THE idea that our ancestors died very young may be a myth based on a
statistical error, according to a team of British archaeologists. They say a
bias in the methods used for analysing human bones can underestimate the true
age of death by as much as 30 years.
Archaeologists estimate the age at which a person died by examining telltale
signs such as the state of skeletal development, and wear on bones and teeth.
Studies of bones found in Britain, dating from around the 11th century, have
suggested that virtually no one lived beyond 55. Remains from a Neolithic site
in the Orkneys pointed to a mean age at death of just 25.
But archaeologists have become increasingly concerned about the reliability
of these estimates, following studies at sites where human remains were
accompanied by documentary evidence of the true age of the skeletons. These have
hinted that the ages of young bones are overestimates, while those of elderly
bones are underestimates.
Advertisement
Now Mark Pollard and his colleagues at Bradford University, working with
Robert Aykroyd, a statistician at the University of Leeds, think they have found
the cause of the problem: the statistical method used to estimate age.
The basic idea of the method, known as linear regression, is simple.
Researchers take a large sample of bones of known ages and plot how a particular
characteristic鈥攕ay, wear on bone joints鈥攃hanges with age. Through
these points they run a 鈥渓ine of best fit鈥�, which they then use to read off the
age from measurements taken from other remains.
Pollard and his colleagues have found that the regression technique can give
faulty results when the correlation between the measured evidence and the age is
not particularly good. 鈥淭here鈥檚 a bias in the regression that depends on the
degree of correlation,鈥� says Pollard. 鈥淭he lower the correlation, the more it
tends to underestimate the ages of the oldest people in the population you are
蝉迟耻诲测颈苍驳.鈥�
Publishing their findings in the current issue of the journal American
Antiquity (vol 64, p 55), the team says the resulting error in ages could
be as much as 30 years鈥攕o bones estimated as coming from someone who died
at 40 may in fact come from a 70-year-old.
Other archaeologists say that the finding could resolve some long-standing
disputes. For example, in the 1950s anthropologists dated the bones of the
7th-century Mayan king Hanab Pakal as belonging to someone who died in his 40s.
Yet when the inscription on the king鈥檚 tombstone was translated three decades
later, it declared that he had died aged 80. 鈥淲hat the new research has done is
show that the inscription can be reconciled with the bones,鈥� says Norman Hammond
of Boston University, an expert on Mayan civilisation.
Pollard adds that the same bias could affect many other areas of research
which rely on regression techniques. 鈥淚t could also have a major impact on
attempts to reconstruct past climates.鈥