1 – The documents

8As far as the historical demography side of our study is concerned, we decided to work on seven localities which all suffered from the plague epidemic that affected Provence from 1720 to 1722. These were Aubagne, Cassis, Puy-Sainte-Réparade, Marseilles, Martigues, Salon-de-Provence and Vitrolles. We based our work on original documents (parish registers, population enumerations, lists of the sick and of the casualties, etc.) that had not been processed until now, or were studied to answer different questions. In all we examined close to 14,000 records concerning these localities (8,000 of which concerned persons who were the victims of a demographic crisis).

9For Martigues the data come from three documents: the general population enumeration made in 1702, that of 1716 and the parish registers (which record baptisms, marriages and burials) of the three parishes within the walls of the urban area which were analysed for the years 1702 to 1725 [1].

10For Aubagne our study was based on two types of documents: the population enumeration made in 1716 and a list of the victims of the 1720-1721 plague epidemic compiled after the end of the epidemic [2].

11For the localities of Cassis, Puy-Sainte-Réparade, Salon-de-Provence and Vitrolles, our study used data from the parish registers during the months of the epidemic (Szabo de Edelenyi, 2000) and the enumeration of 1716.

12The enumeration of 1702 concerns only the town of Martigues. It was conducted in January 1702 upon the orders of the Intendant of Provence to serve as a basis for a redistribution of the poll tax [3]. For the three parishes of the city this enumeration, which was conducted street by street and house by house, mentions for each family the identity, age and occupation of the head of the family; the identity and age of the spouse; the number of children, the ?rst name and age of each, as well as any occupation. Also mentioned sometimes are the identities and the ages of other persons, relatives (brothers, sisters or parents of one of the spouses) or domestic servants living under the same roof. These data may be completed with information concerning the health status of the individuals (maimed, blind, simple-minded) or their socio-economic situation (bene?ting from poor relief, indigent, begging for his bread). This enumeration being for tax purposes, the occupations (sailor, boat master, in the King’s Service…) or estates (working on his own land, day worker, widow…) indicated in the margins of the notebooks were crossed out as the identities were copied onto the poll tax registers (Raufast and Reynaud, 1981).

13Reservations must be expressed concerning the overall value of this document. The ?rst reservation is related to the taxation purpose of the enumeration:

14

“Any enumeration for tax purposes is by nature incomplete and subject to fraud, so anxious were the communities to escape taxation, a feeling shared by all their members”.

15It would also seem that this document is incomplete (Paoli, 1971) in several respects: the enumeration apparently only covered part of the villages in the countryside, and a few streets seem to be missing. Lastly, this enumeration, in principle individual and nominal, concerns the de jure and not the de facto population, which means that all the foreigners in Martigues (even the Provençaux) are not included [4]. In spite of these reservations this document is of great value in providing information about the population structure of this small urban community before the advent of a series of demographic crises.

16Of the enumeration of 1716, carried out upon the request of Intendant Cardin-Lebret, all that remains are the population size and the number of families of each locality in Provence. This enumeration was discovered and partly published by one of the authors of this article (Biraben, 1975).

17For each of the localities concerned by this study there is a list of the victims of the plague epidemic of 1720-1722. In most cases this document is incorporated in the death registers (except for Aubagne). It starts as a normal death registration and then takes the form of a list of victims established after the epidemic.

2 – Demographic data

18The town of Martigues had a population of 5,664 inhabitants in 1702 and of 5,886 in 1716. The plague started in the town at the beginning of the month of November 1720 and ended in June 1721. In all, 2,150 inhabitants of Martigues died in this epidemic (De Villeneuve, 1826) and we have found the death certi?cates of 2,134 people in the parish registers.

19Aubagne had a population of 3,980 in 1716. The plague raged in this town from August 1720 to October 1721 (Barthélemy, 1889) and claimed 2,114 casualties.

20In 1716, Cassis had a population of 2,980. This locality was hit by the plague a ?rst time between 21 July 1720 and 31 January 1721 (26 victims) and a second time between 16 September 1721 and 3 March 1722 (216 victims).

21Puy-Sainte-Réparade, which had a population of 760 in 1716, was hit by the epidemic from 29 August 1720 to 26 January 1721 (43 victims).

