Record Longevity in Chinese History
- Evidence from the Wang Genealogy

by Zhongwei Zhao

Records of extreme human longevity are frequently found in many countries. Some of them suggest that in the remote past there were people who lived to very old ages, or even became centenarians. Yet questions such as to what extent these records can be accepted as historically accurate; what was the highest age that was ever reached by a human individual in the past; was there any person who indeed became a centenarian before the 19th century; and when did the human lifespan first break the one hundred year record in the world still need to be adequately examined. Some scholars, working on historical records in a number of western countries, have recently proposed that there was in fact no centenarian before 1800 (Jeune, this monograph). Could this suggestion be true of the population of the world as a whole? This chapter, by presenting a theoretical model and some empirical evidence, attempts to further improve our knowledge on these questions.

The longest human lifespan recorded, mortality level and population size - a model

What is the biological limit of the human lifespan or whether such a limit indeed exists are questions under investigation. However, if we assume that there is such a limit but it is still far away for the human population to reach - say the limit is 150 years - then in a homogeneous population where the risk of death is the same for everyone at birth and how long a person can live is mainly affected by what he or she experiences after birth, the occurrence of centenarians, or the number of people who could reach extremely old ages, would be primarily associated with two factors: mortality level and population size. The first determines at what speed a population of a birth cohort would die out, and the second determines, under a particular mortality level, the actual number of people in that cohort who could have reached a certain age. If we also assume that mortality in human population more or less follows the Gompertz law,1 the relationship can be illustrated by the graphs in Figure 1.

        The curves in Figure 1 are based on Coale-Demeny family of model life tables, Region west, Females. The number of people who could survive to very old ages has been recalculated by using the Gompertz formula given by Coale and Demeny in order to get more accurate results.2 In this figure, the Y axis represents mortality level and is labeled by its outcome - life expectancy at birth. The X axis indicates population size and the log10 scale is used to resolve the difficulty in mapping numbers with huge discrepancies. The curves show the size of a population of a birth cohort which, under the given age specific mortality rates indicated by the level of life expectancy at birth, could produce one person surviving to age 90, 95, 100, and 105. These curves also show how the number of people who could reach very old ages would change in relation to changes in mortality level and population size.3

        The curve representing one person surviving to age 100, for example, suggests that under the mortality level which gives a life expectation at birth of 20 years, out of a birth cohort of about 40 million people perhaps only one person could survive to age 100. When the mortality declines and the life expectancy at birth rises to 30 years, the size of the birth cohort which could possibly produce a centenarian will decrease to about a million. If these curves are taken as roughly representing the surviving experiences of the human population in the past, then the following observations could be made in relation to our understanding of the recent rapid increase in the number of very old people in the world.

        First, the number of people who could reach very old ages is negatively related to mortality level, and positively related to life expectancy at birth. The lower the mortality, the smaller the number of people from which a centenarian could emerge. If a population experiences a force of mortality which gives a life expectancy at birth of 55 years, for example, in a birth cohort of 10 thousand people, only about one person could reach age 100. While under the mortality which gives a life expectancy of 80 years, the size of the birth cohort which is likely to produce a centenarian may be reduced to less than three hundred. In other words, if population size remains unchanged, falling mortality or increasing life expectancy will lead to a greater number of octogenarians, nonagenarians and centenarians.

        Second, the number of people who could reach very old ages is positively related to population size. This not only means that when population size increases N times, the number of those who could survive to a certain age will also increase about N times, but also that those who could reach extremely old ages are much more likely to be found in large populations. Under a mortality level which gives a life expectancy at birth of 20 years, for instance, if a birth cohort consists of 200 thousand people, then there may be only one person who could possibly survive to age 95 and it will be exceedingly unlikely that any person will reach age 100. If the population size increases for 200 times (from 200 thousand to 40 million), however, the number of people who could survive to age 95 will rise to about 200, and among them, some may be able to live to 98 or 99, and centenarians are also likely to emerge.

