1The major subsistence crises, such as those of 1693 or 1709, were characterized by an exceptional increase in grain prices, coinciding with a no less exceptional upsurge in deaths and a decline in conceptions. But these phenomena cannot be clearly detected unless the counts are made by crop year, and not by calendar year. While crises of this type disappeared in the first half of the eighteenth century, price increases continued to influence demographic behaviours in a manner which, though less visible, was none the less considerable.
2“Through our various investigations, we have obtained proof that the years when the price of corn was highest were also those in which mortality was most severe and illness most common”. It was in 1766, in a memoir entitled Réflexions sur la valeur du blé tant en France qu’en Angleterre depuis 1674 jusqu’en 1764 [Reflections on the value of corn in France and England from 1674 to 1764], later published in Messance’s Recherches sur la population [Research on population] that a key problem was first raised: that of the effect of subsistence crises on the demography of France under the Ancien Régime. A highly complex problem, indeed, which can be broken down into numerous questions, of which some have no answer, at least in the form in which they are habitually posed, while others may be resolved over the shorter or longer term by means of research yet to be undertaken, but whose principle must first be established.
4How can mortality due to subsistence crises be measured? We note the scientific caution exercised by the author of the Réflexions. He starts out from a statement of fact, namely that years of exceptional corn prices coincide with years of exceptional mortality, but goes on to point out that these years were also years of exceptional morbidity. It would thus be rather pointless in statistical terms to seek a specific difference between such closely linked phenomena: mortality from starvation alone, mortality from disease but attributable to malnutrition, and last, mortality by contagion, this contagion being inseparable from the state of dearth which contributed not only to the development of disease but also to its spread. In the modern era, apart from the extreme case of “physiological misery”, what share of deaths can be attributed to food supply problems when the officially reported causes of death make no suggestion of this?
5We can, on the other hand, define with precision the years of exceptional mortality in which excess mortality can be linked to a subsistence crisis. These years are easy to identify, since the scale of the phenomena is so large that concurring testimonies abound. Even historians with little interest in studying economic and social realities cannot be unacquainted with events such as those of 1693 or 1709. Indeed, the numerous monographs at our disposal leave no doubt as to the existence of a causal relationship between price rises, poverty and death.
6But difficulties arise if we try to identify more clearly or to quantify this mortality. Statistics on population change existed during the Ancien Regime, but not until 1772, when the major mortality crises due to an exceptional rise in corn prices were a thing of the past. To study these crises, we need to move further back in time, to the reign of Louis XIV and earlier, when no such statistics existed. However, we do possess the registers of births, marriages and deaths - a vital source for all retrospective demographic studies. They are sporadic and often of doubtful quality up to 1667, but well- conserved and of generally high quality from 1667. The final years of Louis XIVth’s reign thus appear to be the most favourable - or in any case the least unfavourable - period for research of this kind.
7But a trap awaits us, which is so glaring that one wonders why so many excellent scholars have fallen into it, and so serious that it must be very clearly pointed out. We believe we know what we are saying when we talk about crisis years. But what practical reality does the word “year” refer to? Be they meteorological, agricultural, economic or demographic, the fluctuations we analyse are difficult to fit into the arbitrary framework of a “calendar year” starting on 1 January and ending on 31 December. Here, the difficulty is considerable. Exceptional corn price rises occur naturally within the crop year, but are severely distorted if considered in the framework of a calendar year.  But experience shows that the same applies to the mortality peaks observed during years of dearth. A rapid month-on-month analysis of the parish registers is sufficient to prove this point. Examination of deaths from 1 August to 31 July reveals increases that are much less visible under the habitual counting method.
8As a consequence, otherwise laudable publications such as those of Oursel (introduction to volume V of the Inventaire des archives de Dijon), of Brossard for Bourg-en-Bresse, and of Faidherbe for Roubaix, have become worthless, at least with regard to the question that interests us. If, as we believe, a more extensive extraction of data in the parish registers is to be undertaken, in a collective form and on a national scale, then we must require that the original records, which will need to be conserved, include a monthly count and that the published results provide totals calculated by crop year.
10The study of conceptions provides a further argument in favour of our proposed reform. The fall in conceptions seems to have escaped the attention of scholars working on historical crises. It is nonetheless an indisputable and symptomatic fact. It can be observed by shifting baptisms backwards by nine months to follow numbers of conceptions month by month. We can thus obtain a graph that is doubly characteristic of a subsistence crisis year, in which an abrupt upsurge in deaths coincides with an equally abrupt decrease in conceptions. 
