Patterns of Advanced Age Mortality in the Medieval Village Tirup
by Jesper L. Boldsen
The general Medieval rural Danish population rarely entered contemporary written records. These people formed the basis of most of the events which took place, but they have left pitifully few individual marks. The drama of the life of the individual Medieval peasant escapes our comprehension. The marks of their lives and labour can be seen in the Romanesque churches scattered throughout the country and in the names of their settlements. However, the only direct source of knowledge about the individual man, woman and child in the Middle Ages are the skeletons of the people themselves. The unique historical character of each individual life story is generated in a confrontation between the general period and group specific conditions of living and the actual course of events that formed life. Some life history events like birth and death have general importance more or less independent of time and place. But in spite of this, the perception of such events is strongly dependent on ideological fabric creating meaning in the life of the individual in the present as well as in the past and on the less elusive conditions of living and producing.
The demography of the Medieval Danish population is one of the few areas where it is presently possible to approach a thorough description of one objective side of the frame of living in the past. Benedictow (1993) has given an overview over some general trends in the demography of Medieval communities in Scandinavia. Part of this work builds on case studies of osteodemography and population structure of the individual finds (Boldsen 1979 and 1984, Brøste 1945, Gejvall 1960, Holk 1970 and 1987, Högberg et al. 1987, Møller-Christensen 1953, Persson and Persson 1981, Persson 1976, Siven 1991a and 1991b, Sjøvold 1978 and Tkocz 1985). In spite of the fairly large number of case studies of Scandinavian Medieval demography only three totally excavated rural parish cemeteries from the Nordic area have been published, Westerhus in Sweden (Gejvall 1960), Thjorhilde's Church in Greenland (Lynnerup 1995) and Tirup in Denmark (Kieffer-Olsen, Boldsen and Pentz 1986).
The validity of results from palaeodemographic research has been seriously questioned both locally (eg Olsen 1982) and in the general biological anthropological literature (eg Bocquet-Appel and Masset 1982). The controversy following some of this criticism has led to the appearance of papers pointing to other problems of the application of quantitative osteological methods in the study of growth- and health- related processes in past communities (Wood, Milner, Harpending and Weiss 1992 and Saunders and Hoppa 1993). Work is in progress trying to solve some of the problems pointed to by Wood et al. (1992) and Saunders and Hoppa (1993) and utilizing Scandinavian skeletal samples for this end. The purpose of the present chapter is to analyse aspects of the pattern of adult mortality in the Early Medieval Tirup community in eastern Jutland, Denmark, with the aim of gleaning information on the probability of surviving to advanced ages.
Material and methods
The Tirup cemetery was in use from around AD 1100 to after 1300 (Kieffer-Olsen 1993). Only a small part of the area possibly containing the remains of the village to which the church and the cemetery belonged has been investigated. However, the lack of reference to the site in late Medieval written sources indicate that the place and community lost its independent importance by the middle of the 14th century. The name of the community, Tirup, gives a clue to the foundation of the village. The -rup ending of the name indicates a -torp type of village. The -torp type villages were founded in the Viking Age and during the early Middle Ages by moving existing farms and establishing new ones in areas marginal to older villages. The majority of these -torp villages were probably founded in the 11th and 12th century AD (Sawyer 1988). This means that it is likely that the excavated cemetery spans the whole history of the Tirup community. The area which came to form the cemetery had not been in profane use during the Viking age and the earliest parts of the Middle Ages. Judging from the ceramics and structures found under the cemetery the area had not been in regular use since the Bronze Age (Kieffer-Olsen, Boldsen and Pentz 1986). These circumstances indicate that the skeletons found in the Tirup cemetery are unique in representing the total population of a small village (on average some 70 inhabitants - cf. Boldsen 1994) throughout its whole existence of some 200 years.
