematically. But it is possible, I believe, to derive a very good approximation of the structure of wealth in Karanis from them through a series of adjustments and calculations. Some uncertainties and inaccuracies are introduced at each stage, but I shall argue that they have no material effect on the outcome. First, it is an easy matter, where both wheat and barley payments are preserved, to calculate the distribution of the payer's land between public and private. If we use x to represent the number of arouras of private land held, y those of public land, w the number of artabas of wheat paid, and b that of barley, the total landholding (x + y) can be found by dividing b by .88 (i.e. i aroura per .75 artaba). 23 The formula for public land is y = ( ( w / i . i ) ( 2 ) — (x + y ) ) / 2 . In Table i (tables are given at the end of the article) the landholders are listed, in the order in which they are given in P.Cair.Isid. 9, with their payments and the imputed amounts of land owned. Where names appear in the lists but the amounts are lost, a blank is left; where the name does not appear under one of the grains, a zero is entered. Missing fractions are disregarded, and where a ones digit is lost but a tens is preserved, the latter is given. Now it is immediately apparent that the lack of barley values leaves many lines with meaningless numbers. It is also clear that even where both figures are preserved, they produce 19 That does not necessarily mean any simple, egalitarian pattern of leaseholds by 'peasants'. More than two centuries of Roman rule had given plenty of time for the rights to such land to have been divided, subleased, or ceded in ways that will have made the situation with this land very complicated, and some owners of private land may also have leased public land. 2 " As Bowman points out to me, the low percentages (under 20 per cent) of 'public' land in the Hermopolite and Oxyrhynchite in the fourth century are difficult to accept as accurate reflections of how much land was owned by the crown in earlier centuries. 21 P.Cair.Isid. 6 is such a register, from perhaps five to ten years earlier, but a much smaller part of the population

is represented there and a very substantial part of the numbers are missing, so that no useful conclusions about distribution seem to me possible. n These terms meant only different tax rates in this period, see Bowman, op. cit. (n. 7), 148—9. The barley tax rate is given in P.Cair.Isid. 11.25; c ^- tne editors' introduction for the computation of the barley taxes for the village as a whole. 23 This formula and the others are derived in detail in my article on 'Bullion purchases and landholding in the fourth century', Cd'E 52 (1977), 322-36, at 330 n. i. I calculated the landholdings only for members of Isidores' family in that article,

XII 133

L A N D H O L D I N G IN LATE R O M A N E G Y P T : THE D I S T R I B U T I O N OF W E A L T H

irreconcilable results in some cases, principally because the barley payments are too low for any possible combination of landholdings which could generate the wheat taxes paid. Since we know from P.Cair. Isid. 11 that of a total assessment for the village of about 3,715 art. of barley, only about 3,531 art. were collected and handed over to the government, it is a natural conclusion that barley taxes were underpaid by these people. There are only a few cases in which the wheat taxes are too low for the land represented by the barley, a fact in keeping with the observation that wheat collected actually exceeded the assessment by almost 44 art. (P.Cair.Isid. n, introduction). In Table 2, therefore, I have corrected for all of these deficiencies by a series of operations. First, where the figure for barley taxes is missing, it has been set at .815 of the wheat taxes. This figure approximates the ratio of the barley assessment to the actual wheat taxes. For any individual, of course, this is a levelling procedure, since it assumes that his or her proportion of private to public land was the average. In very few cases, probably, will this be exactly true, but the resulting inaccuracies should not affect the distribution materially. Nor would taking a somewhat different figure (based on actual barley, or the wheat assessment) make a difference of more than a percentage point or so. Second, the reverse operation has been carried out where the wheat figure is lacking: the index of 1.227 nas been used to convert the barley figure into an imputed wheat one. These affect only a few cases and all of them are small, so any inaccuracy can have no significant effect on the overall picture. Where barley and wheat figures are present but impossible—that is, where together they indicate landholdings with a significant (over 0.5 aroura) negative amount of one class—I have followed the same procedure. A small group (four persons) for whom no data survive is omitted from the calculations, but they were probably all small holders. A few landowners would have higher figures if all numbers were preserved, but they would not alter the picture materially either.24 All of these proceedings assume, naturally, that people did not pay more taxes in either grain than they were required to; this commonsense presumption may of course be inaccurate on occasion, but if it were significantly wrong we would find that the totals of the 'corrected' list would differ substantially from the actually collected amounts. This is in fact not the case. The totals in Table 2 are within about i per cent of the amounts reported in P.Cair.Isid. 11, in the case of both grains running slightly high. A divergence of this order of magnitude is very unlikely to affect the validity of the calculations. The total amounts of land imputed, about 2,251 private plus 2,020 public, 4,271 in all, compares to the actual 2,212 plus 2,007, or 4,219 in all, a little over i per cent off again.25 These owners can now be listed in descending order of total landholdings, metropolitans first and villagers afterward. The results can be categorized summarily as follows: METROPOLITANS

VILLAGERS

2OO +

I

150 +

100-199

I

100-149

50-99

40-49

o

90-99

i

70-79 60-69 5°-59

I

30-39 20-29 i5-J9

4

5-9 0-4

2 5

IO—I4

TOTAL

I

2

I8

80—89

40—49

I 2 2 2

4 4 6

I2

30-39 25-29 20-24

i5 14 8

10-14 5-9 1-4 No data TOTAL

16 10 4 4 no

I

5~ I 9

24 For example, I computed Gini indexes including the four with no data and excluding them; the difference was only .022. " It may be worth adding that I will gladly make any of the spreadsheets for data presented in this article available

6

on standard diskette on request; they were prepared on Borland's Quattro Pro and should be usable on any Lotuscompatible spreadsheet programme for an IBM or compatible computer,

XII *34

A distribution of ownership by deciles and a cumulative distribution by deciles are given as Figs 3 and 4. 0.3-]

