INTRODUCTION

Once a system of regional mints is established at the end of the third century, mint-marks immediately show that coin tended to remain where it was produced. In general the coin commonest in Italy was minted in Italy, the coin commonest in Egypt was minted in Egypt, and so on. In important regions without mints, such as Spain, coin predominantly came from the closest neighbouring areas. There was also supplementation from neighbouring mints if production was insufficient for local needs, as it probably was in Britain and Africa. In small amounts, coin could certainly travel from one end of the empire to the other. But the dominant circulation pattern in this explicit evidence is highly localised.

Evidence before the Late Empire is inexplicit, since there are no local mint-marks on the main coinage. It has even been assumed that under the Principate money-flows were homogeneous and free from localised patterns. But that is dangerously close to an argument from silence, and it has to suppose that radical change in circulation patterns took place between early and Late Empire. In the earlier period, regional patterning is naturally more difficult to study, and co-ordinated attempts to do so are rare. But material for analysis exists even here, and what it suggests about natural patterns of circulation is as localised as in the later period. Thus the case for significant long-term change in the motor patterns of Roman coin-circulation under the Empire remains uncertain.

High mortality placed a heavy burden on Egyptian women; this population's survival depended upon sustaining relatively high overall fertility. As E. A. Wrigley has observed, in societies with very high mortality, “a population could hardly allow private choice since it must mobilize maximum fertility if it is to survive at all.” Roman Egypt typifies such populations. However, as we shall see below, Wrigley's observation must be qualified. Mobilizing “maximum fertility” was certainly not a demographic goal in Roman Egypt; and restraints on fertility, though different in kind and operation from those used after the modern fertility transition, played a major part in the general social survival of Egypt's population.

Overall female fertility

In 211 instances, census returns preserve the age both of a mother and of her child. In these instances, the age of the mother at childbirth can be reconstructed by subtracting the child's from the mother's age, as we have done in Table C. The results for ages 12 to 49 are plotted in Figure 7.1 (p. 137), with four-year moving averages used to smooth the figures. As the graph shows, the peak period of an Egyptian woman's fertility lies between ages 17 and 30; thereafter the number of known births drops off. This graph strikingly confirms the argument in Chapter 6 (at notes 7–9) that Egyptian women married at an early age, usually by their later 'teens.

However, the curve in Figure 7.1 needs adjustment. We want to know the probability that a woman will bear a child at a given age.

At various points in this book, we use statistical methods to test theories about data derived from the census returns. The nature and purpose of these techniques are described in all standard introductions to statistics; we recommend R. E. Kirk, Statistics: An Introduction (3d ed.; 1990), to which we refer below. As a rule, these tests are easily performed either on a statistical calculator or with a computer statistics program.

Common to many statistical tests is a predetermined confidence level that indicates the degree of confidence one uses in accepting or rejecting a given hypothesis (Kirk, pp. 426–446). For data of the kind this book examines, the usual confidence level is 95 percent. If a test indicates significance at a higher than 95 percent confidence level, the statistical odds are less than one in 20 that the result is fortuitous.

Measures of dispersion

In Chapter 6, Section 2, we examine the gap in age between spouses for 78 cases attested in the census returns, where the gap is defined as the husband's age minus the wife's. The average (or arithmetic mean) of the attested gaps is simply their sum divided by the number in the sample (Kirk, pp. 84–87): in this instance, 7.53 years.

However, this average is not meaningful unless we also know the degree of typical dispersion about the mean, that is, the extent to which the attested figures normally vary from the average. The usual measurement of this dispersion is the standard deviation (Kirk, pp. 119–123), an easily determined statistic that measures dispersion by construing the attested figures as a normal bell curve;…

Since its publication just over a decade ago, this book has played a large role in discussions of ancient demography, both in renewed debate about old problems and in the introduction of new topics. To do justice to this literature and see how far it might modify the views we set forth in the first edition of this book would mean rewriting its entire text. We do not judge that the time is ripe for such a wholesale reconsideration. This reprint edition does, however, offer the opportunity to make available to readers in convenient form the results of papyrological work over this decade.

The list of census declarations from Roman Egypt in the first edition of this book had been brought up to date as of 1 November 1993. This Supplement includes material known to us published after that date and before 31 July 2005, with the exception of the 63 households from the census register from an Upper Egyptian city published by us, with full presentation and analysis of the data, in P.Oxy.Census (Roger S. Bagnall, Bruce W. Frier, and Ian C. Rutherford, The Census Register P.Oxy. 984: The Reverse of Pindar's Paeans [Pap.Brux. 29, Brussels 1997]).