        Third, the ratio between the number of people reaching age X and those reaching age X+N is not a constant. It changes considerably in responding to a change in mortality level. According to the graph, for example, when average life expectancy at birth is 20 years, among some 200 people who reach age 95 there may be only one person who could survive to age 100. In contrast, when life expectancy at birth rises to 80 years, one in twenty of those reaching age 95 may live to age 100. Although these statistics, which are derived from the model life tables, may not accurately represent the experience of any population, the trend indicated by them is in agreement with the findings of recent empirical research (Kannisto 1988, Thatcher 1992, Vaupel and Lundström 1993).

        The above analysis seems to be helpful in answering the questions set out at the beginning of this chapter. Some scholars have claimed that there might have been no true centenarian before the 19th century or before the onset of industrialization. Such a suggestion is likely to be true, if it is applied to only the populations studied by them, because most of the historical populations which have so far been investigated were in fact relatively small in size and in these populations mortality was fairly high before the nineteenth century. Under such circumstances, centenarians would be extremely difficult to find. Yet, if we increase the size of the population being studied or choose populations with relatively low mortality, the above conclusion may not hold. Indeed, the research on Chinese genealogies which will be reported in the following suggests that in China some people might have become centenarians well before the year 1800, although their number must always have been exceedingly small in the past even in that large country.

Data source - the Wang Genealogy

Differing from in most western countries where parish registers made in the last few centuries are widely available and have consequently become the major source of historical demography, in China thousands of family or lineage genealogies exist and they provide the most important data source for the study of Chinese population history.

        The records concerning the longest human lifespan reported in the next section are taken from the General Genealogy of Wang (Wang Shi Tong Pu). Wang is the surname of those recorded in the genealogy. This genealogy consists of 106 volumes. The first generation in the genealogy, according to the records, lived around 500 BC, and some of those who lived at the end of the last century were listed as the descendants in the eighty-ninth generation. Altogether, around 30,000 people were recorded.

        In the Wang Genealogy, personal information was given for each male individual. These records were arranged by generations and by family or lineage branches. People of the same branch were normally descended from a common ancestor and lived in the same area. Initially, these genealogical records were kept by each particular family or lineage. During the second half of the last century, they were collected together and edited by a person named Wang Yongjing and his sons. The records used in this study were published at the end of the 19th century.

        The social status of those who were recorded in the genealogy was relatively high. This was particularly the case for those who lived in the very early period. It may be said that before the year 1000 AD, the recorded people largely belonged to upper and upper middle classes. Thereafter, the component of those from middle, or perhaps lower, class increased. Nevertheless, people who came from upper middle, or upper, class still made up a substantial part. For these reasons, the social demographic phenomena found in this population may not represent those which might be observed in the Chinese population as a whole.

        Because the Wang Genealogy covered a very long period, consisted of a great number of people, and was originally produced by various families and lineages according to different rules, the information recorded varied considerably from entry to entry. A detailed record normally included the name of the given person, the generation to which he belonged, order of birth, name of his father, precise dates of birth and death, age at death, number of wives (specified as first wife, second wife and concubines), number of sons and daughters (their mothers sometimes being specified), wife's age at death (occasionally with date), and political and academic achievements.

        Most records, however, are not complete and the information provided by them is rather limited. A considerable number of entries contain only name, generation, name of father, and number of sons. This is particularly the case for records made for people who lived in the very early period. Moreover, under-registration of various kinds also existed. Apart from the fact that most women were excluded from the genealogy, two types of persons were apparently under-recorded: those who died young, particularly those who died at very young ages; and those whose information could not be traced. The second type of omission was occasionally noted in the genealogy by "Wu Kao" - not traceable.4

        Two general observations can be made about the Wang Genealogy. Firstly, historical records which cover such a long period have never been found in other countries. Even in China, records of this kind are exceedingly rare. This makes the Wang Genealogy very valuable in the study of social demographic history. Secondly, because of various kinds of under-registration, we will probably never know how many people were omitted from the Wang Genealogy, or the size of the population from which these recorded people came. This limits the usefulness of the Wang Genealogy considerably.