22The town of Salon-de-Provence, with a population estimated at 4,185 souls in 1716, suffered from the epidemic from 1 November 1720 to 26 October 1721; 921 inhabitants perished.

23Vitrolles, with a population estimated at 684 inhabitants in 1716, was declared plague-stricken from 2 August 1720 to 1 April 1721; 210 people died.

24Finally, the data concerning the plague in Marseilles come from a list of people hospitalized at the Couvent de l’Observance during the epidemic relapse experienced by the city between May and August 1722 (Signoli, 1998). This list was partly destroyed in a ?re in 1943. We completed our data thanks to a search through the papers of Intendant Lebret.

3 – Structure and growth of the population of Martigues and Aubagne between 1702 and 1725

25The structure of the population of Martigues was established on the basis of the 1702 enumeration. We know that the enumeration was started in the beginning of January 1702, but the present condition of the document makes it impossible to know the exact date at which it was completed. We have assumed that it was ?nished by 31 January 1702. Analysis of the age distribution points to age heaping on digits ending in ?ve, and even more on those ending in zero. We therefore distributed the population of Martigues by ten-year age groups rather than by individual years of age, in order to reduce the bias at least for the ages that end in ?ve (Figure 1).

26To determine the size of the population of Martigues before each demographic crisis we started with the 1702 enumeration and projected the population from February 1702 until 31 December 1725 by subtracting the intervening deaths and adding the intervening births month by month; these data were taken from the parish registers. This method, developed by J. Bourgeois-Pichat (1970), presents the drawback of not taking migration into account. We assume that migration is quantitatively negligible in this context (see footnote 4).

Figure 1

Population pyramid of Martigues at the 1702 enumeration (individuals of reported age, and including nurslings)

Population pyramid of Martigues at the 1702 enumeration (individuals of reported age, and including nurslings)

27We were able to compare the estimated population size for 1716 to the one provided by the enumeration of 1716. There was only a difference of 45 individuals (in excess in our calculation), representing an error of 0.8% which could be due to the lack of information as to the precise dates of both enumerations.

28The same method was used to estimate the population of Aubagne on the basis of the 1716 enumeration. For the period before 1716 we used a backward projection along the same lines (adding deaths and subtracting baptisms, and assuming that migration was negligible).

4 – Compiling a database of plague victims

29For each of the seven communities studied we found the list of victims of the 1720-1722 epidemic. These documents provide the family name, ?rst name, sex, and age of the victims, the date and the place of death, the identities of the father and the mother or the partner, the occupation and miscellaneous comments (relationship with other persons: nephew, parent-in-law…). In all, this information was available for 5,195 individuals. This allows a demographic approach of the plague, but it also makes it possible to follow the role of “family links” in the spread of the epidemic. We counted for each plague victim the number of links with other victims of the epidemic. Two details need to be clari?ed concerning the method used:

• On the one hand, we adopted an epidemic notion of the family. We therefore considered the master and his apprentice, or the members of a religious community as part of the same family.
• On the other hand, the study is based on the list of victims of the epidemic. We do not know the identity of the patients who survived and this prevents us from evaluating their role in the spread of the disease.

1 – Determining sex among adults

32We used primarily the ilium or hip bone which is the most discriminating element in the skeleton since it contains practically all the information relating to sexual dimorphism (Iscan, 1989; Bruzek, 1992; Bruzek, 2002). With appropriate methods it is possible to establish the sex of the individual in 95% of cases.

33When the hip bones were badly preserved or totally absent we determined sex by using other elements of the skeleton, the cranium in particular. We used a list of 17 morphological parameters established on the basis of classic descriptions [5] (Nemeskeri, 1963; Ferembach et al.; 1979; Buikstra and Ubelaker, 1989; Leonetti, 1998).

2 – Determining the age of immature subjects

34For children, the estimation of the age is based on three indicators: the study of the different stages of dental irruption according to Ubelaker’s tables (1978), the study of the diaphysial lengths of the long bones (Martin and Saller, 1957; Stloukal and Hanakova, 1978; Sundick, 1978; Fazekas and Kosa, 1978; Adalian et al., 2002) and the study of the different stages of epiphysial fusion of the long bones (Brikner, 1980; White and Folkens, 1991). These three methods were used simultaneously when the state of preservation allowed, thus enabling us to obtain more precise estimates by comparing results.

