Page Actions

Generation length

From ISOGG Wiki

Generation length is of interest in two distinct contexts:

  • In “deep ancestry”, where early studies in anthropology and genetics over several millennia typically use average generation lengths of 25, 20 or even 15 years.[1]
  • In conventional genealogy, where more recent studies (post 2001) of pedigrees and historical data typically show average generation lengths of 30 or 35 years (see later section of this webpage).

Genetic genealogists seek guidance on which generation length is the most appropriate to use with TMRCA (Time to the Most Recent Common Ancestor) tables when they want to multiply the number of elapsed generations by some average generation length to derive the elapsed time. NB There does not seem to be a consensus for this application, or even on whether the appropriate average generation length should be based on recent historical data, or on assumptions of shorter genealogical lengths associated with pre-historic times. It may even be that the mutation rates themselves have been derived using some assumed generation length, in circumstances where it would be wrong in principle to use any other generation length to convert the product of generations and generation rates back into elapsed time.

There is currently little consensus on definitions, calculation methods and applications.[2] This article seeks to throw some light on these issues.

Definition of generation length

Generation length, also known as generation interval, generation time, generational length and intergenerational interval, is generally accepted as being the number of years between the births of a parent and of a child, that is the age of a parent when a child is born.

In practice generation length can vary widely. For example:

  • A 13-year-old father and 12-year-old mother were recently reported in the UK.[3]
  • The artist Pablo Picasso fathered children at ages 40, 54, 66 and 68.[4]
  • An Indian man fathered a child at 96; the mother was 52.[5]

Obviously, parenthood can extend from puberty to death or, for women, if earlier, to menopause.

Average generation length

Average generation length is the average (i.e. mean) age of parenthood of a number of children, either in one generation or across several generations. For example:

  • If a man had three children when aged 27, 30 and 36, then their average generation length is 31 years ((27 + 30 + 36)/3 = 31). NB The average generation length may differ from age at birth of the middle child (in this example 30), and average age at birth of the first and last children (in this example 31.5).
  • If the same man was born when his father was aged 28, who was born when his father was 40, who was born when his father was 32, who was born when his father was 28, then the man’s patrilineal (i.e. father-son) ancestry has an average generation length of 32 years ((28 + 40 + 32 + 28)/4 = 32).
  • If the same man was born when his mother was 27, who was born when her mother was 31, who was born when her mother was 29, then his matrilineal (i.e. mother-daughter) ancestry has an average generation length of 29 years ((27 + 31 + 29)/3 = 29).
  • If ancestral research extends the maternal ancestry of the same man back a further three generations with motherhood ages of 24, 26 and 24, then the average generation length of his maternal ancestry changes to 27 years ((27 + 31 + 29 + 24 + 26 + 24)/6 = 27).

These examples show that one individual can be associated with several average generation lengths, some of which may not match the averages quoted at the beginning of this webpage.

Biological, social and evidentiary factors influencing average generation length

Generation length may be influenced by many factors, such as:

  • Changes in the average age of puberty and menopause.
  • Changes in death rates of reproducing adults: over the centuries man has been living longer, while women today are less prone to death associated with childbirth.
  • Men and women may postpone parenthood in anticipation of a better financial situation.
  • Poor families, without prospects, may have children earlier than more affluent ones.
  • Local and national economic and societal mores may favor child-bearing at particular ages.
  • In time of war men may be killed before having as many children as they would otherwise, and may hasten fatherhood in fear of death, or contribute to postponing births – e.g. the baby boom of the late 1940s. [6]
  • The Black Death of 1348/50 and the Spanish ’flu epidemic of 1918/19 should have temporarily shortened average generation lengths.
  • In the nineteenth century large families were common, as they still are with Catholic families, and some younger children were an uncle or aunt before they were born.
  • Monarchs, nobles and landowners/lairds descend through elder sons, so average generation lengths of such pedigrees will be shorter than for pedigrees of whole families.
  • Illegitimate births are often from younger parents.
  • Pedigrees of landowners are more easily traced, and of illegitimacies less easily traced, leading to evidentiary bias in associated average generation rates.

These factors suggest that it is rash to assume average generation lengths should be consistent over several centuries, or from country to country, or indeed within in any one particular pedigree.

Calculation of average generation length

The above considerations suggest that calculating or estimating average generation lengths is a pointless exercise.[7] However awareness of typical average generation lengths is not only relevant to TMRCA applications but is also a useful aid for checking if one or more generations are missing from a pedigree. For example some pedigrees claiming descent from the time of William the Conqueror only show three or four generations between the 11th century and the 16th century, implying an average generation length of a century or more over this period, even though the average generation length between the 11th century and the present day may be a more credible 40 years or so.

There is a right way and a wrong way to calculate the average generation length from a series of birth dates. Some websites infer one should subtract the date of birth of an ancestor from one's own date of birth and divide this elapsed period by the number of generations.[8][9] Thus, for example,

  • Generation 1: great great great grandfather, born 1775;
  • Generation 2: great great grandfather, born 1810;
  • Generation 3: great grandfather, born 1836;
  • Generation 4: grandfather, born 1869;
  • Generation 5: father, born 1909;
  • Generation 6: self, born 1937.

Thus average generation length = (1937-1775)/no. of generations = 162/6 = 27 years. However this calculation method is incorrect: in fact the individual generation lengths are 35, 26, 33, 40 and 28 years, which indeed sum to 162 years, but this total should be divided by 5 rather than by 6 (the “five fingers, four gaps” conundrum), to give a correct average generation length (for this data) of 32 years (actually 32.4 years, but decimal places for the average generation length of a single pedigree imply misleading accuracy).