Recorded maximum human lifespan in the Wang Genealogy

In the Wang Genealogy, those who died young (under age 25), especially those who died very young, are apparently under-recorded. Because of that, we are not in a position to investigate infant and child mortality. However, the data analysis suggests that mortality patterns among those who died between age 30 and 75 are similar to those indicated by Coale-Demeny model life table, East region, Level 7 or 8, Males. Among those who were recorded as surviving to 80 and over, the mortality rate calculated from the genealogical records is likely to be lower than what actually existed in this lineage population. This is mainly due to the fact that in the genealogy people who died at very old ages tended to be over represented. Since our major focus is recorded maximum lifespan rather than mortality patterns as such, these details are not discussed here.5

        The limitation imposed by various types of under-registration is apparent. Because of the selective bias which is created by these problems, we should be very cautious whenever we try to use the findings derived from such a sample to generalize the situation of the entire population. Nevertheless, these problems do not prevent us from describing the character of the sample itself. Although the over-representation of those with longer lifespan tends to make the age specific mortality rates less reliable, it has no serious impact on our searching for the maximum length of lifespan reached by human populations in the past, if the recorded ages in the genealogy are accurate.6

        In the Wang Genealogy, dates of birth and death are available for some 2,500 males, and they show a very high degree of consistency with recorded ages at death. Since these records are ostensibly more reliable than others, they have been used as the major data source in this study. Among these 2,500 males, about two hundred were born before the year 1000, some one thousand between the year 1000 and 1500, the remaining thousand being born after the 15th century.

        In this study, the traditional Chinese calendar year has been converted into the modern calendar year and the age of every person has been calculated in the conventional way.7 There were 209 of the sample of nearly 2,500 men who were recorded to have survived to age 80 and over. Half of these 209 lived before 1500, and half of them after that date. Of the 209 men, 22 died between 90 and 94, six between 95 and 99, and one died at age 102 in the year 1513.

        Those who survived for more than 95 years are listed in Table 1. Three of the males died at age 96. One of them was born in the 13th century and the other two in the 15th century. There were also three who were reported to have reached age 97, 98 and 99, and were born in the 12th, 13th and 16th centuries respectively. The one male centenarian lived between 1411 and 1513.8 In addition, there were also six old women who were recorded to have survived to age 95 and over, and are listed in Table 2. They were aged 95, 96, 97 (two persons), 100, and 103, and were born in the 17th, 16th, 15th, 17th, 16th and 13th centuries respectively. Most of these old people, especially these old women, are found in the families or lineages where genealogical records of higher quality were kept. Because there are no individual entries for females in the Wang Genealogy, no dates of birth and dates of death can be used to check the reliability of these reported ages. However, other information may be brought up as reference. For the lady who was recorded to have reached age 100 in the 17th century, for instance, the genealogy recorded that by order of the imperial government an archway was built in front of her house and she was given a title to honour her very impressive success in archiving great longevity.

        According to these records, therefore, centenarians already existed in China at a period as early as the 13th century. Mortality was high in Chinese history, and it was considerably higher than that recorded in European countries at least during the last one or two hundred years. But China had and still has the largest population in the world. Research suggests that the Chinese population had already reached one hundred million by the 11th century, and hundreds of millions of Chinese people lived and died during the last thousand years. In addition to the Wangs, there were also hundreds of thousands of Lis, Zhangs, Lius, Yangs and many more people with other surnames. This huge population size over time may have given China the chance of producing the first few centenarians in the world history.