35Immature subjects were divided among the following age groups: 0-1, 1-4, 5-9, 10-14 and 15-19. This distribution is based on the mean age of the children and does not take into account the margin of error associated with these methods. If the margin of error were taken into account, the majority of subjects would straddle two age groups (86.3% for the Observance series and 71.4% for Délos). The resulting biases have already been demonstrated (Masset, 1973a and 1973b; Sellier, 1996) and new methods are being developed (Buchet et al., 2001).

3 – Determining the age of the adults

36The fusion of the spheno-occipital synchrondrosis (sutura sphenooccipitalis) marks the passage to adulthood. It was also possible to use the presence of certain degenerative pathologies as an indicator of the age at the time of death. We considered a whole set of parameters (degree of synostosis, dental wear, cervical osteoarthritis, osteopenia) to classify each individual within broad categories of age: young adult (Y), mature-young adult (M-Y), mature adult (M), mature-elderly adult (M-E), elderly adult (E). Given the lack of precision of the techniques for determining the individual age of the adult skeleton we decided to use the qualitative method, which is based on the following categories:

• Young adult (Y): closed spheno-occipital synchondrosis, epiphyseal clavicle, third molar fully out;
• Mature-young adult (M-Y): dental wear and/or possible beginning of cervical arthritis;
• Mature adult (M): average dental wear with some loss of teeth, partial sutural synostosis, moderate vertebral arthritis;
• Mature-elderly adult (M-E): marked dental wear, more important dental loss, advanced sutural synostosis, marked vertebral arthritis or hyperostosis, thinning of corticals;
• Elderly adult (E): substantial dental wear and loss, total or subtotal sutural synostosis, marked thinning of the corticals and rarefaction of the trabecular bone, spinal compression, marked arthritis or vertebral hyperostosis.

1 – Historical background

37The plague epidemic of 1720-1722 started with the arrival in Marseilles on 25 May 1720 of an infected ship from the Ports of the Levant (Middle East). Following a series of administrative errors the contagion spread from the in?rmaries throughout the city where it claimed its ?rst casualty on 20th June (Bertrand, 1723; Gaffarel and Duranty, 1911; Carrière et al., 1968). Rapidly, and in spite of quarantine measures no doubt taken too late, the epidemic gained momentum in the villages around the city; and later spread throughout Provence and to other regions (the Comtat-Venaissin, the foothills of the Alps and the Massif Central). Communities were unequally affected (only one casualty in Lambesc for 2,680 inhabitants, 1,068 deaths in La Valette for 1,600 inhabitants). Some localities, such as Fuveau, although situated in the middle of the plague zone, remained unaffected by contagion. In all, this epidemic affected 242 localities and resulted in 119,811 deaths out of a population of 394,369 inhabitants before the epidemic (Biraben, 1975). The last pockets of infection died out at the end of 1722 (Avignon, Chasserades and Orange). During this period several cities infected a ?rst time experienced phases of epidemic resurgence [6] or relapse [7].

2 – Epidemiological approach based on historical archives

38The examination of all the documents presented earlier enabled us to identify a total of 5,195 victims of the plague epidemic of 1720-1722 for the seven localities [8] selected for this study.

Distribution by sex

39The study of the distribution by sex of the casualties reveals that among the adult population (20 years and older) women seem to have been more affected than men (Table 1). The only exception to this trend is the sample of the casualties from the epidemic relapse in Marseilles. However, it should be noted that the latter consists of only 61 individuals. For Martigues the population enumeration of 1702 and the follow-up of population covering the period until 1 January 1720 indicates that the population included at these dates 51.3% and 48.8% of women respectively. Therefore the 57.5% of women among the adult casualties of the epidemic of 1720-1721 may represent excess female mortality. However, a comparative statistical test of the two distributions (Chi square) does not con?rm this conclusion (differences not signi?cant at the 99% level).

40We have noted elsewhere that women also represented a majority of the casualties in previous epidemics of the plague, for example in Toulon in 1664-1665 (52.3% of the casualties aged over 20 were women).

