For thousands of years human beings have been attempting to increase or decrease their fertility. Modern interest in this history, for reasons that are transparent, has tended to concentrate on ways of limiting fertility. Precise notions about how to do this are first recorded in an Egyptian document of about 1850 BC, the Kahun Medical Papyrus, which includes three prescriptions for vaginal suppositories with allegedly contraceptive properties. There is no reason to think that such prescriptions were then new.

As recently as 1960 a great historian of the Roman Empire, Ronald Syme, could still write that there was “little or nothing on record” about contraception,1 but it is now recognized that artificial means of contraception were discussed at length by Greek and Roman medical writers and were alluded to from time to time in various classical literary works. And the Greeks and Romans, like us, commonly practiced abortion while often censuring it.

A paradox emerges from the evidence we have about contraception in classical antiquity. The methods that medical authorities and other men of learning recommended are regarded by modern scholars as in many cases ineffective, and at their worst as downright ludicrous:

There is also a third kind of phalangium, a hairy spider with an enormous head. When this is cut open, there are said to be found inside two little worms, which tied in deer skin on a woman before sunrise, act as a contraceptive, as Caecilius has told us in his Commentarii. They remain effective for a year.2

Wear the liver of a cat in a tube on the left foot, or the testicles of a cat in a tube around the umbilicus, or else wear part of the uterus of a lioness in a tube of ivory. This is very effective.3

It is generally agreed that some of the contraceptive methods that Greek doctors of the Roman period recommended, particularly the use of spermicidal substances and of other substances that blocked the os uteri, are at times likely to have worked. But so much bad advice is mingled with good that the effects on fertility—this, at any rate, has been widely assumed—cannot have been significant. All but four of the twenty contraceptive techniques mentioned by Dioscorides, the foremost pharmacological writer of the Roman Empire, are said by modern experts to be ineffective.4 The greatest authority of classical medicine, Galen, wrote in the second century AD (in a passage which Riddle does not explicitly discuss) that he was unwilling to discuss contraceptives or abortifacients, since most of them belonged to the category of remedies that were absurdly ineffective or dangerous.5 (In other texts, it is true, he was more forthcoming, and gave specific prescriptions.) Classical methods of abortion have for the most part been described by modern scholars as having been quite primitive.

On the other hand, diverse Roman writers, for instance Musonius Rufus and Juvenal, indicate unequivocally that practical and effective methods of contraception were in use. The paradox cannot be eliminated by saying, though it may well be true, that Musonius and Juvenal were exaggerating the effectiveness of available techniques. As for abortion, texts written during the Roman Empire make it reasonably clear that in some circles at least it was regarded as a practical option.

If the Greeks and Romans failed to identify effective methods of contraception, it would not be altogether surprising. Not even the best classical physicians—though in many respects they were remarkably advanced—understood the relationship between fertility and the menstrual cycle. Soranus of Ephesus, for instance, a physician of the second century, supposed that the period of maximum fertility came in the last days of menstruation (he is usually taken to mean just after menstruation, which makes his mistake appear slightly less extreme, but that is not what he actually says), and that abstinence during those days would prevent conception.6 In ancient conditions, this mistaken view could easily go unrefuted, for the simple reason that women who restricted their sexual relations to the times Soranus regarded as infertile were often lucky and did not become pregnant anyway. The same applied to every nostrum that was supposed to have contraceptive power: it was impossible to know, in any month during which pregnancy did not begin, that the nostrum itself had been ineffective. Since, furthermore, Greek biochemistry was rudimentary in the extreme, and no physician seems to have been inclined to carry out a statistical survey, progress was inevitably halting.

But at this point we need to rein in our arrogance a little. Is it not possible that our ancestors learned something from their experience of these matters? No ancient person could know why conception did not occur in any month in a woman’s life in which she had potentially fertile sexual relations. But the ancients could perfectly well have learned by painful experience which methods did not work. Perhaps we should treat their experience and their intelligence with more respect. Such caution seems all the more necessary since we have tended, until relatively recently, to think of physical impediments—condoms, diaphragms, and IUDs—as the effective means of contraception, whereas most of the methods recommended by the classical medical writers were oral.