1 I

FIG. 3. KARANIS VILLAGERS : DISTRIBUTION OF LAND BY DECILES

1

0.25T3

|

0.2-

~° ffl 0.15-

I—I

"o Q)

ra

I

0.1-

C/5

I

0J—n.n.l 1

2

l,l I.I I.I 1,11,11,11,11— I

0.05-

3

1

1

I

4 5 6 7 Deciles of Population

8

9 1 0

1 -j

1

r~

0.9-

1

FIG. 4. KARANIS VILLAGERS : CUMULATIVE DISTRIBUTION OF LAND

?0.8-

3 0.7-

"b o>

I

0.6-

cc W 0.55

i a4 " I

"

oz

0.3-

"

n

I—I

n

I

I

I

_.rn n .PI , I I I I H I I I I I 1

2

3

4 5 6 7 Deciles of Population

8

9

1 0

The results for metropolitans are interesting but of limited applicability. For one thing, the number is small. For another, it must be kept in mind that the metropolitans included in this roster may well have had landholdings elsewhere; indeed, they are almost certain to have had. Among the Hermopolites analysed by Bowman, those owning over 100 arouras were very likely to have held land in multiple locations, particularly when it is remembered that their holdings in the pagus nearest the city are not included. There were two large metropolitan accounts at Karanis, people with really extensive holdings (133 and 205 ar.); we have no evidence for the character of these estates, whether single tracts or fragmented, and the owners do not appear in the archive of Aurelius Isidoros as lessors. The remaining holdings are much smaller, with one each at 47, 36, and 28, then six in the teens and seven under 10. The large range and radical inequality described by Bowman for the Hermopolite city residents in their country estates is manifested here again, with a Gini index of .638. It is clear that any given data set involving urban ownership in a particular village may vary greatly from others; we would expect the results from one village to another to be less consistent than those from one nome to another. Above all, however, as in the case of Philadelphia these figures can

XII 135

L A N D H O L D I N G I N LATE R O M A N E G Y P T : T H E D I S T R I B U T I O N

OF WEALTH

legitimately be compared only to other data for metropolitan land in a particular village, not to data for metropolitan ownership over an entire nome (as in the Hermopolite registers).26 For the villagers, however, the situation is rather different and more meaningful. There is a continuum, with the main bulk of the population (71 of 106) in the range of 10 to 50, fourteen below that range and twenty-one above it. The source of our documents, Aurelius Isidores, had holdings of about 36, but was part of a family with more extensive holdings. The median holding of those holding tax-collection liturgies seems to have been in the 2os, with men owning as little as about 12 arouras sometimes chosen.27 By that standard, there were about eighty-five landowners eligible for liturgical service, or more than three-quarters of all landowners. It is, of course, possible that some of the villagers also owned land in another village's territory. The Gini index for landowners for whom full or partial data are preserved (omitting the four cases mentioned above) is .431, somewhat (10 per cent) higher than that at late Ptolemaic Kerkeosiris but markedly lower than the metropolitan figures. There was a broad middle ground among the villagers. The Lorenz curve, shown in Fig. 5, is similarly strikingly different from that for Hermopolis. 1

—

^ ./

0.9-

c

Ol8

5

°' 7 ^

S^

~ /^

0 w 0.6co 6 0.55 tf

£

// /

1 0.3-

/

O 0.2-

./

// /

0.1

/

/

/ /

/

FIG. 5 . KARANIS VILLAGERS: LORENZ CURVE DISTRIBUTION OF LAND

/ /

/

.^

^^^^

0 Y^-—^~~~^\

0.0

.s

/^

/

/

//

0.4-

0.1-

s'

.// /

0.2

1

,

.

,

r

0.3 0.4 0.5 0.6 0.7 Cumulative Share of Population

,

0.8

,

0.9

1.0

It is important to underline that our findings for Karanis cannot be considered typical without further evidence. For one thing, this was a village in trouble, as has long been recognized. The archive here is full of complaints about the failure of the irrigation system and the dryness of large parts of the fields. Karanis was only a shadow of its former self, and a landowner with 35 arouras may have had only a fraction that many in cultivation. It was certainly not the only Fayum village in this kind of trouble, but there are no good grounds for thinking that their difficulties were typical of the rest of the Fayum or of Egypt as a whole.28 The official figures in P.Cair. Isid. 11 show that about 9.5 per cent of the private land and 24 per cent of the public land was categorized as unsown; but in all likelihood the land assigned by e/>i'we/wesw for compulsory cultivation was largely unsown as well. If so, the percentages of unsown land rise to about 13 per cent for private land but 46 per cent of royal land. Overall, 29 per cent of the land would then be unproductive. The average holding, then, high as it may seem, must be discounted by about 30 per cent to allow for unproductive land.29 For these reasons, calculations about the estate necessary for a reasonable livelihood are even more difficult than normal. 26 By the same token, of course, the reservations expressed about the metropolitan data from Karanis are irrelevant to the Hermopolite figures. 27 See Bagnall, 'Property-holdings of liturgists in fourth-century Karanis', BASf 15 (1978), 9-16. 28 cf. my 'Agricultural productivity and taxation in fourth-century Egypt', TAPA 115 (1985), 290—308 for a detailed discussion of this problem. 29 It is entirely possible that this is too pessimistic, as Alan Bowman points out to me. No doubt the actual figure

fluctuated substantially from year to year. If the correct were instead 15 per cent (for example), the figure consequences would be that the village holdings in the nome model (below, iv) would be increased in size by about 21 per cent from the figures I have used, lowering^ for the nome to some degree. The figures for/? within the village, however, do not depend on any assumption about unproductive land, being computed from the gross holdings.

XII I3 6

Moreover, investment in land in such a village was less attractive for anyone interested in stable income than it would be elsewhere; only an entrepreneur active on the spot would have found the risks worth taking. Absentee landlords were, as Dennis Kehoe has pointed out in the case of Pliny, on the whole averse to risk and change; predictability, avoidance of need to invest in improvements, and stability were their aims.30 The relatively small percentage of outside ownership (about 13.7 per cent in Table 2) compared to the 25-30 per cent computed by Bowman for the Hermopolite districts may well be a reflection of the unattractiveness of Karanis to metropolitans. Some middle-range villagers could have picked up additional land cheaply, particularly in the late third century when things seem to have been worse in Karanis than a couple of decades later. The issue of typicality of villager distribution remains. The closeness of the Kerkeosiris figures to the villagers' index at Karanis seems the most significant fact. One possible test is to distinguish 'public' from 'private' land, i.e., land now owned by farmers but formerly leased from the state (which I have argued should have been more equally distributed when it was leased) vs. land which had been private for three centuries or more. It turns out that for the non-residents the distinction is minor; private land overall has a .626 index (vs. .638 for all land together). That is not surprising, since the non-residents had presumably acquired virtually all of their land by purchase or inheritance from a purchaser, and private land preponderates.31 Among the villagers, the Gini index rises from .431 to .478 when only private land is considered: significantly closer to the Philadelphia figure, it is noteworthy, although still somewhat more equally distributed. Even with all due reserves, this evidence suggests that the landholdings of Egyptian villagers tended to have only a moderate degree of inequality, an inequality least in holdings of 'public' land and greater (by about .100, perhaps) in holdings of 'private' land; and there is no particular trend visible over the more than four centuries across which our evidence is scattered. Landholdings of village land by non-residents have a wider range of possible degrees of inequality, particularly where small numbers of persons are involved; the upper range is perhaps almost double the index for village residents' holdings of public land.

