Record Longevity in Swedish Cohorts Born since 1700

People reaching an unusually high age have always been of particular interest. This is especially true for those who attained the age of 100 or more. Many tales have been told about very old people in Sweden. We can take Jon Andersson as an example. He died 18 April 1729. According to the death register he was born 18 February 1582. This means he was no less than 147 years old at death. This high age was of course commented upon and Jon was said to be the oldest person in Sweden ever where the age could be verified (Valentin 1966). Another example is Nils �hrberg who died 12 October 1816, 116 years old according to the death register. He was married three times. Nils seems to have survived both wives and children. He was a sexton until the age of 108 when he had to retire (Valentin 1966). There are many more tales like these about persons who got very old.

The question is to what extent data on age can be relied upon. A person living to the age of 147 is of course impossible, especially as far back in time as 1729. The oldest authenticated person ever in the world is Madame Jeanne Calment. She passed her 120th birthday in February 1995. A person 116 years old at the beginning of the 19th century seems impossible too. There are many more persons reported as more than 100 years old at death during the 1749-1900 period. In this chapter I will try to answer the question how reliable the Swedish population data really are.

For more than 300 years each parish in Sweden has kept a complete and continually updated register of its population. From 1686 it was compulsory for each parish to keep registers on births, deaths, migration, marriages and divorces as well as a population register. In some parishes this registration had started long before 1686. In other parishes the registration seems to have started at a later date.

Apart from the usual church registers Sweden had a very special register, "husf�rh�rsl�ngden" (the catechetical examination register). This register was originally introduced to keep track of literacy, understanding of the bible and other facts of interest to the clergy. This register after some time turned into a full-fledged population register including all persons living in the parish. Just like a census the register was arranged by household. Among other facts the date of birth was registered for each person. What makes the register very special is the fact that it was continuously updated. That is, the general population registered showed the actual population (de jure) living in the parish at any time (providing that the clergy updated the register regularly as they were supposed to do).

The various church registers were tightly linked. When a birth was registered the child was entered in the population register too. When a person moved out of the parish the migration was registered in the migration register, and he was removed from the population register. From time to time the clergy made some mistakes. Some, if not most, of these errors were detected in the yearly updates of the taxation register. Quality in the church registers gradually improved and by 1860 data can be considered as virtually correct. One reason for this was the rising interest in population matters among scholars and the general public. The clergy became aware of the importance of keeping the registers accurately. Another reason was the very strict control of all statistics that was introduced in 1860. The compilation of population statistics was from that year and onwards carried out by Statistics Sweden. The compilation was based on copies of all parish registers sent to Statistics Sweden. This made it possible for the first time to make a very careful check of all data. When in doubt the local parish was contacted and data corrected if necessary. Age was for the first time centrally calculated based on register information. The only remaining problem was, and still is, that all migration was not registered. In principle we have the same kind of base for the population statistics today as we had in 1749. In 1946 population registration was made a concern of the state. The church continued to keep the base, the local registers of the population up to 1991 when this task was taken over by the local tax offices. The system was computerised in 1967 and the introduction of personal identity numbers in 1947 made the computerisation efficient.

The special interest in the oldest old can be seen in the statistical tables. When the collection of population statistics started in 1749, deaths were to be reported in five-year age groups up to the age of 90. For all 90 years of age or older each death had to be specified by age and sex in a special part of the form reserved for comments. It was not uncommon that the clergy added to these more detailed facts a short memoir of the deceased. That is, detailed information on centenarians is available in Sweden for a period of more than 240 years.

For the 1749-1859 period population statistics were put together in each parish. The clergy was responsible for this task and they used information from the church registers. Before the statistical tables were sent to the forerunner of Statistics Sweden they passed several authorities. The parish tables were sent to the rectorial district where data for all parishes were summarised into one table for the rectorial district. That table in turn was sent to the rural deanery and so on. In this process some information naturally can have been left out. Especially extra comments that were not explicitly asked for can have disappeared in this process. According to Lublin (1957), however, the county tables seem to include most if not all comments and detailed information of those who were 90 years of age or older. However, it was not quite clear how to fill in the form. Deaths were to be reported in five-year age classes. The classes were of the form ...75-79, 80-85, 85-90. Was a person who was exactly 90 years old at death to be reported in the 85-90 age group or not? This possible source of misunderstanding was more or less eliminated in 1774 when new forms were introduced. From 1802 there was no longer room for any misunderstandings. The five-year age groups 0-5,5-10 .. up to 95-100 asked for, were in the table followed by the one-year age groups 101,102 ... up to 115. A person exactly 100 years old was no longer by mistake added to the 95-100 age group but to the 101 group (100 up to but not including age 101).