        Furthermore, average mortality, as a combined result of low living standard, frequent famines, wars and diseases, was certainly high in Chinese history. But because of the better living standard which was enjoyed by the upper class, the long history of Chinese medicine, the prosperity of the early Chinese civilization and socio-economic development, the mortality experienced by certain social groups might not have been so excessive. Indeed, the mortality level prevalent among the Chinese elites could even have been lower than those found among their European counterparts.9 The people recorded in the Wang genealogy were largely a collection of those who belonged to this social group, and some of them were likely to have reached very old ages. For these reasons, it is not impossible that a couple of centenarians could be found in China several hundred years ago.

Reliability of the recorded age

The above discussion inevitably leads to the question to what extent these longevity records can be accepted as historical facts, or to what extent the age records in the Wang Genealogy can be seen as reliable? Age exaggeration has been found in many populations and it has been particularly a problem among old people (Wilmoth and Lundström 1995). It is true that in Chinese history there were always incentives for people to exaggerate their ages and old people were respected by the society. But, disincentives were there as well, and the punishment for exaggerating age could be severe. In the Wang Genealogy, recording people's ages at death seems not to have been affected by these factors, and there was no observable fraud.10

        Admittedly, dates of birth and death are not available for the majority in the Wang Genealogy. But this does not in itself mean that the recorded date of birth, date of death and age at death are unreliable. In Chinese history, detailed records of date and even hours of birth were made for each newborn in many families partly due to the fact that they believed that a person's fortune and destiny were largely determined by the time of his or her birth. Although the major users of these records, until recently, were not demographers but fortune tellers, the fortune tellers and those who asked about their destiny also treated the quality of these records quite seriously. Moreover, if the genealogy compilers had intended to falsify ages for those whose ages at death went unrecorded, the number of these people would be very small rather than large. Data analysis indicates that in spite of the problems caused by various kinds of under-registration, no obvious age heaping can be identified among those whose ages at death were given. Recorded dates of birth and death, and recorded ages of death show a very high degree of consistency.

       Another circumstance indicating the reliability of the recorded ages is that in the Wang Genealogy, as in other Chinese genealogies, dates of birth and of death were registered according to the traditional Chinese calendar rather than modern calendar reckoning. This made it rather difficult for people to forge age for themselves and even more so for their ancestors.

        Chinese history has been divided into some twenty dynasties since the "Xia" period (from about 2100 BC to about 1500 BC). Every dynasty, depending on the length of the time it lasted, was ruled by a number of emperors. Each of them, ascending the throne, would start a "new" period by issuing a new "Nian Hao" - the name of the period to be governed by the new imperial figure. The first year after such a "Nian Hao" began would be year one, and the second year two and so on. When the old ruler died or when the new one took over, the "Nian Hao" would normally be changed and the year would be counted from one again. In some dynasties, an imperial ruler might announce several different "Nian Hao" during the period of his or her control.

        In Chinese history, when the year was recorded, people would normally record both such a "Nian Hao" and the number of years since the given "Nian Hao" started. Some time the number of years was alternatively recorded according to the traditional "Sixty Sexagenary Cycle" method.11 This means that counterfeiting ages for those who lived in the past could turn out to be a somewhat intricate task. In a society where the modern calendar is used, forging a death age and the date of death is a simple matter. All you need to do is simply to decide the date of birth, and then by adding the assumed age at death you will get the date of death. But under the traditional Chinese calendar reckoning system, forging an age at death may have required extensive knowledge of dynastic history and involve some fairly complicated calculations.

        Given the above considerations, it seems reasonable to suggest that although many problems exist in the Wang Genealogy, in most cases recorded ages seem to be reliable, especially those where dates of birth and death are given.

Concluding remarks

Because of high mortality in the past and the under-registration problems which exist in Chinese genealogy data, suspicion about the reliability of these longevity records may exist. Although we should be cautious about the credibility of these records and further investigation should be carried out on other historical materials, it would be unwise to completely reject the suggestion that some people could survive to very old ages or even become centenarians in the past.