V.

SIXTH-CENTURY APHRODITO

Another interesting data set, though from a later century, has become available since Bowman wrote, the land register from Aphrodito in the Antaiopolite Nome, fromc. 525/6.32 It lists i ,470.875 arouras of land in the category oi astika onomata, subcategorized by arable land (93.5 per cent), reed land (.8 per cent), vineyards (1.5 per cent), and orchards (4.2 per cent). The astika were holdings on which the taxes were paid to the treasury of Antaiopolis, as distinct from the kometika, taxes on which were paid to the village treasury. The editors deny any comparison with the distinction of politai and kometai in the Karanis fourth-century accounts.33 A study of the list, however, shows that of the holdings owned by individuals or personal estates, as opposed to institutions, 70 per cent were the property of persons identified as citizens of a city or holders of an imperial office and thus presumably not villagers. The remaining 30 per cent are not identified; there is no one identified as a villager in the register. Now some of these are likely to have been villagers, but I do not think it can be denied that the holders of the astika were in fact largely urban residents. On the other hand, it is true that some urban holders had property classified as kometika.

30 'Allocation of risk and investment on the estates of Pliny the Younger', Chiron 18 (1988), 15-42. Some of the same attitudes can be found in the British landed elite in early modern times; see Lawrence Stone and Jeanne C. Fawtier Stone, An Open Elite? England 1540-1880 (1984), 11-15 f°r the landowners' willingness to accept a low return in exchange for security and prestige. 31 It is true that some metropolitans or Alexandrian

citizens may have been resident at some time in a village where they had landholdings. There is very little evidence for this in the fourth century, but it may have been more common a century earlier. 32 Jean Gascou and Leslie S. B. MacCoull,'Le cadastre d'Aphrodito', Travaux et Memoires 10 (1987), 103-58, pis i—10. 33 eidem, 113.

XII 137

L A N D H O L D I N G I N LATE R O M A N E G Y P T : T H E D I S T R I B U T I O N O F W E A L T H

The astika amount to 24.9 per cent of the total area of holdings in the village. That percentage, very much in the range attested for urban holdings in the fourth century, combined with the names of the categories and the identities of the known holders, suggests to me the hypothesis that at some unknown point property ceased to change from one status to another when ownership changed. That is, once astika, always astika. The proportion of tax revenue which was supposed to flow through the different collecting agencies would in this way be stabilized; if an urbanite bought kometika in the future, he would continue to pay to the village treasury (which was in turn responsible to the imperial government), no matter what his own status. Conversely, a villager who bought astika would pay to the city treasury for it. Between them, institutions and urbanites own more than 84 per cent of all of the land in this register, a strong sign that little land had passed from urban to village ownership since the fixing of the categories. In effect, the terms would on this view be frozen terminology, perhaps analogous to 'public' and 'private' in the fourth century in having some fiscal significance but no other meaning.34 Of the sixty-four accounts attested in the register (many of which appear several times), forty-nine are individuals and fifteen institutions. The latter, only 23 per cent of accounts, own 46 per cent of the land. For such a small number of 'persons' computations of R are subject to many cautions, but because of one very large account the figure is .710. For the forty-nine individuals, R = .623. For individuals and institutions together, if the figure has any meaning, R — .684. The most interesting figure, I think, is that for individuals, which is very close to the Antinoite figure in the fourth-century register (.631). It must be remembered, however, that once again we have the holdings of some urbanites in a particular village, rather than a global picture of the holdings of that group. All the more striking, then, to recall that at Karanis holdings of metropolitans have R = .638. VI.

A NOME MODEL

The general consistency of village results encourages the attempt I shall now make to construct a model of an idealized Hermopolite Nome as a whole. That model rests on a series of assumptions, all of which I believe to be roughly correct but each of which introduces some uncertainty. These are as follows: 1. The total area of the Hermopolite Nome (including the Antinoopolite chora) was about i,140 km 2 , or 413,820 arouras.35 On an Oxyrhynchite analogy, we will suppose that 72 per cent of this was arable, taxable land, or about 298,000 ar.36 Somewhat arbitrarily, we will then assign about 10 per cent of this to the small Antinoopolite chora, leaving 270,000 ar. of arable land in the Hermopolite.37 2. Bowman's figures for landholdings by urban residents are essentially accepted: 10,000 ar. for Antinoites, 65,oooar. for Hermopolites, and 20,000 ar. in the seventh pagus, around Hermopolis, all of which is assumed to be owned by city residents. (That assumption no doubt is excessive and distorts figures somewhat.) 3. The remaining 180,000 ar. were owned by village residents. This is certainly an oversimplification. Corporate holdings by cities may not have been significant (F lists several totalling just under 150 ar., which would extrapolate to 600 ar. for the nome, not counting the seventh pagus}. A single entry for church property in G 534 is less than 30 ar. The imperial estates were probably significantly greater, but F 747—752, where they are concentrated, totals 34 The extensive struggle of the village of Aphrodito to maintain its autopragia, of course, shows that it took the matter seriously. Even though the taxes ultimately wound up in the same place, there were considerable consequences for control of the flow and for responsibility implicit in the distinction. * The basic figure is derived from Karl W. Butzer, Early Hydraulic Civilization in Egypt. A Study in Cultural Ecology (1976), 74, but adjusted for the fact that most of the territory of the Kynopolite nome on the

West Bank was divided between the Oxyrhynchite and Hermopolite by the Roman period. Factors of uncertainty are the precise dividing point and shifts in the Nile's bed in this period. % R. S. Bagnall and K. A. Worp, 'Grain land in the Oxyrhynchite nome', ZPE 37 (1980), 263-4. 3 ' These are somewhat lower than the figures offered by Bowman, op. cit. (n. 7), 147, who does not subtract for the difference between gross and net area,