Apart from misunderstandings over how to use the table resulting in minor errors the more serious error remained. At least up to 1786 the clergy had rarely, if ever, information in the registers on date of birth for centenarians who died. The deceased were born before 1686 when registration of all births in a special birth register started. In order to calculate age the clergy had to rely on information from relatives only. This of course meant that the reported age at death was sometimes exaggerated, especially for very old persons. The table below shows both the absolute number of centenarians and the relative number per 1 million of the total population for the 1751-1990 period.

According to the official statistics the number of deceased centenarians was very high in the 1750s. The absolute number of deaths as well as the relative number then decreased to a minimum around 1835 and then slowly started to increase again. This development is not what one would expect and clearly indicates an error in data.

Period	Number	Per 1 million of total population
1751-1760	449	24
1761-1770	252	13
1771-1780	160	8
1781-1790	155	7
1791-1800	82	4
1801-1810	75	4
1811-1820	38	2
1821-1830	56	2
1831-1840	31	1
1841-1850	32	1
1851-1860	30	1
1861-1870	34	1
1871-1880	43	1
1881-1890	43	1
1891-1900	63	1
1901-1910	75	2
1911-1920	154	3
1921-1930	252	4
1931-1940	303	5
1941-1950	275	4
1951-1960	374	5
1961-1970	584	8
1971-1980	1156	14
1981-1990	2412	29

According to the table there were twice as many deceased centenarians in the first 50-year period as in the following 100-year period. Furthermore, in the 1751-1800 period 99 of the deceased were registered as 108 years old or older. The maximum age at death was as high as 127 (Lublin 1957). Assuming the same relative number of deceased centenarians as in the mid 1850s there would have been only some 100 deaths at the most in the 1751-99 period instead of 1098.

Census data show a similar pattern with a high proportion of persons aged 90 or more in 1750 declining up to 1800 (Figure 1). This fact with a high proportion of old people in the middle of the 18th century was commented upon in a statistical report some 100 years later. It was said that there could have been some mistakes when registering the age as many persons were born before the keeping of church registers had started. Comparing with other countries, however, Statistics Sweden found that Norway had very high rates. This was taken as an indication that the Swedish data might not be so wrong after all (BiSOS 1851-55 Del 3 p.25). This is of course not true. There are clear errors in the data. This was also pointed out by Sundb�rg (1903).

Using the extinct cohort method for the 1820-60 period the official population size can be compared with the calculated population aged 90 . The calculations clearly show that there was some kind of error in the data (Table 2). The difference between official and calculated population size indicates errors in the official population size and in age at death. Persons who had disappeared from the parish some time ago or perhaps had died were still included in the population register. The explanation for this error is the tightly linked registers. As no death or migration was reported the population register was not updated. This error was not taken care of until 1870 with the introduction of a register of 'residence unknown'. When the whereabouts of a person was unknown he was after some time entered in the register of 'residence unknown' and his name was taken off the population register.

In the individual cases the only way to check the age is to find register information of some kind, for example registration in the birth register. For the period 1860 to 1992 it is not necessary to verify age at death as that has already been done by Statistics Sweden. From 1907 a list of all centenarians was published yearly with information on name, date of birth, age at death and other facts for each individual. For census years a list of all centenarians was published too with detailed information about each person. It is only the death of Dorothea Andersdotter in 1860 that may be questioned. She was registered as 110 years old at death but in the copy of the church register age was just given in years and not as in all other cases specified in months and days too.

For the period before 1860 there are certainly a number of deaths where the stated age is not correct (Figure 2). The quality of data seems to deteriorate as we go back in time. There is a clear change in pattern of deaths by age before and after 1860. In the 1816-59 period there are many observations of very old persons, a pattern that changed abruptly in 1860. Furthermore there was a concentration of persons 100 years of age at death in the 1816-35 period clearly indicating some error in data. The only way to check age at death is to track down each individual, trying to find the person in the church registers. This is a very time-consuming method. You probably end up with certain authenticated centenarians and others who probably were as old as the clergy registered in the death register. For all other deaths nothing can be said at all. For those who died before 1800 and were said to be 100 years old or older it is highly unlikely that age can be verified at all.

The only thing that could be completed was a list of all deaths 1816-59 with information on year of death, age at death, sex and in which parish the death occurred - information which has not been readily available before. This list can form the starting point for a more thorough search of the parish registers.

The question when the first centenarian emerged in Sweden cannot be answered as yet. Only very uncertain speculations are possible at this time indicating that the first centenarian in Sweden perhaps emerged in the 1820s or 1830s.

The oldest old female in Sweden ever registered and verified was a woman called Hulda who died in 1994, 112 years and 105 days old. The second oldest woman died in 1986 and was 111 years 2 months and 28 days old. The oldest male in Sweden ever registered and verified was Nikolaus who died in 1993, 109 years and 236 days old.