        Some people tend to believe that mortality change has been a linear decline - mortality was higher in the past and is lower now, and the earlier the period, the higher the mortality. Because mortality was high in the past, the occurrence of centenarians would be impossible. This suggestion, however, may not be true. The work on the Wang Genealogy shows that over a period of nearly a thousand of years, long-term mortality patterns were fairly stable in the population being studied. Data provided by Liu also indicate that in a much larger lineage population, the mortality level might not have changed or only had some marginal changes from the 16th century to the 18th century, perhaps even to the early 20th century (see Liu 1992 and Zhao 1994). If this indeed is the case, then it would be justifiable to say that what we have found in an 18th century or a 19th century population could also be found in a 15th century or a 14th century population. Similarly, if people could reach an age of a hundred of years in the 18th or the 19th century, there would be no reason why they could not become centenarians in the earlier time, if the population size was large enough. Mortality was high in the past, and this certainly considerably constrained the chance of people surviving to very high ages. But the high mortality did not make it impossible. In this regard, the evidence found in the Qing Imperial Genealogy, which is now being investigated by James Lee and his colleagues, has provided an important reference.12

        As far as the quality of and the detailed information provided by the data are concerned, the Qing Imperial Genealogy may be one of the best historical demographic data sets in the world and is certainly one of the best historical data sets in China. In this genealogy, 96 percent of personal records contain a precise year, month or even day of death; and the proportion of those with precise date of birth are even higher. The records made for males are more complete and more accurate than those for females.

According to Lee and his colleagues, mortality in this population was very high, in spite of the fact that these people were the most privileged in the country. Life expectancy at birth was less than 25 years among those who were born between 1700 and 1720. It was lower than 30 years among those who were born between 1720 and 1750. The mortality experienced by the next eight (ten-year) birth cohorts was slightly lower than that experienced by their predecessors, but life expectancy at birth was still lower than 35 years for most of the birth cohorts. In other words, the life expectancy at birth recorded in these 13 (ten-year) birth cohorts varied between those indicated by Coale-Demeny model life tables, Region west, Level 3 and Level 9, but was never higher. Adult mortality in this population was very high. In these birth cohorts, for example, life expectancy at age 30 and 50 was noticeably lower than those set by the Model life tables, Region west, Level 5 (Lee et al 1994). However, among those recorded people, a couple of centenarians have been found. According to the genealogy, two men reached an age of 100 and 3 survived to 102. In addition to that, there are also 13 men who were recorded to have reached an age between 95 and 100. For all these people, detailed dates of birth and death are available, listed in Table 3.

        In the population selected from the Wang Genealogy and the population recorded in the Qing Imperial Genealogy, the proportion of those surviving to very old ages, as has been addressed, is relatively high. Some researchers may accordingly feel uncertain about the longevity records because at very old ages the mortality pattern observed in these selected populations is not as "regular" as that suggested by the model life table. In this respect, two more points need to be added.

        Firstly, if the relatively high proportion of those surviving to very high ages is primarily caused by the over-representation of those very old rather than by people exaggerating their ages, the validity of the record longevity will not be affected. Secondly, although model life tables have been widely used and indeed the first section of this chapter is primarily based on the Coale-Demeny model life tables, we should not rely on only the model life tables to judge whether the mortality pattern found in a population is reliable. This is particularly important in the study of mortality patterns at very old ages, since the "regular" mortality patterns mapped by model life tables, especially those under very high mortality, are largely obtained through theoretical modelling rather than empirical investigation (because of the lack of reliable empirical data).