By taking the classical physicians seriously, John Riddle resolves the paradox, and successfully overturns established views about contraception and abortion in the classical world. He is able to do so because he looks hard and systematically to find out whether and how ancient remedies could have been rational ones. He is the first scholar to have faced the complex task of determining whether the many substances, mainly plant-derived, to which classical writers attribute contraceptive or abortifacient effects, actually work. He is superbly equipped to do this because, in addition to knowing the classical texts, he has a vast knowledge of botanical science (not the least of the difficulties scholars have faced is the exact identification of the plants to which the classical writers on medicine refer) and he is in command of the medical research which in recent decades has tried to assess the antifertility potential of a great number of plants. Much of this research has been undertaken in developing countries in which the need for inexpensive antifertility drugs is felt most keenly.

Riddle’s study is a true turning point in the history of contraception and abortion, which may have large implications for the history of, among other subjects, the medical and psychic experience of women in antiquity (they apparently had more control over reproduction than anyone had thought), folk medicince, and premodern demography.

What Riddle finds over and over again is that the oral contraceptives and abortifacients known to the Greeks and Romans were both effective and also safer than one might have expected. (His preface, however, warns the reader not to try any of the recipes he describes; it may be the first time such a warning has been thought necessary in a book about ancient medicine.) To take a fairly typical example, Soranus of Ephesus gives the following prescription for an oral contraceptive:7

The seeds of leukoion [=Matthiola incana L. and/or Cheiranthus chieri L: Cruciferae] and myrtle…three obols each; of myrrh…a drachma; of white pepper…two [seedpods]; give to drink with wine for three days.

Much of the research leading to Riddle’s judgement that this recipe was likely to have been effective was carried out by biochemists in India.

There are those who put excessive confidence in folk medicine, but Riddle is not among them.8 He has submitted all the proposed remedies to scrutiny and he is careful to avoid exaggerating their effectiveness. Nonetheless it is very striking that Dioscorides and Soranus, in particular, knew so much about plants and their combined effects, all the more so, as Riddle remarks, when one considers that there exist approximately 750,000 species of plants.

But granted that Dioscorides, Soranus, and others knew of effective drugs, did people use them? The question requires us to ask how knowledge was diffused in antiquity, a world in which books were expensive and illiteracy was widespread. People would not normally have found out how to prevent or terminate pregnancy by reading a text. Riddle elegantly and credibly suggests that the Greek medical writers were recording folk usage rather than, as readers would assume today, the results of scientific research. In all probability, knowledge about contraceptives and abortifacients was mainly transmitted by women. As Riddle imagines it,

A woman could take her herbs—willow, rue, birthwort, or whatever was available locally that she knew—according to instructions from her mother, her adviser, or, less likely, her physician.

But can we tell whether great numbers of people actually attempted to limit their fertility? On the one hand the classical physicians’ numerous recipes presuppose a great deal of human experience. And allusions in nonmedical texts assume that methods of contraception were widely known, and used as a matter of course. Such is the case, for instance, at the end of Ovid’s Amores 3.7, an amusing poem which the Loeb Classical Library once considered too shocking to print even in Latin. At the end of the poem, Ovid’s mistress of the moment gets out of bed to douche herself in order to give her maids the impression that she has had sex and to conceal from them the fact that Ovid has proved to be incapable of performing.

On the other hand, it would be gratuitous, to say the least, to assume that most Romans wanted to limit fertility rather than increase it. We need to know more than Riddle, or anyone else, has so far told us about the central question, people’s intentions. What was regarded, by Romans of different social classes, as the ideal number of children? We do not know. The abandonment of unwanted infants by their parents, which was commonly fatal, must also be taken into account: it seems to have been practiced on a fairly large scale by the inhabitants of the Roman Empire, especially perhaps by its Greek inhabitants. That fact makes it somewhat implausible to suppose that people were acquainted with means of contraception or abortion which they could trust to work. Riddle sees this difficulty, but fails to deal with it adequately.


How safe were the available methods thought to be? On this subject Riddle’s book is least satisfying, for some of the abortion-inducing drugs mentioned in the texts can undoubtedly be fatal if taken in excessive quantities. No wonder Galen was dubious about mentioning them. Amores 2.13 finds Ovid praying for Corinna, whose life has been endangered by an abortion, and who no doubt would have had available the best medical advice. Most women who sought abortions were in some danger of receiving treatment that would threaten their lives. The recipes in the medical texts normally give precise quantities, but even they leave many things vague, such as whether a dose could safely be repeated.