The Tirup cemetery was excavated in 1984. The area used for burials was surrounded by a well-defined ditch. The whole area within the ditch and a strip of 5-10 meters outside of it was excavated. No graves were found outside the circumscribed area. Evidence of 618 burials were found (Kieffer-Olsen, Boldsen and Pentz 1986). It is assumed that around 100 infant burials were not recovered due to the grave not penetrating beneath the top soil (Boldsen, Kieffer-Olsen and Pentz, work in progress). This chapter is only concerned with adult mortality. This means that the absence of some infant burials in the material does not affect the results. The material for the present analyses consists of 116 male and 95 female skeletons from Tirup.
The material analysed here consists of skeletons in which it has been possible to determine age at death and sex. The indicators of sex are very vague for individuals younger than 18-20 years at death. This means that only skeletons clearly indicating an age above 18 years are included in the analyses. The methods of the determination of sex and of age at death are described by Boldsen (1988). The empirical mortality profiles have been reconstructed from the two sets of age at death intervals (male and female) using a method first described by Turnbull (1978) and developed for the reconstruction of palaeodemographic profiles by Boldsen (1984 and 1988). This method treats the skeletal data as interval censored data. The method gives unbiased estimates of the distribution of age at death but it does not facilitate rigorous statistical testing.
Under the stationary population model the cumulative distribution of age at death corresponds to one minus the survival function. Mortality rates estimated from the mortality profiles estimated in this way are virtually independent of natural population growth or decline; but they are vulnerable to age and sex specific migration. The Tirup community is assumed to have gone through an initial phase of natural immigration mediated population growth followed possibly after a period of fluctuating population size by a depopulation phase. Nothing indicates that the community went extinct in a demographic catastrophe (Boldsen, Kieffer-Olsen and Pentz, work in progress). The Tirup community was probably one of the rural communities that supplied the young adult immigrants found in some Medieval urban cemeteries like Sct. Mikkel in Viborg some 80 km north of Tirup (cf. Boldsen 1984). It has been estimated that as much as 40% of adult females buried in this cemetery were immigrants from surrounding rural communities like Tirup. However, the Medieval urban communities contained a maximum of 10% of the population and the migrants consisted of women under the age of 22 years. As the estimates of the mortality rates are independent of events (death and/or migration) taking place prior to the age under study this means that emigration from the community had a negligible effect on the mortality rates estimated from the mortality profiles.
The endpoints of the age intervals forming the data for the analyses in this chapter show a clear digit preference. The vast majority of endpoints (both upper and lower) have a last digit of '0' or '5'. To correct for the bias introduced by this digit preference the mortality profiles have been smoothed out in the way described by Boldsen (1988). The yearly mortality rates estimated directly from mortality profiles are quite unstable even after having smoothed the profiles. In order to make the mortality rates more stable they have been estimated as ten year moving averages as follows:
where 'µ(t)' is the mortality rate at age 't', 'M(t)' is the value of the smoothed mortality profile at age 't' and 'ln' is the natural (base e) logarithm. For comparison the mortality rates of the contemporary Danish populations (1990-91) have been estimated in the same way just using the survival function rather than the mortality profile as the basis for the calculations (Statistisk Årbog 1993).
Results and discussion
The basic differences in the mortality patterns among the peasants from Tirup and recent Danes are clearly illustrated in Figure 1.
This figure illustrates the empirical mortality profile for the Tirup cemetery conditioned on age at death over 20 years and the survival function conditional on survival to age 20 years for recent Danes. It is well-known that survival is higher for both sexes and at all ages for the recent Danes than is was for the people of Tirup. It has also been shown before that early adult survival was greater in males than in females in Tirup. The fact that the Tirup mortality profiles are based on relatively few observations is illustrated by the generally unsmooth curves that these functions draw.
The curves in Figure 1 have been translated to the mortality rates illustrated on a logarithmic scale in Figure 2.