        1Benjamin Gompertz suggested that mortality rates increase exponentially at old ages. This suggestion has been referred as the Gompertz law. Whether mortality change among very old people follows this law is a question which is under debate and is being empirically tested. Recently, studies show that this law may not be universally applicable to all species and some even suggest that in some human populations mortality at very old ages may not exponentially increase but level off or decrease after a certain point. However, because the major focus here is whether centenarians could exist at all in the past rather than what law was precisely followed by human mortality, the Gompertz law has been used to produce a theoretical value of the number of very old people which may be achieved under the given mortality level. This value would be the low limit, if mortality does not accelerate exponentially at very old ages but stabilizes at a certain level or slows down. For further discussion, see Gompertz 1825, Barinaga 1991, 1992, Thatcher 1992, Carey et al 1992.
      2The formula is expressed as lx = l80 exp [-(u(80) / k) (ek(x-80)-1)] and is provided in Coale and Demeny (1983 p.20). It should be noted that another formula which they presented for calculating "k" and set out in the same page contains a mistake. The formula should be written as k = log (u(105) / u(77.5)) / 27.5, or k = (log u(105) - log u(77.5)) / 27.5, instead of k = log (u(105) - u(77.5)) / 27.5.
       3These curves suggest that if a population of a birth cohort of a certain size went through a set of age specific mortality rates, which are represented by life expectation at birth in the figure, then among them one person would reach a particular age. It should be aware that the theoretical value of this kind is different from that which could be observed in an actual population where mortality levels have not been constant and the occurrence of centenarians or those surviving to very old people is also affected by many other factors. Nevertheless, such a value can been seen as an average which is likely to be found if demographic conditions similar to our assumptions indeed exist.
        4In most Chinese genealogies, there are no individual records for women, although they may be mentioned as wives in the records of their husbands, or as daughters in the records of their fathers. For discussion of under-registration of other kinds, see Zhao (1994).
        5A detailed discussion of mortality patterns in this lineage population is presented in Zhao (1994).
        6Indeed, under the condition that the recorded ages are accurate and reliable, this disadvantage can even be seen as an advantage. According to figure 1, for example, under a high mortality regime, a population of millions might produce only a small number of nonagenarians and possibly a couple of rare centenarians. Yet, detailed personal records of a population with such a size are not available for any population living before 1750. Even if such data are available, data input and analysis may be very time consuming. In this sense, the selection by the genealogy compiler of those who died at very advanced ages saves us an enormous amount of work.
        7In the genealogy, age at death was counted in the traditional Chinese way - i.e. when a person was born he or she was counted as one year old. In this paper, however, a person's age have been computed by subtracting the date of birth from the date of death.
        8In the Wang Genealogy, there were also several other males who were reported to have become centenarians, but because their dates of birth and death are not available, these cases are not presented here.
        9Analysis of the Wang Genealogy and results published by some other researchers indicate that before the mid-18th century, mortality among some Chinese elite groups might be very close or even lower than those found among the European upper class (Zhao 1994).
        10According to the genealogy, for example, Wang Yongjing, the genealogy compiler, lived to age 54. During his lifetime, he married twice and took in four concubines. His first wife died at age 35. The four concubines died at age 20, 17, 23 and 42 respectively. The age at death of his second wife was not recorded and it was likely that she was still alive when the genealogy was published. Here, the genealogy compilers do not seem to have attempted to exaggerate the death ages for these people at all.
        11This method treats sixty years as a cycle and it uses a particular combination of two Chinese characters to name each year. The same combination of Chinese characters will be used again when a new cycle starts.
        12Soon after the Qing replaced the Ming in 1644, the Office of the Imperial Lineage was established (in 1652) to register lineage members, supervise lineage activities, and maintain the lineage genealogies. The first genealogy appeared in 1662, and by the time of the last update in 1921, 28 editions covering a period of two and half centuries had been produced. The recorded population increased from about 1,000 people to 200,000, with 50,000 were alive in 1921. So far, records of 80,000 individuals from the principle imperial line have been included in a computer database, and the research findings reported are mainly derived from these data. See Lee et al (1993).



This research, which was partly funded by a grant from the Alfred P. Sloan Jr Foundation to the University of Minnesota and by a grant from the U.S. National Institute on Aging to Duke University and Odense University Medical school, has been conducted in the Cambridge Group for the History of Population and Social Structure and the East-West Center of Hawaii. I would like to acknowledge the help and support from the above institutions. I would also like to thank Peter Laslett, James Vaupel, James Lee and Feng Wang for their help and comments.



Updated by V. Castanova,   March 2000