What became of all this knowledge after the disintegration of the Roman Empire? Written information about contraceptives and abortifacients began to thin out, and continued to do so over an astonishingly long period ending only in the nineteenth century. New substances were occasionally recommended, but in the long term available knowledge diminished. The primary, or at least the obvious, reason was the spread of the Christian religion. But the Church’s opposition to abortion and its somewhat less decisive opposition to contraception appear to have been very slow in taking effect. Ecclesiastical teaching was still being consolidated in the eleventh and twelfth centuries, and Petrus Hispanus, who became Pope John XXI in 1276, still, in his Thesaurus Pauperum, seems to give advice on birth control.

Men of learning tended to know less and less about the subject, particularly, according to Riddle, after the professional training of physicians became institutionalized in the thirteenth and fourteenth centuries. The medical curriculum of the later Middle Ages, which derived in the first place from the school at Salerno, an important conduit of classical and Arabic medical knowledge, did not even mention methods of limiting fertility. At least in some places, however, folk knowledge was preserved. In the interrogations of villagers at Montaillou made famous by Le Roy Ladurie, the people questioned knew, or thought they knew, of an herb that prevented conception. Plenty of other evidence shows that in various parts of Europe folk medicine continued to include such information. And with typical energy Riddle discovers that Queen Anne’s lace or wild carrot (daucus carota L.), known as an abortifacient to Dioscorides and others in antiquity, is still prized as a contraceptive in Watauga County, North Carolina, not far from where he teaches.

Powerful and persuasive though it is, Riddle’s book is likely to be criticized by historical demographers because it takes no notice of the great demographic transition which separates the modern Western world, typified by low birth rates and the survival of most children through infancy, from the ancien régime of high fertility and high infant mortality.9 He does not explain how high infant mortality can have coexisted in the ancient world with effective means of contraception.

Indeed, he starts his book somewhat unfortunately by asking, in effect, what caused the population of the Roman Empire to decline in the first centuries following the birth of Christ. This was the wrong place to start, because we simply do not know whether the population increased or decreased, before the great troubles of the third century. The most recent discussion known to me, by the Dutch historian Harry Pleket, leans toward the view that the population of the empire actually increased.10

None of this invalidates Riddle’s conclusions, and we should be prepared to accept that the demographic patterns of the Roman Empire may have been to some extent at odds with current ideas about premodern demography. But it is a pity that Riddle did not give a credible and detailed description of what those patterns would have been had effective contraceptive and abortion techniques been available. Such a discription would, I think, have been possible.

If Riddle is essentially right, Greeks and Romans had at least some control over their fertility. Were there significant demographic consequences? Unfortunately the demography of the Roman Empire has been poorly understood, partly because of scant evidence but partly also because technical skill in demography and technical skill in ancient history are seldom found in the same person. A young ancient historian, Tim Parkin, has now attempted to overcome this problem by studying demography. The result is Demography and Roman Society, in which he explains the elements of this science in admirably clear though occasionally rather patronizing terms. These explanations can do nothing but good, but his own conclusions are anticlimactic.

Parkin writes primarily about what might be called pure demographic issues, those that concern life expectancy and the population’s age structure. In previous attempts to answer these questions, all sorts of information were dragged into service, including epitaphs, poll-tax receipts from Roman Egypt, and the remains of Roman skeletons, as well as the very small quantity of surviving ancient census data (also from Egypt). With crusading zeal, Parkin attempts to show that not much of this material gets us close to an answer. On occasion he is, I would think, preaching to the faithful: most scholars probably take it for granted nowadays that Roman life expectancy cannot be calculated by reference to the numerous surviving epitaphs which give the age at death, chiefly because the epitaphs concern a skewed sample of the population, under-recording the deaths of the poor and the very young. 11

But more can perhaps be made out of the Egyptian census documents than Parkin allows, even though these documents would be regarded as pathetic material by modern demographers, consisting as they do of a few hundred often incompletely preserved census returns of the first to third centuries AD, written in Greek, which report the ages of the persons listed. Yet with suitable precautions, these texts can be made to yield a modest amount of information about the population’s age structure,12 and they seem to confirm what we would expect: that Roman Egypt had a high level of mortality.