Marriage in Roman Egypt differed arrestingly from its modern equivalent. Although marriage was a legal state with consequences especially for the legitimacy of children, the government regulated it lightly. For the most part, among native Egyptians, Greeks, and Roman citizens alike, marriage was a private matter between the spouses (as well as, in many cases, their families); both formation of marriage and its termination through divorce or otherwise usually occurred without the government intervening or even being notified. In discussing the census returns, it is important to bear in mind the relative informality of marriage institutions in Roman Egypt.

For demographers, the sequence of marriage, separation or widowhood, and remarriage (collectively referred to as nuptiality) is important less in itself than for its relationship to the age at which regular sexual relations normally begin and end for women, as well as to the forming and dissolution of families and households.

Age at first marriage

The census returns do not declare the ages at which spouses were married. However, this question can be approached indirectly. Table B summarizes the current marital status of 206 free females and 218 free males whose reported age is 10 or older; data are taken only from returns in which the principal resident family is complete or nearly complete, so that marital status is likely to be determinable. Persons are counted as married if the census return attests a current spouse, regardless whether that spouse is known to be co-resident; and as formerly married if described as such, or if there are other signs of prior marriage, especially the presence in the household of apparently legitimate children.

Although the Egyptian census returns are often poorly preserved, they contain much valuable information on the form of ordinary households; in better-preserved returns, the kinship between household members can virtually always be reconstructed with considerable confidence. As will emerge in subsequent chapters, the form of Egyptian households is of considerable importance to Egyptian demography.

We begin with a brief discussion of Egypt's probable population during the early Roman Empire, then treat in more detail the composition of households, especially the large differences between the metropoleis and villages.

The population of Egypt

Roman Egypt's total population has long been the subject of dispute, mainly because the two principal literary sources contradict each other. The historian Diodorus Siculus, writing toward the end of the first century BC, places current Egyptian population at 3 million (1.31.6–9). By contrast, Josephus, in the latter half of the following century, gives a population of 7.5 million for Egypt exclusive of Alexandria, an estimate allegedly based on the amount collected from the poll tax (BJ 2.385, in a speech). This would imply a total Egyptian population on the order of eight to nine million.

Many historians accept Josephus' estimate as close to accurate; accordingly they denigrate or explain away Diodorus' figure. However, not only is Josephus' estimate of doubtful provenance—avowedly not derived from an actual count, in any case—it also supposes a population level that Egypt would not reattain until the late nineteenth century, after the introduction of perennial irrigation and Egypt's partial integration into European industrial economies. Dominic Rathbone has recently argued that Diodorus deserves more credence than Josephus, and that for various reasons, Roman Egypt's population must in any case have been far lower than Josephus.

In this book, we have tried to reconstruct the most probable demographic characteristics that surviving census returns support for Roman Egypt. As we have repeatedly stressed, some aspects of the demography of Roman Egypt are baffling, and probably destined to remain so, particularly because of the paucity of our evidence. Unless major new discoveries are forthcoming, the difficulties that surround, for instance, infant mortality, male life expectancy, and the Egyptian sex ratio may well remain insoluble except by way of reasonable conjecture.

By contrast, many important aspects of Egyptian demography are, on present evidence, at least relatively clear. These include especially household structure, the female life table, patterns of female and male first marriage, female fertility rates, the usual means of fertility control, and the way in which these features combined to preserve the delicate balance between mortality and fertility. What is perhaps most surprising about these clearer aspects is this: despite persistent inexactitude in the measurement of detail, there is very little about Egyptian demography that would not have been anticipated, very little that lies outside the boundaries of the normal for pre-modern Mediterranean populations. To be sure, local peculiarities obtrude, of which brother-sister marriage is the most conspicuous example; and allowance must also be made for the usual Greco-Roman social and institutional framework with respect to, especially, slavery, marriage, and infanticide and exposure. But what stands out about Roman Egypt is, in the end, not its oddness but its conformity. The very mundaneness of Egypt gives rise to an expectation that less certain aspects of its demographic regime lie also within the normal range.

The Egyptian census returns preserve reported ages for 337 women. These ages are collected in Table A, which lists females resident in villages and nome metropoleis, and the total for each age. Table 4.1, derived from Table A, groups the same females into five-year age groups.

Even casual inspection of Table 4.1 indicates that Egyptian females were subject to heavy mortality. For ages 0 to 24, there are an average of more than 35 females in each five-year age group. By contrast, for ages 55 to 79, this average drops to only five females in each age group. Although the figures in Table 4.1 display many statistical inconsistencies, it appears that only a small proportion of Egyptian females—perhaps as little as a fifth—survived from their teens to their sixties.

To what level of mortality do the figures in Table 4.1 correspond? This chapter attempts to determine this at least approximately. But we begin with a more theoretical introduction to our working methods.