XII '38

about 395 ar., which would extrapolate to about i ,600 ar. leased to urban residents. Even with the extrapolations for missing data described in the next paragraph, none of these would be major factors. It must be admitted, however, that the land registers do not preserve the portion for Alexandrians or other outsiders to the Hermopolite Nome. Moreover, land belonging to the imperial house but leased to villagers is not included. None of these gaps affects the distribution of land ownership among the population at issue here, but clearly they had some impact on the total distribution. 4. Patterns of distribution are assumed to be those of F Ant., F Herm., and Karanis. The individual holdings in F Herm. are increased by .46 (probably too much in some cases) to adjust for the absence of data for the seventhpagus and for incomplete data in F Herm.; those for Karanis are reduced by .30 to allow for the fact that those holdings include as much as 30 per cent unproductive land. This is clearly the most questionable assumption, given that patterns in the Fayum and in the Hermopolite may well not have been identical, but I can see no basis on which to correct it.38 5. It is assumed that F Herm. represents a quarter of all Hermopolite (metropolitan) landholders. We have no way of knowing if the correct figure is a fifth or a third instead. 6. A model population is supposed in which there are 7,400 rural landowners with an average of 24.3 ar. each (Karanis average discounted by 30 per cent); 952 Hermopolite holders with 89.3 ar. each (actual F Herm. average increased by 46 per cent); and 256 Antinoites, with an average of 39 ar. each. This population shows/? = .560 ± .002, significantly higher than the village index but far below the Hermopolitan.39 It can be argued that it is almost certainly exaggerated somewhat by the treatment of the seventh pagus, part of which was probably owned by villagers and part by urban residents who held no land in otherpagi, but it is unlikely that accurate figures for that would change the result by more than .002, which is trivial. 40 Far more serious concerns are the representativeness of the populations we know and are using, a question unanswerable at present. If this quarter of Hermopolis proved unrepresentative, and Antinoopolis much closer to normal, overall R would be lower (somewhere around .500, depending on assumptions).41 An interesting sidelight is provided by the male rfemale ratio of landholdings. At Karanis, women own about 17 per cent of the villager-owned land (and about 18 per cent of the metropolitan-owned land, but that is a small sample). At Philadelphia, women owned about 25 per cent of the land reported in the register.42 By contrast, Hermopolite women own only about 8.5 per cent of the land in F Herm., and Antinoite about the same (8.4 per cent). And the bulk of that belongs to a handful of women who are part of major families and whose male relatives are known. The combined model has men owning 86 per cent, women 14 per cent, of the total. The very considerable concentration of wealth in the overall model has significant implications for the generation of surplus income after taxes. On the reasonable assumption that 10 arouras was sufficient to support the average household, 59 per cent of the villagers' holdings were surplus to their personal requirements.43 For the urban population, the concentration of surplus is of course even greater, some 88 per cent for Hermopolites.44 In all, the nome's landowners may have generated a surplus (not counting tax revenue) of some 380,000 art. at a conservative figure of 2 art. per ar. We have no means of computing how much of this was used to feed the non-landowning population, much of which worked the land of the 38 The relatively high Greek settlement in the Fayum may have had some impact on patterns, but it is hard to know precisely what that may have been. 39 I have computed it in two ways. A 10 per cent sample of this hypothetical population was constructed, using i in 8 of the Antinoites, 2 in 5 Hermopolites, and 7 times the Karanidians; that yielded R — .558. A simple calculation weighting the separate R's yielded .561. (Herm., .829, weighted at .309; Ant., .631, weighted at .0363; Kar., .431, weighted at .6545.) 40 Within the computational error range induced by different sampling procedures. 41 Assuming a lower percentage of urban-owned land,

of course, would lower it further, perhaps another .020 if the real split were 25 per cent urban, 75 per cent village, 42 Remembering that this does not include public land. 43 This figure is derived by adding the holdings of those with more than 10 ar., subtracting the number of individuals multiplied by 10, and dividing the remainder by total holdings. ^ I have used n as a floor for urban residents, since their net yield was no doubt lower than that of villagers who actually worked much more of their own land. Moreover, 10 ar. at Karanis is equivalent to 7 ar. of productive land, probably close to what it actually took to support a family.

XII 139

N D H O L D I N G I N LATE R O M A N E G Y P T : T H E D I S T R I B U T I O N O F W E A L T H

owners, and how much was exchanged for other purposes, but if all of it were used as wages for non-owners, it would have supported something like 16,000 households, or perhaps 80,000 persons. We have supposed about 8,600 landowners in the model, which would imply that in this case only 35 per cent of all households owned any land. The figure of 16,000, however, is undoubtedly too high, since not all surplus was spent on labour. VII.

COMPARATIVE PERSPECTIVE

It remains to try to put all of these figures into some kind of perspective.45 What constitutes a high or low index of inequality? Just how tricky matters can be may be seen from an elaborate study of wealth in mid-nineteenth-century Wisconsin.46 The figures most closely comparable to the Karanis figures are the indexes for the distribution of farm sizes, or land acreage, which stood at .400 in 1860. Despite a near-tripling in the size of the average farm over the next century (from 54 acres to 153), and numerous other social and economic changes, this index had fluctuated in a narrow zone, ending up at .360 in 1959.47 These figures are not greatly dissimilar from .431 for Karanis villagers and .374 for Kerkeosiris, and even the .518 index for Philadelphia is not grossly dissimilar if one bears in mind that it represents only private land; if we had the royal land figures, its overall distribution would probably be reasonably close. And they have in common that they all exclude the zero cases, i.e., the landless. That, it should be pointed out, is an important exclusion and reminds us that we are not dealing with the totality of the rural population. In Wisconsin, including non-owner adult males in the reckoning of land ownership drove the Gini index from .400 to .670.48 We have no way of reckoning the number of such adult males either at Kerkeosiris or at Karanis. 49 It is not likely to have been so large, nor the step from land to wealth so steep, because a significant share (about 40 per cent) of inequality in Wisconsin was attributable to the distinction between native and immigrant. That is, recent immigrants to the United States tended to have much less wealth than natives.50 This factor is not likely to have been operative in any significant way in Roman Egypt.51 That caveat aside, we must next point out that the distribution of landholdings and the distribution of landed wealth are not necessarily identical. The Gini index for the distribution of the value of farms in Milwaukee County was 15 per cent higher than that for the acreage (.440 vs. .380). Some land is inherently better, and some has a more valuable working capital in buildings and equipment added to it. It is certainly likely that this was true also in Roman Egypt. For example, if one assumed that orchard land was worth twice as much as arable land at Philadelphia, R for villagers would rise from .518 to .552. The divergent value of land does not in itself, of course, lead to a higher Gini index, since such inequalities could be distributed in the same pattern. Whether the Gini index for value of farms in Roman Egypt would be the same as for acreage we cannot know with present evidence, but the Philadelphia register suggests not. In the case of only the urban population of a nome (or, as we actually have, a portion of it), another factor enters in, namely that landed wealth is only a portion of total wealth, and it produces only a portion of total income for at least some of the people involved. These factors have complex effects. It might be thought that adding in non-landed property would help to balance out differentials in landholdings and reduce the index. This is not necessarily the case. In 1860 Wisconsin, non-farmers had a substantially higher Gini than farmers (R — .830 vs .690, measuring wealth for all adult males in each group), and inequality in Milwaukee 45 The other figures for R in the Roman Empire derived by Duncan-Jones, op. cit. (n. 12), do not seem to me very useful, for reasons he pointed out himself, quoted above (at n. 13). The best of the lot is probably that from Ligures Baebiani, which gives/? = .435, close to that of Karanis. 46 Soltow, op. cit. (n. 16). 47 idem, 121—4. It was .410 in 1870 and .380 in 1910. These are for not quite comparable sets; that for 1860 is 'improved' land; the Gini for all farmland was .364 (p. 57). 48 idem, 124; his table there for Milwaukee County alone shows a similar rise from .380 to .640 when zero cases are included.