This figure illustrates the general similarity and the great difference between the patterns of mortality in the Tirup community and in Denmark today. All four curves in Figure 2 show evidence of an increase of the mortality rate from young adulthood (20-30 years) to later adulthood (50-60 years). In contemporary Denmark this development is monotonous whereas in Tirup the youngest adults (ages around 25 years) experienced a higher mortality rate than did people in their early thirties. This is just a minor difference between the two sets of mortality rates. The most striking difference is that the mortality rates in Tirup were never less than five times as high as the mortality rates for the same sex and age in contemporary Denmark.
The surplus mortality in the Tirup community is illustrated in Figure 3. It appears that female surplus mortality in Tirup was larger than male surplus mortality in all ages from 25 to 55 years; but the curves converge from age 40 and onward reaching very similar levels at 55 (mortality rate ratio of 8.9 for females versus 7.6 for males). The determination of age at death is at best very inaccurate and the number of skeletons available for study is small for ages over 50 years in all pre-modern societies. The general flat curve for the male mortality rate ratio for ages over 35 years and the converges of the male and female curves at the upper end of the age spectrum illustrated in Figure 3 indicates that it might be fruitful to study old age mortality in a community like Tirup by multiplying the mortality rates for recent Danes by a constant.
Following this reasoning Figure 4 probably illustrates the best possible reconstruction of the survival function for a Medieval village community in Denmark conditioned on survival to age 50 years. It might be a point of interest that using the mortality rate ratio factor 8 the chance of surviving to age 100 was 3.8*10-19 for males and 7.1*10-16 for females given that the person had lived to reach age 50 (see Vaupel and Jeune, in this monograph, for discussion of this calculation). Less than 25 percent of the males and 15 percent of the females in Tirup reached age 50. The mean survival time at age 50 is 25.6 years for males and 29.9 years for females in contemporary Denmark. Following the same method used to reach the figures behind Figure 4 it can be calculated that the mean survival time in the Tirup community at age 50 was 9.8 and 12.6 years for males and females, respectively.
The Tirup village probably experiences some 700 births through its whole history. This means that an expected 34 males and 25 females would have reached the age of 60 years before dying. Of these 5 men and 8 women are expected to have reached the age of 70 years. The expected number of both men and women reaching the age of 80 years is less than one (0.03 males and 0.4 females). A reasonable estimate of the number of people born to the rural Danish population through the Middle Ages is close to 20,000,000 (based on a crude mortality rate of 0.05 and an average population of between 500,000 and 1,000,000 people in a 500 year period). Of these in excess of 1/4 of a million would have reached the age of 70 years, around 12,000 would have reached 80 and only four of them (all females) are expected to have reached the age of 90 years. The estimated number of persons reaching the really high age of 100 years is 8.8*10-10. This means that the chance for any member of the rural Medieval Danish community or of any community with a similar pattern of mortality of living a full century is effectively nil.
Wilmoth (in this monograph) has given a very strict definition of the emergence of the first centenarians. He requires a 99% probability of at least one centenarian appearing in the world population each century. This corresponds to a mean number of centenarians each century of close to five (4.6). Following this criterion a world population with the pattern of mortality seen in the Medieval Danish village of Tirup should consist of more than 1017 people. This many people have never existed and it must be concluded that the Tirup mortality regime would never have produced even a single centenarian. The difference between this conclusion and the conclusion reached by Wilmoth (in this monograph) is probably not a consequence of the empirical data used for the estimation of the chance to reach 100 years. Both approaches vastly extrapolate any results that could be reached on empirical grounds. This extrapolation is most serious for the oldest age mortality, i.e. mortality from 90 to 100 years of age. This means that although the present analysis reaches a conclusion incompatible with the one reached by Wilmoth (in this monograph), the differences are solely due to differences in assumptions, and I for one would not insist that my own more empirical approach gives more reliable results than the more model-orientated approach used by Wilmoth. Further, the analyses indicate that the solution to the contradiction lies in the acquisition and analysis of more and more reliable data on mortality in antiquity.
Updated by V. Castanova, March 2000