41Obviously the explanation for this inequality cannot be any epidemiological selectiveness on the part of Yersinia pestis. This mortality difference is linked, either to an over representation of women in the living population subjected to the epidemic risk — which may be of structural origin or result from the temporary absence of men from the community (see for example the importance of seamen in Martigues) — or to the entrusting of certain burial practices to women (the washing and shrouding of the bodies).

42However, this observation cannot be generalized: other studies concerning the same epidemic showed a slight preponderance of men among the victims. In Saint-Rémy-de-Provence out of the 426 deaths during the epidemic of 1720-1721, 50.7% were men (Biraben, 1975).

Table 1

Distribution by sex of the adult victims of the plague epidemic of 1720-1722 in selected localities of Provence and comparison with the structure of the population of Martigues (%)

Men Women Plague victims(1) Aubagne 44.1 55.9 Puy-Sainte-Réparade 32.1 67.9 Salon-de-Provence 43.2 56.8 Vitrolles 39.8 60.2 Marseilles 58.1 41.9 Martigues 42.5 57.5 Population of Martigues 1702 (enumerated) 48.7 51.3 1720 (estimated) 51.2 48.8 (1)Population aged over 20. Source: Signoli, 1998.

Distribution by sex of the adult victims of the plague epidemic of 1720-1722 in selected localities of Provence and comparison with the structure of the population of Martigues (%)

Distribution by age

43The ages in the lists of plague casualties reveal a preference for digits ending in 5 or 0, more marked for women more than for men. As in the case of the living population, we chose to work with ten-year age groups.

44The distribution of deaths by age groups reveals a plague-related death toll which was very different from the one observed during a period of mortality unaffected by crisis (average for the years 1702 to 1719, leaving out the crisis years of 1705, 1709-1710, 1720-1721, see Figure 2). During a plague epidemic, individuals aged between 10 and 69 are exposed to far greater risks than in “normal” periods. On the other hand, the two extreme age groups (0-9 and 70 and over), who usually pay the heaviest toll, seem proportionally less affected during plague periods (Figure 2c).

45It may also be noted that the age distribution of the victims may be different during the secondary phases (resurgence or relapse). The proportion of young victims (notably those aged 0-4) is weaker in the resurgence or relapse phases than during the ?rst epidemic episode (Figure 3).

Figure 3

Plague victims aged 0 to 9, by single year of age, by epidemic phase (peak or resurgence) in Aubagne and Salon-de-Provence in 1720-1721 (as % of all victims)

Plague victims aged 0 to 9, by single year of age, by epidemic phase (peak or resurgence) in Aubagne and Salon-de-Provence in 1720-1721 (as % of all victims)

The demographic toll of the plague is very different from the one induced by other mortality crises affecting the ?rst quarter of the eighteenth century. The parish registers of Martigues reveal two other major crises: a smallpox epidemic in 1705 and a famine in 1709-1710 (Figure 4). The age distribution of deaths is different in each of the three crises (Figure 2). However, the general tendency brings observed mortality closer to the epidemic crisis of 1705 than to that of the “Grand Hyver” (“Great Winter”, 1709-1710) and to that observed outside of the years of crisis. Compared with the latter, the crisis of 1705 led to excess mortality among the ?rst two age groups (and below-average mortality among the older ages). The opposite happened during the crisis of 1709-1710: the oldest subjects were affected more than in “normal” times. Finally, the mortality pro?le of the plague is totally different from mortality without crisis as well as from mortality observed during the two previous crises.

Figure 2

Age distribution of the victims of the smallpox epidemic of 1705 (2a), of the famine of 1709-1710 (2b), and of the plague epidemic of 1720-1721 (2c) in Martigues and comparison with the age distribution of the population in 1720 and that of the “normal” mortality recorded from 1702 to 1719 (%)

Age distribution of the victims of the smallpox epidemic of 1705 (2a), of the famine of 1709-1710 (2b), and of the plague epidemic of 1720-1721 (2c) in Martigues and comparison with the age distribution of the population in 1720 and that of the “normal” mortality recorded from 1702 to 1719 (%)

Figure 4

Number of baptisms, marriages and burials in Martigues between 1702 and 1725

Number of baptisms, marriages and burials in Martigues between 1702 and 1725

46The comparison between the mortality pro?le in the absence of a crisis, the pro?le of mortality from the plague (1720-1721) and the structure of the living population in 1720 (estimated by projection, cf. supra) shows that the plague does not select its casualties. In fact the age distribution of the victims of Yersinia pestis follows the age distribution of the living population quite closely but departs from that of mortality in the absence of a crisis (Figure 2c). The victims of the plague constitute a representative sample of the structure by age (but not by sex) of the pre-epidemic population [9].