49 Their absence significantly limits the significance of the Gini index; cf. Soltow, op. cit. (n. 16), n. 50 The average native-born farmer aged twenty or higher had 169 per cent of the wealth of the average foreign-born farmer aged twenty or higher. See Soltow, op. cit. (n. 16), 47. 51 Soltow, op. cit. (n. 16), 127, makes the general point that Wisconsin in the middle of the last century was not yet a stable society. On the other hand, his overall findings are that income inequality has decreased greatly (by a half, roughly) since then, while wealth inequality has decreased very little (p. 139).

XII 140

County, the most urbanized part of the state, was still higher, at .890. In other words, nonagricultural assets tended to be distributed still more unevenly than land.52 That does not show that the same was true for a metropolis of Roman Egypt, but it at least suggests the possibility. As to the land itself, it is apparent that no more than a fraction of the urban population owned any at all: 20 per cent would be a reasonable guess.53 A Gini calculation putting all the rest in as zero would drive the index close to its theoretical limit. There is nothing inherently improbable in that. The Gini index for landowners of i acre or more (excluding both zero cases and those with plots smaller than i acre) in the England of 1875 was .858.54 Wealth, to be sure, is not the same as economic welfare. Income is probably a better measure of well-being than assets. In an agricultural economy, there is a substantial correlation between wealth in land and income, because most income is derived from the land.55 A distinction should be made, however, between rural and urban populations, and within the urban population, between those dependent mainly on income from land and the landless.56 The rural population, which works the land, will see a major distinction between those whose income depends solely on labour—hired hands—and those who own land. Within the latter group, landed wealth will be the largest single component in inequalities of income. The urban rentier class will have a still higher correlation between wealth and income, since their own labour is not involved. The non-rentier remainder of the urban population, however—and we have seen that it is the vast majority of city residents in Roman Egypt—will not necessarily find much correlation between wealth and income. Historical trends of the last few centuries underline this point. Numerous studies have documented a significant fall in the R for income in all developed countries (they are much higher in the underdeveloped countries57). But a similar decline in R for wealth has not been found; inequality of wealth has declined much more slowly, if at all.58 The reason lies mainly in a decline in the wealth to income ratio, which indicates in the main a rise in the proportion of income derived from salaries and wages (as well as transfer payments) and a decline in the proportion derived from property (rents, dividends, interest).59 The decline in the Gini for income is almost entirely a product of a growth in the income received by the middle classes; lower-income groups have not gained significantly.' For these reasons, the structural importance of the very high Gini indexes we find for Hermopolite urban landowners depends very largely on what role we suppose that wage labour and income from businesses with little capital needs (cabbage-sellers, for example) played in the economy. The absence of income data with which to test the wealth data is thus particularly damaging to our ability to use the latter. A comparison of distribution of landholdings by deciles of the populations studied above helps to flesh out the bare number provided by the Gini index. In the 'decile' column are shown the tenths of the village population (i = the poorest tenth, and so on), while the columns for Philadelphia, Karanis, FHerm., and the Hermopolite Nome list the percentage of the land (private land in the case of Philadelphia, all land in the case of the others) owned by that tenth. For the sake of perspective, the figures for income in modern Honduras (the country with the most unequal distribution of income, and R = .630) and Britain (with a 52

idem., 5-7. 53 There are 234 Hermopolite owners recorded in one quarter of the city; quadrupling would give at most about i ,000. For an estimate of 7,000 houses in Hermopolis, see G. Roeder, Hermopolis, 1929-1939 (1959), 107. Since two quarters are known to have had over 4,200 (East and West City, see SPP v 101), that estimate may be too low. But we do not know how far women listed as landholders may be part of the same households as listed men. And some houses held multiple households. At all events, a total urban population of 35,000 (perhaps too high) would imply 7,000 households. ^ Soltow, op. cit. (n. 16), 126. The exclusions comprise 73 per cent of the landowners. These were, to be sure, 'cottagers', not farmers, averaging about a fifth of an acre each. 55 See Lee Soltow, Toward Income Equalitv in Norway (1965), 61. 56 Within these groups there are also distinctions (not recoverable with present evidence) between those who

own urban property and those who do not, and even those who own property other than their residence and those who own just a place to live. 57 cf. Atkinson, op. cit. (n. u), 247. 58 See Soltow, op. cit. (n. 55); Soltow, op. cit. (n. 16), 128-39; Atkinson, op. cit. (n. n), esp. 22-7, 251; Jan Tinbergen, Income Distribution, Analysis and Policies (1975), 17. For example, various Norwegian towns' income Ginis fell from .47o-.56o in 1865 to .28o-.293 in 1960 (Soltow, op. cit. (n. 55), 17). The US in 1967 is estimated by Atkinson at .380 (income from all sources); Tinbergen (17) gives a drop from .500 to .330 for the US (wages only) from 1903 to 1956, essentially in line with data for other developed countries. Atkinson (25—6) points out that the US and UK had essentially similar income inequality, but the UK had a higher inequality of wealth. 59 Soltow, op. cit. (n. 55), 42, gives wealth—income ratios for Norwegian towns, snowing a decline in the median from 5.4 in 1855 to 1.2 in 1950.