11We may thus consider that the characteristic index of the crisis is the ratio of deaths to conceptions or, put another way, the percentage of deaths with respect to conceptions counted, if not monthly, then at least in the timeframe of a crop year. This index can then be compared, not against the price of corn itself, but against percentages that reveal the strength of the increase with respect to the immediately preceding time period. In the exceptional years that interest us, the increase produced a brutal effect, a shock effect, very distinct from the social effects of other economic fluctuations. It was felt above all by the lower classes who were living “from hand to mouth”; hence its immediate demographic repercussions, whose rapidity and intensity may evoke surprise. While wages and non-wage incomes were slow to adapt to a price increase - indeed, any price rise, even moderate, produced hardship - we can estimate that after several years this increase was absorbed, and that even if prices remained high, they were affordable to many. Moreover, the effects of scarcity that persisted without worsening were lessened by the fact that the weakest members of the population died within the first few months. Conversely, the opportunities made available to labourers during periods of low prices were often wasted by many among them. The texts that discuss such questions are often too partial to be trusted. While they very likely exaggerate the thoughtlessness and negligence of these “poor idlers”, there is no doubt that after a period of low prices, a sudden increase was a rude awakening for many.
12This is our justification for choosing the method used to plot the graphs presented here, in which the percentages of deaths with respect to conceptions in Dijon and in the Gien region are compared with the percentages of wheat prices in Rozoy-en-Brie with respect to the median prices observed in the five preceding years. We note that the fluctuations in percentages of deaths in Dijon are greatly attenuated by the artificial transformation operated upon the data calculated by calendar year in order to render them comparable with those recorded by crop year.
13This said, the results are quite clear and the comparison between two time periods, 1680-1713 on the one hand, and 1755-1789 on the other, is striking. The reign of Louis XIV was punctuated by subsistence crises of such a clearly exceptional nature that this characteristic alone would be sufficient to differentiate them. Correlatively, the ratio of deaths to conceptions presents peaks which, when the data were collected using the method described above, are of a comparable, and no less exceptional, intensity. The national character of the crisis is indisputable, and replacing the prices quoted on the Rozoyen-Brie market with the prices of Dijon would not make the concordance any more significant.
14Under Louis XV, and even more so under Louis XVI, everything changed. There was no longer any apparent correlation between price spikes and demographic indices. While there was still a demographic problem of subsistence, it was of a very different order of magnitude, and this difference in quantity is already, in itself, a difference of quality. Between the era of deadly crises and the era of latent crises, a revolution took place, a major revolution that can only be mentioned briefly here and which remains to be studied.
15Yet so far, we have simply used illustration to highlight known facts, and made a statistical review of the habitual historical documentation. Can we hope to go further one day, and measure the demographic consequences of subsistence crises? At first sight, this would call for extensive counts in diverse regions, since we would very likely detect major local variations in the degree of dearth. But before undertaking a project on this scale, we need to make a more detailed analysis of the phenomena at play.
16“I buried twelve hundred corpses this dreadful year, dogs were eating the dead bodies they found scattered along the roadside”. Let us not dwell upon the macabre scenes conjured up by this note which begins the register of Gien-le-Vieil for 1709. But, after a thorough count, we find only 241 deaths recorded between January and December, plus a further 17 who are mentioned anonymously, sometimes with their place of origin, but sometimes with no mention other than “poor beggar”. In any case, we are far from the number of burials given in the initial note. Did the priest have an over-lively imagination? More probably, the figure of 1,200 was a gross approximation. Many deaths were not counted in the registers - deaths of strangers recorded neither in their parish of origin nor at their place of death, and hence documented nowhere.
17To understand the importance of this observation, we must bear in mind that the historical population of France included, even in normal times, an immense proportion of errants (vagabonds), an expression borrowed from Mr Georges Lefebvre and which is the evocative title of one of the chapters of his book on the La grande peur de 1789 [The Great Fear of 1789]. One could almost say that there were two peoples: the sedentary and the nomadic. And all the evidence points to the conclusion that the same coefficients of mortality, fertility and nuptiality cannot be applied to both populations. It was precisely the onset of famine that modified these coefficients, by swelling the ranks of the homeless, with professional beggars and odd-job men being joined by the newly dispossessed who, from then on, often remained permanently so.
18In any case, it is impossible to make an accurate count of this floating population, so any direct enumeration of losses due to famine is bound to be flawed.