Parkin is also hard on the figures about life expectancy which can be extracted from the sole surviving Roman discussion of the subject, an account by the third-century jurist Ulpian, who had a professional interest, because a lawyer might need to calculate the value of a lifetime annuity. The account conveys the Romans’ own ideas about life expectancy at various ages.13 These ideas were undoubtedly crude, but they should not be dismissed as mere guesswork. This evidence suggests that, in Ulpian’s time, life expectancy at birth was under twenty-five years.

Still, Parkin’s skeptical survey of the supposed evidence is acute and learned, and it should be read by anyone concerned with pre-modern demography. There are admittedly some errors: it is misleading, for instance, to say that “Soranus…speak[s] out against both contraception and abortion.” While he appears to disapprove of both when they are used in cases of adultery or to care for one’s good looks, he gives ample information on the subject.14 But Parkin shows that traditional evidence fails to give precise answers to his main questions.

Can they be answered in other ways? For some time scholars have realized that it is possible to arrive at credible demographic models of ancient populations by making use of the relatively precise demographic statistics that have been compiled for the populations of certain non-developed countries of the present century.15 The argument is essentially from analogy: populations with similar levels of health care and economic development, and similar marriage patterns, are also thought to have shared patterns of life expectancy. It has been plausibly suggested, for example, that Ulpian’s figures correspond to the well-documented demography of the island of Mauritius in the 1940s.16 What makes Mauritius a useful comparison is that it combines accurate statistics with low adult life expectancy—which is assumed to be characteristic of the Roman Empire. The chief effect of this approach has been to bring home to us the stunningly high level of mortality in the first year of life which must have prevailed in classical antiquity, even before we take account of the deliberate abandonment of infants—stunningly high from our point of view, that is, for it is hard to generalize about the intensity of the emotional reaction of the Romans to the loss of their very young children.

The standard descriptive form used by historical demographers is the life table, which supplies life expectancy figures, at various ages, for a real or a hypothetical population. Parkin feels the allure of the precise figures that life tables provide, as others have in the past. But in truth he does not succeed in extracting from them anything more than the approximations already offered by his predecessors. His conclusion about Roman life expectancy at birth is the familiar one: that it was probably somewhere between twenty and thirty years. That is a very vague answer from a demographic point of view, since it leaves the average number of a woman’s child-bearing years so unsettled.

The disagreeable truth is that while life tables offer possible age structures of stable populations with given characteristics, such as a given life expectancy at birth, a historian who wishes to apply them to a population like that of the Roman Empire, whose main demographic characteristics are somewhat hypothetical, has to make a choice about which life table to use. The choice inevitably depends in part on assumptions about imponderable factors, such as the quality of childhood nutrition in the Roman world and its relation to childhood mortality. In other words, we cannot be altogether confident that any particular life table based on a modern population can be transferred to an ancient population.

Life tables describing modern populations can guide us to approximate conclusions about ancient populations. But they should be used in combination with other approaches, and not in isolation from the admittedly troublesome evidence from antiquity.

Demography and Roman Society can be read as an introduction to the wider and more discriminating treatment that the demography of the Roman Empire now requires. We should break down the heterogeneous population of the empire, for there are ample reasons to suppose that the demographic characteristics of, say, the senatorial elite, of peasants, and of poor Greek townspeople were very different. Parkin does not do enough to distinguish urban and rural populations. The next study of this subject should try to ascertain, for each important element in the population, what can be discovered about the age of women when they married, the conventionally desirable size of the family, the incidence and effects of the abandonment of infants, weaning practices (very relevant to fertility), nutrition, epidemiology, and population growth and decline. At the end of his book, Parkin comes to the conclusion that fertility declined under the Roman Empire, but he does so abruptly, and without sufficient argument. The next demographers of Rome will have to confront the question of declining fertility again, and, as Riddle has now shown, they will also have to consider the clear likelihood that effective methods of contraception and of inducing abortion were both known and used.

This Issue

November 18, 1993