Reading the newspaper on Saturday I came across an
article about the Japanese imperial family. This family is on the throne
for a 125 generations, but now, for the first time in history, the family is
about to die out. There are only seven males left, of which only one is younger
than 45. If Prince Hisahito (now 5) doesn’t have any sons, he would be the last
one on the throne. This due to a law which only allows male to sit on the Chrysanthemum
Throne and excludes females after marriage. This narrows the line of succession
to a minimum.
A Roman family; a relief from the Ara Pacis (Uffizi Gallery, Florence) |
In 1983 Hopkins wrote a book about the openness of the Roman senate.
He concludes that the aristocratic (political) elite were unable to
biologically reproduce themselves. Only 4% of the consuls in his research
(249-50) had consuls in six consecutive generations. 40% had a consular father
and 32% had a consular son. 27% only had one consul in the family. Five or six
consuls are quite rare. Although the Scipiones had seven generations starting
earlier, nevertheless this is quite an accomplishment. Hopkins bases his theory
on so called model life tables. Because there is no real– only partial and/or
defective – demographic information left, historians use model life tables for
a indication of life expectancy for example. These tables are based on modern
demographic data. There are different sets of model life tables available. All
based on different data. Hopkins for instance uses the United Nations tables,
whereas Parkin prefers the Coale-Demeny tables. For Parkin, the United Nations
tables are to rigid when it come to variations of the mortality figure. He
prefers the more flexible Coale-Demeny tables.
To see the difference between both sets of tables, it
is best to look at a arbitrary cohort of 100,000 people. At the end of each
year we look at the number of people who died (dx) and the people
who survived (lx). For a better comparison we take from both sets
the tables with a average life expectancy at birth of 25 (e0 = 25).
This means that every child born alive, had an average life expectancy of 25.
If the child reached the age of one, the average life expectancy would be higher.
This figure cannot be lower than 20, because it would be impossible for a human
population to reproduce itself.
So, we have an arbitrary cohort of 100,000 people and
an average life expectancy of 25. This table shows the numbers used per 5 year,
with the exception of the first year (to show infant mortality). After one year
of our 100,000 people, around 70,000 are still alive. At age 30 only 40,021 are
alive; which means that more than half died by then.
This
table also shows that the differences between the tables used by Hopkins and
those used by Parkin are minimal. Parkin does appear to have a slightly better
outcome. The greatest deviation can be found at the age of 60. In Hopkin’s
table 13,155 people are still alive (13.16% of the original cohort); for Parkin
16,712 people (16.71%). Is a deviation of 3.56% statistically negligible? Until
year 6o in Hopkins’ table dies an average of 6,203 persons per period; for
Parkin this is 5,949. A difference of 254
persons per period. That’s 0.000254% of the original cohort. I’m not a statistician,
but for me that’s not much. So I think the difference between both tables are
negligible.
The big difference is in choosing the average life
expectancy at birth. The outcome for e0=25 or e0=30 can
make a huge difference. I made a separate set for Hopkins’ e0=25 as
well as e0=30. This table shows that another choice in average life
expectancy can lead to different outcome.
For
instance, if we look at the number of persons alive at 60, for e0=25
13,155 persons are alive, for e0=30 this is 19,353. A difference of 6,198
persons. This difference is higher than the aviation between the model used by
Hopkins’ and that of Parkin! So the choice of average life expectancy is more
important than choosing the model.
Do
these models be of any value when we look at the Scipiones? A child born in
this family had an average chance of 70% to reach his first birthday. The most
important political functions can only be reached at a later age: consul at 42
year and praetor at 39 year. The average chance of reaching this age is 32% (40
year). For the Scipiones we know only the men who reached this age and
therefore played a role on the political playfield. Of only two Scipiones we
know their age at death. The first is L. Scipio, the son of Scipio Asiagenus. His
epitaph mentions his age as 33, “annos gnatus XXXIII mortuos”. The second Scipio is Scipio Comatus. His epitaph
says he died at 16, “annoru gnatus XVI”.
Of none of the other male Scipiones the age at death is known, even not from
those whose epitaphs we know. In those cases only an estimation can be made. For
these cases a model is not a solution.
So
a model life table can only be used for the more mundane questions, for giving
an global image of a population in a certain time and certain place.
Based
on:
A.J.
Coale and P. Demeny, Regional model life
tables and stable populations (2e print; New York, 1983)
K. Hopkins,
Death and renewal Vol. 2. (Cambridge,
1983)
T.G.
Parkin, Demography and Roman society (1992)
United
Nations, Methods for population
projections by sex and age Vol. 25. (New York, 1956)
R. Woods,
“Ancient and early modern mortality: Experience and understanding”, in: Economic
History Review 60.2 (2007), p. 373-99
Further
reading suggestions:
B. Frier,
“Roman life expectancy: Ulpian's evidence”,
in: Harvard Studies in Classical
Philology 86 (1982), p. 213-51
R.P.
Saller, Patriarchy, property and death in
the Roman family (reprint; Cambridge, 1996)
W. Scheidel,
“Roman age structure: Evidence and models”,
in: JRS 91 (2001), p. 1-26
No comments:
Post a Comment