Monthly movement

47We were also able to establish the monthly distribution of deaths from the plague. Figure 5 shows the speed with which the plague killed most of its victims: in two months approximately half the total number of deaths had already occurred in Aubagne, Martigues and Vitrolles; in three months this proportion reached close to two thirds (64.9% in Aubagne, 68.9% in Martigues and 62.6% in Vitrolles). Inversely the last two months of the epidemic claimed few victims, even in cases of epidemic resurgence or relapse (0.4% of the deaths in Aubagne, 0.9% in Martigues, 8.3% in Salon-de-Provence and 2.6% in Vitrolles).

Figure 5

Distribution of plague deaths by month during the epidemic of 1720-1721 in Aubagne, Martigues, Salon-de-Provence and Vitrolles (%)

Distribution of plague deaths by month during the epidemic of 1720-1721 in Aubagne, Martigues, Salon-de-Provence and Vitrolles (%)

Note: A resurgence of the disease may be observed, starting with the fourth month in Vitrolles, and with the eighth month in Aubagne and Salon-de-Provence.

48However, the importance of the plague mortality rate (number of deaths from the plague in relation to the size of the population before the epidemic) varies greatly between the communities and is not proportional to the number of inhabitants (cf. supra and Table 2).

Table 2

Death rate from the plague in different localities of Provence during the epidemic of 1720-1722

Number of inhabitants at the 1716 enumeration Number of inhabitants estimated at the beginning of the plague epidemic Number of victims Death rate (per 1,000 inhabitants) Aubagne 3,980 4,065 1,801 443 Cassis 2,980 3,034 214 71 Puy-Sainte-Réparade 760 817 42 51 Martigues 5,888 6,090 2,150 353 Salon-de-Provence 4,185 4,177 960 230 Vitrolles 684 807 196 243 Source: Biraben, 1975.

Death rate from the plague in different localities of Provence during the epidemic of 1720-1722

Family relationships

49The quality of the documents examined ?nally enabled us to assess the importance of “family” links (cf. supra) in the spread of the epidemic within a community. The results of this study show their extreme importance in the progression of a disease as contagious as the plague. Thus in Martigues at least one link with another victim could be established for 61.6% of the casualties of the epidemic (Figure 6). This proportion was 71.5% in Aubagne, 56% in Salon-de-Provence, 55% in Puy-Sainte-Réparade and 47% in Vitrolles. The frequency of these links varies according to the nature of the epidemic phase: during a relapse or resurgence the role of family links seems less important than during the peak of the epidemic. For example, in Marseilles during the relapse of the spring of 1722 we were able to show that nearly 70% of the victims had no relationship with another person who died during this epidemic phase. In the absence of lists of the sick we were not able to go any further and measure the contagiousness of the patients having recovered or the degree of lethality.

Figure 6

Distribution of the victims of the plague epidemic of 1720-1721 in Martigues, by number of “family” links established with other victims of the epidemic (%)

Distribution of the victims of the plague epidemic of 1720-1721 in Martigues, by number of “family” links established with other victims of the epidemic (%)

3 – The contribution of archaeology and of the techniques of molecular biology

50Archaeo-anthropological analyses now make it possible to look simultaneously at the archaeological facts and at the modes of burial. Only ?eld anthropology offers this opportunity to understand how the bodies were handled during major demographic crises.