XII 141

L A N D H O L D I N G IN LATE R O M A N E G Y P T : THE D I S T R I B U T I O N OF W E A L T H

relatively equal distribution of income and R = .400) are given.60 A graphic summary of the cumulative effects is given in Fig. 6. DECILE

PHILADELPHIA

KARANIS

FHERM.

HERM.NOME

HONDURAS

UK

1

I.I

1-5

O.2

I.O

1.4

2.O

2

I.Q

3.2

0.4

2.1

1.6

3.1

3 4 5 6 7 8 9 10

2.8 3-9 5-3 7.2 9.6 14.1 19.1 35.1

4.3 5-8 7-7 8.5 11.4 13.7 16.6 27.2

0.6 °-9 i-3 1.7 2.6 4.0 9.7 78.6

3.0 4-3 5-5 6.6 8.3 10.o 13.7 45.5

1.8 2.5 3-4 4.8 6.7 10.i 16.8 50.9

4.2 6.0 7-5 9.1 ii.o 12.9 14.9 29.3

It can be seen that the difference in Gini index between Philadelphia and Karanis derives from Karanis' higher figures for the first 70 per cent of the population and lower figures for the top 30 per cent. It is striking that in the case of Honduras the inequality is produced largely by the concentration in the top 10 per cent; the next two deciles do not (as at Philadelphia) share in the concentration very much. -.—