19Are we at an impasse, then? In reality, only experience will tell us. With marriages, baptisms and deaths, we have three data series spread over time: we can always try to make use of them. Had the Ancien Regime administrators made accurate and frequent exhaustive censuses, the problem as posed here would be resolved. Of that there is no doubt. The difference observed between two censuses would give a clear enough picture of the effects of scarcity. For want of a better solution, can we make up for this absence of data by estimating population change? We have seen that this cannot be done for the actual years in which the crises occurred. But in normal times, even the most wretched vagabond rarely died without his death being registered, and even the least legitimate children were baptized. By working on sufficiently varied indicators and over a long enough period, it is certainly possible to gain an idea of population growth or decline. To the extent that the coefficients of nuptiality, disease and mortality can be considered as relatively stable, each of the three data series can be used as a basis for evaluation.
20But it is clear that each of these coefficients is subject to secular change - of which the current period provides us with many memorable examples - and, moreover, may fluctuate momentarily under temporary influences of an accidental or cyclical nature. Scarcity upsets the composition of human populations and, as a result, the overall value of the coefficients. However difficult it may be to determine age-specific mortality, the sampling we have undertaken appears to indicate that mortality in years of dearth was quite different in this respect to that of normal years. Scarcity not only modified the age distribution after the crisis was over, but also substantially modified not only the number of marriageable individuals and the number of households capable of bearing children, but also the mean physical quality of each generation whose disease resistance was momentarily increased as a result of rigorous natural selection.
21It is even more important, we believe, to consider the compensatory phenomena that occur in the aftermath of the crisis. I imagine that no-one would dream of calculating a country’s demographic losses after a large- scale modern war by looking at pre- and post-conflict marriage statistics. However, over a relatively long period, the coefficient of nuptiality, in our country at least, is among the least unstable. But everyone knows that the months that follow the cessation of hostilities are marked by a flurry of “delayed” weddings. The calculation would thus point to the absurd conclusion of a population increase.
22Yet several phenomena of this kind can easily be detected in the aftermath of major subsistence crises by observing simple series of raw data. Deaths practically always fall sharply, so sharply in fact that it is difficult to explain this fall in relation to population size alone. We can only conclude that scarcity brought forward deaths that would have occurred in the following years. Conversely, we observe not only the existence of delayed weddings, but also a much larger number of conceptions. On our graphs, this surge in births, combined with the decrease in deaths, contributes, after its sudden rise, to an abrupt decrease in the ratio between the two. Finally, given the magnitude of these compensatory mechanisms, one wonders whether these terrible demographic crises were not, demographically speaking, surmounted within relatively few years, in which case, efforts to calculate their intensity by extrapolating upon previous conditions are quite vain.
24We are hardly in a position to avoid this objection, given the additional argument that a large share of the victims belonged to a very distinct demographic category, destined to disappear gradually without trace. This was a category where fertility was certainly low and infant mortality high, a category afflicted daily, even in normal years, by crises that had little impact on the sturdy bodies of ploughmen rooted in the land.
25Among other qualities, the author of the Reflexions cited above had the merit of conducting a hospital survey. While the survey was small, one aspect of its findings is worthy of mention. In years of rising corn prices, the ratio of hospital deaths to deaths in the rest of the population was much higher than in a normal year. And this was true not only dining the terrible price rises under Louis XIV, but also under Louis XV. In the years 1740 and 1741, there were 48,858 deaths in Paris, a quarter more than normal. But at the Hotel-Dieu hospital, in these same two years, 15,085 deaths were counted, a number well over one-and-a-half times the 9,796 deaths recorded in two ordinary years. Note, however, that the “hospital” of the seventeenth and eighteenth centuries was very different from the institution bearing that name today. It looked after sick people, but these people represented the morbidity of only part of the population. They were all penniless - those who were not “poor” in the strongest sense of the term, i.e. indigent, did not go to hospital. Scarcity, which took a massive toll upon these human beings, on the roads, in their hovels, in the barns where they sought refuge, and in the hospitals, simply concentrated within a few dramatic months a history which, at other times, played out through insignificant episodes of daily destitution.
26The great subsistence crisis, which, despite the confusion created by the expression, one is tempted to designate by its traditional name of “famine”, disappeared under Louis XV. Yet it was not until much later, in the second half of the eighteenth century, that France saw a major upsurge in population growth, announcing the arrival of a new era. Indeed, severe crisis and latent scarcity existed side by side. This latent scarcity persisted even when market prices were not manifestly excessive. Given its permanent nature, this is perhaps the demographically most interesting aspect of the subsistence problem. But can one ever hope to address it? The year of scarcity, prior to Louis XV. is a precise item of data which can be used, up to a certain point, to make rigorous calculations. Beyond that, we find a complex situation of endemic unemployment and growing debt culminating in the seizure and abandonment of farms, in which, without doubt, the price of corn played a role. But it did not kill, either immediately or in a massive sweep; it was a slow process of erosion.