|
Roger S. Bagnall
The ancient authors give many numbers, of course, but even where these are free from
corruption or distortion they are often no more than guesses. Surviving documents, like those on
papyrus, might seem to be more promising, but even there scholarly hostility has been strong: "In my
opinion, the data furnished by the papyri scarcely lend themselves to statistical investigations," said
one scholar-attacking an article of mine, it should be said.2
To be fair, there are reasons for such skepticism. For example, numismatists have often used the
number of attested coin dies and the amount of duplication of dies in the surviving coins to estimate
how many dies were originally used. Such estimates give a sense of the relative intensity of coinage.
But some scholars have gone beyond this to try to calculate the absolute number of coins originally
produced and on this estimate to base elaborate historical reconstructions. Such views have now
come under heavy fire and seem indefensible: The number of coins produced from a single die can
vary by a factor of 10 to 15. Many elaborate studies of ancient demography have depended on ages at
death reported on tombstones, but critical analysis has shown that simply counting these ages results
in age distributions never found in any known population.3 It
seems that some people, particularly young children, simply didn't get tombstones and are thus
missing from any population described simply on the basis of gravestones with ages. In the face
of such difficulties, one could just give up trying to meld scientific precision with historical inquiry.
There are lots of other ways to write ancient history, some of them familiar since antiquity-political
and military narrative, for example-some the product of the Renaissance and the
Enlightenment-descriptive antiquarian cataloguing of offices, taxes, trades, and the like. More recent
fashions borrow from literary criticism or anthropology, looking at ideologies and mentalities, or
picking up the taste for "microhistory"
developed in the study of more recent periods, where numbers do not matter because one is looking
at a particular case in great depth.4 Or do they?
Microhistory is very much concerned with how individuals shaped their lives in the face of the
conditions and constraints that faced them, and many of those constraints demand quantitative
description, whether of demographic realities or economic possibilities. If we do not know how
young most people died, or how much land they farmed, how can we even begin to understand the
choices that faced them?
The data available for antiquity will never approach the richness of our information for early modern
Europe, but that does not mean that there is no hope of learning some important things. Demography
is a prime example. Studying the census declarations of Roman Egypt, for example,
has allowed us to get a fairly good notion of life expectancy for women, a slightly less precise idea of
the life span of men, and usable figures for such demographic basics as household size, the
percentage of slaves, the number of children the average woman had, and rates of marriage and
remarriage.5 Most of these figures stand up respectably to
standard statistical tests. Because they are based on a scattering of individual census returns,
however, they tell us nothing about the aggregate population. Hell
enistic Egypt (323-30 BCE), by contrast, has few individual census returns but plenty of lists of
households and their members. From these, a forthcoming study is able to establish good figures for
the overall population of one important region of Egypt in the third century BCE and for the
breakdown between urban and rural areas; it also yields much more precise figures for the
male-female ratio than we have for the Roman period.6 But the
lists lack ages, and life expectancy is thus impossible to compute. Neither of these bodies of
evidence thus tells us everything we would like to know. And it would be rash to combine them
without careful analysis, because Roman rule brought profound changes to the society of Egypt. But
together they establish a fair number of the main lines of Egypt's population over a long period. What
is striking from a methodological point of view is that these studies are totally different in approach.
The Roman material is not inherently statistical in character; these declarations become quantitative
information only as the result of our counting and analysis of them. The Hellenistic lists, by contrast,
were compiled precisely in order to allow the government to produce numbers-numbers of people
subject to particular taxes, as it happens-and some of them include totals. Some even draw total
figures from various other documents in order to get an overall picture. The bureaucrats of Egypt
were the statisticians. It would be idle to pretend that quantifying the ancient world is easy. The
two studies of demography required an enormous investment of time and expertise in reading,
correcting, and tabulating the texts in Greek and Egyptian. Models based on information from other
times and places can help greatly, as they have in these demographic investigations, but one still
needs the raw information. The data are not just lying there for anyone to use; most of the time they
must be extracted with great pains, then subjected to analysis and statistical tests, including careful
thought about their representativeness. Not all numbers are significant: If the number of texts of Homer doubles from one century to
another, it matters whether the total number of known papyri doubles at the same time. It is easier to
pretend that this analysis cannot be done-and of course not everything can be studied quantitatively.
But a refusal to use statistics and models eventually comes down to a refusal to look at some of the
most important characteristics of a society, and the study of antiquity cannot afford that kind of
evasion. History may never be a pure science, but science has much to offer history.
1. Jones, A.H.M. The Later Roman Empire 284-602 (Oxford 1964) 1: viii. 2.
Wipszycka, Ewa. "La christianisation de l'Egypte aux IVe-Vle siécles: Aspects sociaux et
ethniques," Aegyptus 68 (1988): 165.
3. See Parkin, T.G. Demography and Roman Society (Baltimore 1992) for a detailed
discussion of these studies.
4. E.g., Keenan, J.G. "A Constantinople Loan, AD 541," BASP 29 (1992): 175-82; on
microhistory generally, see Levi, G. in Burke, P., ea., New Perspectives on Historical
Writing (Cambridge/Oxford 1991): 93-110. 5. Bagnall, R.S. and B.W. Frier. The
Demography of Roman Egypt (Cambridge 1994).
6. Clarysse, W. and D.J. Thompson. Counting the People (Leuven, forthcoming).
But we became classicists so we wouldn't have to deal with numbers,"
protested a colleague. The use of statistical analysis in historical study has had its partisans and its enemies, no matter what period
we look at. But the ancient world is in a class by itself, with what one historian called "the total
absence of statistics."1 Many are happy to have it that way,
perhaps feeling that without quantitative methods, there is no threat of history's turning into a (social)
science.
Related links:
ROGER S. BAGNALL, PhD, is professor of classics and history at Columbia and chairman of the
Classics Department. His research specialty is ancient Greek history and the Roman East; his most
recent book is Egypt in Late Antiquity (Princeton, N.J.: Princeton University Press,
1993, 1996).