0.8-i CD 0.7- —

T~

0 0.6

/

|

-—I..-?- - -

0.5

*-' 0.4

~™

§ 0.3

—

~~~~

———

—

~_

_.y--xo\)VTi ev TO) 9Ap(aivoiTR). The date of that document is A. D. 378, and the hand of the present fragment suits the suggested period very well. Moreover, the name Vitalianus was sufficiently uncommon in Egypt to give the coincidence some value.» The last observation is an understatement, as the these three documents are the only attestations of the name in the papyri.7 REMONDON summarized his activity succinctly: «en 378, retire de Parmee, «proprietaire» dans PArsinoite, residant a la ville, il donne en location un lot de quarante aroures de terre a ble . . . et achete en une seule fois mille corbeilles pour ses vendanges.»8 CARRIE describes him as follows: «ce prefet de numerus transplante d'Ascalon a Arsinoe d'Egypte . . . en 359 apparait quelques annees plus tard comme un important proprietaire dans le Fayoum. . . . Vitalianus, Abinnaeus ont trouve sans peine des terres a proximite de leur nouvelle affectation. Leurs offres d'achat font prime, et Ton imagine la perturbation que pouvait creer sur le marche foncier local la presence d'une garnison.»9 Now Hiera Nesos is not particularly near Arsinoe, whether it is the village of that name near Karanis or the one near Tebtunis.10 And 5

Republished as M.Chr. 271 and FIRA III 135; cf. BL 3.10, 5.11. P. Grenf. I 54; cf. BL 1.183 for extensive corrections. 7 F.PREISIGKE, Namenbuch, Heidelberg 1922, s.v. cites also SPP III 208, but the termination of the name is not preserved there, and Vitalius or Vitalinus would also be possible. 8 REMONDON 1965,140. 9 CARRIE 1976, 169. 10 See W. PEREMANS and E. VAN'T DACK, Prosopographica = Studia Hellenistica 9, 1953, 63-64. S.DARIS, in: A. CALDERINI - S.DARIS, Dizionario dei nomi geografici e topografici 6

XIV Military Officers as Landowners in Fourth Century Egypt

49

when one reads the lease closely, it turns out that Vitalianus does not own the land at all. Rather, he himself is a lessee: ac; mi aoi (= cro) £%ic oc re TOV d^toAoyovvToc /crA.The formula is precisely that of a loan in kind such as P.NYU 22 or 24 (to cite only two of numerous examples). • For example, SB I 4504. 7 These papyri were published by R. H. Pierce, Three Demotic Papyri in the Brooklyn Museum (SymbOsl Suppl. 24, Oslo 1972). They are reedited in the Recueil de textes demotiques et bilingues by P. W. Pestman, with collaboration of J. Quaegebeur and R. L. Vos (Leiden 1977), nos. 4-6 (I owe to Professor Pestman's kindness the use of this work in proof).

XV 88

PRICE IN 'SALES ON DELIVERY'

not decisive.8 Indeed, faced with these texts and posing the question of their proper title, Pestman remarks: "Nous croyons du reste que la question que nous avons soulevee n'avait aucun interet pour les personnes interessees a la transaction. Elles avaient en vue une convention de credit et ne se sont surement pas demande si leur transaction constituait un pret ou une vente a livraison diferee/'9 But the critical question of motives can be answered more confidently. Pestman, though admitting that the merchant's occupation might lead one to suppose that his interest was in acquiring the merchandise which would be delivered to him, points out (1) that the quantities involved are too small to be of any real commercial importance, and (2) that the merchant reserves the right to be repaid in cash rather than produce if he wishes. He concludes, "Nous penchons, par consequent, pour la supposition que le marchand n'a pas execute ces transactions en vertu de sa profession, d'autant plus que nous avons des raisons d'admettre que rinitiative de ces transactions emane des debiteurs plutot que du creancier. . . Les debiteurs se sont probablement adresses au marchand parce que dans leur communaute il a du etre un homme relativement aise et d'une certaine puissance financiered10 We will do well to acknowledge that not all transactions need spring from the same motive. In a few cases we can judge that the buyer was principally interested in consuming the produce, as was the buyer of BGL7IV 1055 (Alexandria, 13 B.C.), who purchased a stamnos of milk a day for three months, or in reselling it, as in the case of the wineseller of P.Flor. Ill 314 (Hermopolite, A.D. 428), who purchased 224 knidia of wine. No such interest, however, is discernible in most documents. Since one of the principal merits of Packman's approach is that it takes into consideration the motives of both parties, a summary of her argument will be useful before we proceed further: (1) The first Tetoueis document, SB VI 9603a, is dated 3 February 372, yet it calls for delivery in Epeiph (June-July) of 373. A delay before * The acknowledging party says, dj=k n= j 5wn (n) hd rtb sw wjd 3%, translated by Pestman as "Tu m'as donne la valeur en argent de 3^ artabas de froment frais." This could almost be a literal translation of the typical Greek phraseology. 9 Pestman et al. (supra n.7) II 38-39. 10 Pestman et al. (supra n.7) II 38. Cf. the remarks of Pierce (supra n.7) 85-93; he is attracted by notions of speculation similar to those advanced by Packman, but he concludes that the Egyptian contract is probably an imitation of the Greek and corresponds roughly to the Greek Sdveiov.

XV 89

delivery of more than a year in such a contract is unparalleled, and in loans of commodities which are similar, only one other such period is found, also in a document dated early in 372, P.Vindob.Sijp. 13. The only reasonable explanation of this anomaly is the assumption that the scribes have written the wrong consulship (i.e., vTrareiac instead of fjiercc rfv vTrartiav). Such scribal errors in dating formulas of this period are common. (2) The editors, Day and Forges, took the price of barley in this document at face value and supposed a rise in price from 500 to 600 talents per artaba between February and the date of SB 9603c, December 372. N. Lewis, on the other hand, supposed that the price stated represents a price less interest; that is, that interest was deducted before a price was paid.11 A difference of price is therefore a difference of the time for which the money was used by the borrower; the longer the period, the lower the price. (3) If the true date of SB 9603a is 373, the situation is reversed from what Lewis supposed. The price is inflated by the incorporation of interest, not reduced by deduction of it. The parallel to commodity loans, where the amount that is stated as received is iri fact that owed, i.e. with inclusion of interest, suggests that here too the amount of indebtedness is stated. This is true both for the money and for the goods to be delivered. (4) The specification of price appears generally when there is no penalty clause, and vice versa. The only reasonable explanation of this phenomenon is that the two were essentially equivalent in function. The only explanation of this situation is that a sale was not legally binding until the delivery was completed, and that some provision to protect the buyer in the event the seller did not wish to make delivery was necessary. (5) This variability of the transaction allowed a wide range of possibilities to the two parties for speculation according to their wishes, against possible changes in the value of the goods sold in advance. Packman proceeds to elaborate on the opportunities for speculation opened up by this kind of transaction. One is forcibly reminded in reading her remarks of the operation of a modern options exchange, 11

Lewis remarks briefly on the text in AJP 83 (1962) 185-87.

XV 90

PRICE IN 'SALES ON DELIVERY'

in which there is no intention on the part of most players to exercise the options, but rather that the options themselves, as hedges against price movement in both directions, are repeatedly traded. The scheme proposed by Packman allows for almost as many and as sophisticated tricks as the modern exchange. This brief summary does not do justice to the force and ingenuity of Packman's argument, but I hope that the main lines are clear enough. Before proceeding to BGU 2332 I wish only to observe that this scheme is simply too ingenious. The mentality which is presupposed for the complicated financial maneuvers involved no doubt existed in antiquity, but it must have been rare, for it is alien to the ancient mind in general. It appears to me utterly impossible that a small-time moneylender in Ptolemais Euergetis could have understood this manner of doing business, let alone that a poor peasant in a Fayum village could have made any sense of it. The economic life of the farming class was extremely simple—if often difficult—and I know of no other evidence that members of that class interested themselves in speculation.12 Even if one rejects stage (5) of the argument, as I do along with the consequences suggested in (4), the central tenet of stage (3)—that the price and amount of goods given include interest—is not thereby automatically disproved. We turn therefore to BGU XIII 2332, a loan of money to be repaid in kind, concluded probably in Ptolemais Euergetis on 12 November 374. The text of the critical section, lines 8-15, reads as follows: ojLtoAoyco /coerce T7]vSe TTJV

8

12