51Few burial sites of plague victims have been studied in a global manner, from the ?eld to the laboratory. To date only three sites satisfy these conditions: the pit of the Observance (Marseilles), the Délos site (Martigues) and the cemetery of the plague in?rmary of Fédons (Lambesc) dating from the end of the sixteenth century (Reynaud, 1996a and 1996b; Signoli, 1998). The cemetery of Lariey, in the municipality of Puy-Saint-Pierre (Hautes Alpes), which is contemporary of the epidemic of 16281632, is currently being studied (Signoli, 2001; Tzortzis et al., 2002).

52In the study of demographic crises the plague is an original theme since among all the epidemic crises, those ascribable to Yersinia pestis present at least two particularities: on the one hand the speed and intensity of the mortality toll on the population (requiring an adaptation of burial rites); on the other hand, the absence of selection of the victims in relation to their age. Thus the paleodemographic pro?le of a burial site of plague victims offers a snapshot of the age distribution of the living population before the epidemic.

4 – The contribution of osteological data

57Paleodemographic data from the two osteoarchaeological series obtained from the sites of Délos and the Observance were taken from doctoral research (Signoli, 1998). They were drawn from the two samples by the same observer and according to the same protocol. The state of preservation of the hipbones made it possible to determine the sex of the adults for 91% of the subjects from the Observance site and 71% of those from the Délos site. Distribution by sex of the exhumed skeletons would tend to indicate a slight excess male mortality at the Observance and Délos. It should be stressed, however, that the difference observed in terms of sex ratio is not substantial for the Observance site (+ 5.5 points for male individuals) and that the under-representation of female individuals in the Délos series may be related to the number of subjects for whom it was impossible to determine the sex (38.5% of the sample) as well as to the small size of this series.

58Determining the age of the immature subjects was possible for all of the individuals of the Observance and for all of the subjects of Délos. Concerning the adults, the estimation of age at death, based on qualitative criteria, was possible for 91% of the skeletons of the Observance site and for 66% at Délos. The distribution obtained reveals an anomaly in the composition of the sample from Marseilles, where immature subjects aged under a year are absent and those of the second age group (1-4 years) appear to be clearly underrepresented. This anomaly is a common feature in traditional cemeteries, but it should not be found in the burial sites of plague victims, where theoretically all of the victims are buried. The sample from Marseilles is peculiar in this respect.

59Research conducted jointly on the osteological series from the burial sites of plague victims and the archive documents enabled us to compare our paleodemographic results with those of historical demography. Thus, in Martigues the same proportions of adults and immature subjects were found in the Délos cemetery as those computed on the basis of the historical archives (54% and 60% of adults respectively). This similarity clearly con?rms that this mass grave was created at the peak of an epidemic [10]. Similarly, at the Observance site (mass grave contemporary of the epidemic relapse in Marseilles in 1722), the paleodemographic pro?le resembles the age distribution of the victims registered in the hospital archives (Figure 7). The total absence of the 0-1 year age group is noteworthy, in the list of casualties as well as among the skeletons exhumed. During a relapse, the 1-4 age group is also under-represented compared with the peak periods (cf. supra). That is why our result differs from the plague mortality pro?les that are usually published and which only concern the more lethal epidemic phases (Mallet, 1835; El Kordi, 1970; Hollingsworth and Hollingsworth, 1971; Biraben, 1975).

60We believe that there are two main hypotheses that could explain this under-representation of the youngest individuals in the mass grave in Marseilles:

• A selection at the recruitment level. During the ?rst phase of the epidemic of 1720 the children infected with the plague were in hospitals reserved for them. It may be assumed that this was also the case for the epidemic relapse of 1722 and thus that very few children and newborn had been hospitalized at the Observance. However, the list of people hospitalized in this requisitioned convent shows that children died there between May and June 1720 [11].
• The epidemiological hypothesis. The Observance pit having been dug and used during the epidemic relapse of 1722, the small number of children aged 1 to 10 could be explained by the very high toll on the population during the ?rst epidemic phase of 1720-1721 and by the impossibility of reconstituting these age groups before the spring of 1722.

Figure 7

Age distribution of the victims of the epidemic relapse of 1722 in Marseilles according to historical data (list of victims) and to paleodemographic data (Observance pit) (%)

Age distribution of the victims of the epidemic relapse of 1722 in Marseilles according to historical data (list of victims) and to paleodemographic data (Observance pit) (%)

Note: The 0-1 year age group was totally absent in both cases.