Life tables and stable populations

Life tables were introduced above in Section 1 of Chapter 2, where we discussed the Coale-Demeny Model West, Level 2, for females. Table 4.2 (p. 77) reintroduces that model, but with three additional columns of interconnected figures. As will be recalled, q(x) is the probability that females of exact age x (that is, on their birthday x) will die before their next indicated birthday, usually within five years; l(x) gives the number of survivors to exact age x from a theoretical “cohort” of 100,000 newborn females; and e(x) represents the average life expectancy of females aged exactly x.

The scientific study of human populations

Demography is commonly defined as “the scientific study of human populations, primarily with respect to their size, their structure and their development.” This definition brings out three essential aspects of the discipline. First, its primary concern is with aggregate populations, large groups of people historically bounded by geography and time; a “population” is the basic unit of study. Second, the mode of study is “scientific” in that demographers seek to understand a population objectively, above all through statistics that describe it. Third, a demographic description of a population focuses especially on the numerical size of a population, its composition, and its change over time. What demographers isolate are objective explanations for gross population trends.

There are two main demographic approaches. First, we may look at a given population, like that of the Roman Egypt in the first to third centuries AD, as an entirety, and ask, for instance, how many persons this population had, whether it was growing, how it was geographically distributed, and so on. Second, we may instead study the more basic demographic functions that bring about population change: mortality, the rate at which persons leave the population by dying; fertility, the rate at which persons enter the population through birth; and migration, the rate at which persons enter or leave the population by physically relocating.

The second approach is more interesting because it incidentally raises issues of great social importance. These issues include the level of overall welfare in a given society, its structures of family life, the emphasis it places on bearing and rearing children, its treatment of women and the elderly, and so on.

During the early Roman Empire, the provincial government of Egypt conducted a periodic census of all residents. Among the tens of thousands of documentary papyri from Roman Egypt, there survive just over three hundred census returns filed by ordinary Egyptian declarants. Extant returns run from AD 11/12 down to the last known census in 257/258, but the vast majority date to the second and early third centuries. Their state of preservation varies: some are scraps, but many contain complete or nearly complete registers of Egyptian households. In all, nearly eleven hundred registered persons can now be made out, of whom sex is known for more than a thousand, and age for more than seven hundred. About threequarters of surviving returns come from the Arsinoite and Oxyrhynchite nomes (administrative districts), which lay in Middle Egypt to the southwest of the Nile Delta; other nomes are represented only intermittently.

Although the Egyptian census returns are not free of the flaws that beset all pre-modern censuses, their surprisingly high demographic quality has long been recognized. Early discussion of them from this standpoint culminated in Marcel Hombert and Claude Préaux's classic Recherches sur le recensement dans l'Egypte romaine, published in 1952. Since then, however, research has been spasmodic. The only major demographic contribution is Keith Hopkins's 1980 article on “Brother-Sister Marriage in Roman Egypt.” Hopkins devoted much of this article to exploratory demographic comments on the census returns, which he argued “are far from perfect, but … the best data we have” for ancient populations. Hopkins was the first historian who applied to the returns the sophisticated modelling techniques that modern demographers use when analyzing imperfect data.

Migration is the third major demographic function, along with mortality and fertility. In pre-modern populations, migration over very long distances (between countries, or, in historical empires, between provinces) usually had slight impact since populations were largely sedentary. But internal migration over shorter distances, for instance between an urban center and the surrounding countryside, could play a major role in shaping the demographic characteristics of a population.

It may help to have some advance idea of what we might expect regarding internal migration. Late medieval Tuscany exhibits a fairly typical pattern:

As will emerge below, this pattern is at least partly found in Roman Egypt as well.

The attached list of census declarations from Roman Egypt, brought up to date as of 1 November 1993, has been produced according to a fixed scheme. Households are given numbers consisting of the digits of the first julian year of the regnal year of the census, two letters designating the nome, and a (largely arbitrary) serial number within the group; they are organized in order of these household numbers, i.e. first by census, then by nome, then by household. Persons are given by status groups, and within these in the order in which they are given in the papyrus. Cross-reference numbers operate within the status group unless indicated otherwise, except that the declarant is always referred to as such except when ambiguity would result.

All returns that were used in generating the data base are marked in this Catalogue with an asterisk (*) at the beginning of the entry. We provide stemmata for households in which there is a kin relationship between two or more members, or in which a relationship can plausibly be restored. In the stemmata, a triangle denotes a male; a circle denotes a female; and a diamond means that the person's sex is unknown. We give the person's age where it is known or likely. Between persons, a solid line indicates that the relationship is considered secure; a broken line means that the relationship is reconstructed or uncertain.