Where a date of birth is not known it may be estimated. Ages at marriage for successive generations is another proxy for generation length, but more indicative of the average for eldest children than that for several children.

Studies of average generation length

Many recent studies of average generation lengths have challenged earlier assumptions:

  • Tremblay and Vezina studied 100 catholic French Canadian families of 1850-1990s, finding average generation lengths of 34.5 years for males and 28.9 years for females.[10] They noted that some of the earlier genetic studies were based on mtDNA studies only, which may explain some of the bias now apparent in their average generation lengths.
  • Helgason et al studied Icelandic genealogies of 1742-2002 and found average generation lengths of 31.9 years for males and 28.7 years for females.[11]
  • Matusmura and Foster studied 1805-1974 genealogies for 225 families in North Greenland and found average generation lengths of 32 years for males and 27 years for females.[12]
  • Jack Fenner undertook detailed analysis of many pre-2005 genetic and genealogical studies and found average generation intervals amongst hunter-gather societies of 31.5 years for males and 25.6 years for females (although with a large variance), amongst less developed nations of 31.8 and 28.3 years, and amongst developed nations of 30.8 and 27.3 years respectively. He recommended the use of 31 years for males and 25 years for females.[13]
  • Donn Devine, after reviewing published research on Quebec, Icelandic, and Botswana Dobe !Kung hunter-gathererer populations, recommended using 33 years for male-line generations and 29 for female lines. Comparing his own known ancestry, he found close agreement, with average generation lengths of 34 years for males and 29 years for females.[14]
  • King and Jobling adopted 35 years for English males.[15]
  • The pedigree of Confucious over 80 generations, recently revised, has an average (male) generation length of 32 years.[16]
  • The earliest pedigree in the Cruwys surname DNA project shows average intervals of 30-35 years for males.[17]
  • In the Irwin DNA Project nine Irwin paternal pedigrees dating from between 1323 and 1660, largely Scottish, with a wide range of socio-economic backgrounds, have intervals of between 31 and 38 years.[17]
  • The MacDonald surname DNA project recommends 31 years for males.[18]
  • The Williams surname DNA project finds 28-33 years for males.[17]
  • Tetushkin suggests using 31-32 years as the average generation length for males, and a little less for females.[19]
  • Katherine Borges gives examples of average generation lengths of 18.2 years, 43 years and 58.5 years.[7]

Conclusions

Recent studies, particularly by Fenner, suggests that earlier recommendations of average generation lengths of 25 years or less are inappropriate, even for pre-historic societies. Devine's "rule-of-thumb" that males typically span 3 generations per century, which is the same as the "genealogical law of three generations" quoted by Tetushkin (i.e. an average generation length of 33 years) and females 3.5 generations per century (i.e. an average generation length of 29 years) appears to be a useful and reasonable tool for both genetic and conventional genealogy. But, as Borges rightly emphasises, individual pedigrees may have generation rates very different to these averages.

Recommended reading

References

  1. Walsh B. Estimating the time to the MRCA for the Y chromosome or mtDNA for a pair of individuals. Genetics 2001; 158: 897-912.
  2. How many years are in a generation? Yahoo Answers. Accessed 11 January 2015.
  3. Gysin C. The boy who became a father at 12 and now calls a jail cell home. Daily Mail, 20 February 2009.
  4. Saper R. The women of Pablo Picasso 1881-1973 Saper Galleries website, accessed 11 January 2015.
  5. Kim J. World's oldest dad, 96, fathers another child. Time magazine, 18 October 2012
  6. Laden G. How long is a human generation?. Greg Laden's blog, 1 March 2011.
  7. 7.0 7.1 Borges K. The hampster wheel of generation length. ISOGG Newsletter Jan./Feb. 2012.
  8. Lancaster FM. Generation intervals and their significance Genetic and Quantitative Aspects of Genealogy. Published online October 2005 and updated in November 2007)
  9. Brunet M. How to calculate a generation.eHow website, accessed 11 January 2015.
  10. Tremblay M, Vezina H. New estimates of intergenerational time intervals for the calculation of age and origins of mutations. American Journal of Human Genetics 2000; 66: 651-658.
  11. A Helgason, B Hrafnkelsson, JR Gulcher, R Ward, K Stefansson. A populationwide coalescent analysis of Icelandic matrilineal and patrilineal genealogies: evidence for a faster evolutionary rate of mtDNA lineages than Y chromosomes. American Journal of Human Genetics 2003; 72: 1370-1388.
  12. Matsumura S, Forster P. Generation time and effective population size in Polar Eskimos. Proceedings of the Royal Society of Biological Sciences 2008; 275: 1501-1508.
  13. Fenner JN. Cross-cultural estimation of the human generation interval for use in genetics-based population divergence studies. American Journal of Physical Anthropology 2005; 128: 415-423 (subscription required).
  14. Devine D. How long is a generation? Science provides an answer. Ancestry Magazine Sept/Oct 2005: 53-55.
  15. King TE, Jobling MA. Founders, drift, and infidelity: the relationship between Y chromosome diversity and patrilineal surnames. Molecular Biology and Evolution 2009; 26(5):1093-1102.
  16. The longest family tree. GenealogyInTime Magazine, accessed 11 January 2015.
  17. 17.0 17.1 17.2 Irvine JM. Towards improvements in y-DNA Surname Project Administration. Journal Of Genetic Genealogy Fall 2010; 7.
  18. STRs and haplotypes . Clan Donald website
  19. Tetushkin EY. Genetic aspects of genealogy. Russian Journal of Genetics 2011; 47(11): 1288-1306.

See also