ac(f>dXia,v ec^/ceVca Trapa cov Sia XLP°C apyvplov raXarav fjivpidSav filav 8ic^i\€i(a.) 6V[rcc etc r]ifjLr)v oivov evro\7T\Lov evapecrov rrj[c e\7rl rov K€po[v] (/>avr]coiJi€v('r)c) TlfJifjC

KOV^l^CUfJieVVV

TOV

TplTOV

KCcl e77oW[y]/COV TT]V CCTToStOCtV ACrA.

The editor translates, "I acknowledge that in accordance with the present agreement I have received from you in cash twelve thousand talents of silver, being the current price of local, good wine, reduced by one-third, and I shall necessarily make the delivery. .." The critical phrase for us is that in line 13, which is corrected by the 12

For example, cf. M. I. Finley, The Ancient Economy (London 1973) 23, 110-11.

XV 91

editor to KOV^OJJL^C ra> rpirco. For this the editor offers two possible explanations: "1) the price has been reduced because of the deferred delivery; that is to say, were the transaction immediately consummated, Aurelios Hoi could have demanded the full price. Since it is not, he has to take a cut rate. 2) Perhaps this grain is a requisition for the army, therefore sold to it at a reduced price."13 The answer becomes more evident if the central part of the passage is translated correctly. We must render it, "being for the price of good local wine, the price which is current at the time being reduced by the third/' From this it is evident that the price to be used is not decided upon at the time of the contract but is to be whatever is current in Mesore. A hypothetical set of numbers may be useful. Hoi has borrowed from Adelphios 12,000 T. in November. In July, when the wine comes in, Hoi repays the loan with wine. Wine is then 1,000 T. per knidion, and the money would normally buy 12 knidia. But because Hoi must pay interest of 50 per cent, the price is reduced by a third, and Adelphios receives 18 knidia (worth 18,000 T.). The number of units obviously is hypothetical, but the essence of the transaction appears to me to be the only one possible given the text of the papyrus. In other words, interest has been deducted in advance; Hoi obligates himself to pay 18,000 T. on the basis of having received only 12,000 T. The situation is that suggested by Lewis for the Tetoueis documents. This type of transaction is not common among the advance sales, and the new text is the only one to state (in the main portion) that the price is to be reduced. The other examples of the use of the current price at delivery are the following:14 P.Gen. 8 (Dionysias, 141), X[ax]ccvoc7r[€pfjLo]v

rfjc

[e]co[/i]eV^[c]

T€ifJLrjC

SPP III 123 (Arsinoite, VIp), n^v [xoprov rfj rore] Tipfj KOI

X6ya> TOKOV avrov fjLrjviaiajc . . . fjiavSaKW P.Lond. I 113, 6c (p.215) (Arsinoite, VI-VIIp), hay at rfi (f>ouvofji€vr]

13

rifjifj

aroKcl.

'Grain* is a slip for 'wine' here. I would translate airoScocic as 'repayment', so also in line 25. 14 1 exclude several of the examples given by Montevecchi, op.cit. (supra n.l) 131-33, as being dubiously restored or interpreted.

XV 92

PRICE IN 'SALES ON DELIVERY'

Two of these explain what interest is to be charged; in one case it is nothing, in the other, a mandakion (of hay) per month. It has been established by P. W. Pestman (see n.25 infra) that &TOKOC referring to a loan does not indicate that no interest was charged, but rather that no interest was charged on the amount stated, i.e., interest was included in the amount of the loan. In the London piece there was probably interest charged, but it cannot be recovered. In SPP III 123, the rate of interest may not have been evident until the end of the loan, since the amount of interest in kind was fixed but the principal in kind was not. P.Gen. 8 yields no information. The provision is in fact awkward if something to regulate interest is not put in, like that in BGU 2332, and in fact its use is a curious borrowing from a wellknown penalty clause in contracts of this type where the amount to be repaid in kind has been specified but not the amount of money borrowed. Here the borrower, on default, must pay the cash value of some multiple (usually 1.5 or 2) of the sum of produce, the value being determined by that current at the time of default.15 In this situation, the use of the then-current price is a convenient means of providing an equivalence which does not openly show usurious tendencies, where the lender's profit is already secured by the size of the penalty. In the main body of the contract, however, it is difficult to manage unless one wishes, as in the new Berlin text, to reveal indirectly the use of a usurious rate of interest, and most lenders did not want the rate revealed. We may note that a phrase somewhat similar to that about reduction of price occurs in a penalty clause in PSI III 239: lav 8e p,y, A. ...

OCt JJL€ TO V7ToX(oi7TOv) T7JC TLfJiTJC

KCCTO, TTjV (fxXVrjCOfJi(€Vr)v)

TlfJirfv,

KO.I

aicre e^ie Trapaxajpfjcat, COL IK rrjc rifjifjc vo/xt'qu,(aroc) rpirov. That is, an unpaid amount of produce would be repaid in cash with the cash reduced in accounting by a third of its value; the true interest rate would be then 50 per cent as in the Berlin text. When we return to the Tetoueis documents, we notice that they— like such documents generally—state consistently the amount to be repaid but are not consistent about giving the amount borrowed. In the case where the amount borrowed is not stated (SB 9603c), the amount to be repaid is the only amount given in the contract. (There 15 Examples of this phraseology may be found in P.Mil. 4 and 5 (1.5x original); SB VI 9569; P.Osl II 43 (both 2x plus interest); P.Hamb. I 71 (1.5x); SB III 7175 (2x); P.Hamb. I 21 (2x).

XV 93

is no penalty clause.) I take it, therefore, that this is the true amount to be delivered.16 Now 'advance sales', other than those where only the money amount is stated (discussed above), divide into those where only the amount in goods to be delivered is given and those where the price is also stated; the ratio is about 5-2 between these types.17 In the former, as in SB 9603b, we must accept that the amount of goods to be returned is the true amount of the contract. Scholars have long been suspicious of the reticence of these texts about how much cash was advanced, and with reason. The documents try too hard to emphasize that the price was fair and agreed-upon. The standard Byzantine formula is rfjc reAeiccc Kal aglac riprjc, and some variants are stronger; common is the inclusion of some form of cvpfavea) to indicate mutual agreement. Cf. P.Select. 2 (III/IVp): rrjv ptTagv cvfJLa)V7]6[€'Lcav] Kal cvvapecacav rrpoc aAA^Aovc Tip,rj(i>y. One's SUSpicions are also increased by P.Athen. 24 (Arsinoite, A.D. 283), where the amount is stated, and it is called ovcac XOITTCCC ri^rjc KpiOfjc apra^wv rpidKovra. (A similar phrase is found in P.Bad. 25.) One asks, 'balance' after what? Deduction of interest is the natural response, considering BGU 2332. When one turns to the transactions where the price is given, one is struck, in those cases where one can judge from good contemporary parallels,18 by the low prices paid; these also point to a deduction of interest before the amount is stated. Some examples are P.Corn. 2 (Philadelphia, 249 B.C.), where the editors conclude that the profit over fifteen months must have been nearly 100 per cent; P.Mich. XI 608, where 4 solidi buy 18 artabas of barley, 18 of wheat, 4 of vegetable seed and 200 jars of wine; and P.Michael. 35, where a solidus (of 22^ carats) buys 20 artabas of wheat. The editor of this last text was led 16 Packman argues that paragraph 104 of the Gnomon of the Idios Logos (BGU V 1210), [a]Tpvynr)Ta yemj/zara OVK €^6v nI^ S 'I^H'l^^I'^ S 'I^H'l^M'^ 9

9UIBS

9UIBS

'I^M'I^M'^ 01

-jdy ^S

9UIBS '09Q J?£

'09Q OS uminf

91UBS "UBI'IB^I'B 6

-uBi'iB^-B gj: V?wo/¥

9UIBS

»tu^s 9tuBS 808

9UH3S am s ^ 9iuns V = 91

^OQ "(f

S^-SS 'PI ^g~9S 'PI SS~8l 'PI

?5

Ll~ll 'PI OT-T'88i: -;oo 'd

(80S) 9UIBS 9UIBS 9UIBS

^08 uvjjnf

9TUBS 9UIBS

(^ = 9T) JVd£

OS~^I 'PI £1~L *PI

9-T'T£ 'WH'd doudjdfdg

X

HH

X

XIX

a brief note (P. Panop. Beatty 2.215-221 n.) that these exactions are connected with imperial purchases of bullion at fixed prices and are thus not properly speaking a tax. This idea was picked up by Rea in his recent study ; he regards it as « much more attractive » than the aurum tironicum hypothesis, but nonetheless he states that « as far as I can see there is as yet no way of absolutely proving this hypothesis ». The bulk of our texts concerning these collections are receipts given by officials ; in the case of P. Thead. 33, by one official (nome level) to another (village level), in the other cases (P. Mert. I 31 and the Columbia papyri), by a collector at either level to the contributor directly. Since the Merton papyrus and the three Columbia ones all concern the same people and their contributions for the same tax year (year 16= 4, A.D. 307/8) and fall within that year, it will be useful to have the material set forth in tabular form (*). The collectors in these texts are either epimeletai, usually councillors of the municipality, or a tesserarius, a village-level liturgical appointee (2), who, as P. Thead. 33 shows, ultimately turned his collections over to an epimeletes. The motive in these receipts is always given in a phrase something like Kara Oslav nqoara^iv VTZSQ Kr^aedx; aov KO)ju,r]u 'Apivvecp KA,Tiu