Statistical Investigations 343 or a total loss of 533 surnames. Here the population increases since

tl+2t,+3t3+4t4+5t,

is greater than unity. As before the extinction rate is quick to begin with

but soon slackens down, as the number of persons holding each surname

increases, while the number of surnames diminishes. On the above hypo

thesis nearly a quarter of newly-created peerages would become extinct in

the first generation and half of them by the sixth generation. With any such

hypothesis there is no need to appeal to sterility as rendering rapidly extinct

a large proportion of the peerages created for ability. It will be clear that

if we take not the number of sons, but the number of children, in computing

the t's, the problem becomes that. of the extinction of definite stirps; it is

highly probable that families die out in approximately the same manner as

they die out in the male line. If mankind has not sprung from a single

pair, it seems possible that even the most numerous nation may tend with the

ages to be the product of a very few stirps, if not of a single pair. The fable of

Adam and Eve may be somewhat truer for an old world than for a young one!

Beside the data noted in the paper on the stature of boys from urban and rural schools', several schools provided material of a more extended kind, notably Marlborough School, which had established something like an anthropometric laboratory 2. The school medical officer and the natural science master took the measurements : namely weight, stature, horizontal circumference of the head, chest girth, girth of the flexed arm over the biceps muscle, girth of the leg over the calf, both the last two being the maximum measurements. The ages of the boys ranged from 10 to 19 and there were 550 of them. The authors 'of the paper give three correlation tables for age with stature, weight and head circumference, but make no reductions, citing merely in the case of the extreme boys in each measurement the other measurements of those boys. One remark deserves citing. The authors state that they "are unable to trace any distinct connection between intellectual vigour and head measure

ment ; for although many of those who possess the higher girths of head are intelligent boys of considerable ability, it must be confessed that many boys whose heads measure less than 22 inches, are in ability, perseverance, and general culture, quite equal to those who possess

the higher measurements." (p. 129.)

This remark bears on a point already referred to in this Life (p. 94).

Galton's short accompanying paper confines itself to one character, stature, and he tells us that he proposes to illustrate the statistical methods which will be adopted, when sufficient material of a homogeneous nature is available. He takes the boys for each year of age and finds their means, which give for the central ages 122, 132, etc. the law of growth. He thus obtains what we should now term the regression line. But here he strikes a new point: he finds that the arithmetical means of the arrays are not identical with

' See our p. 125.

2 "On a Series of Measurements for Statistical Purposes, recently made at Marlborough College." By Walter Fergus, M.D., and G. F. Rodwell, F.R.A.S. Journal of the Anthropological Institute, Vol. iv, pp. 126-30. "Notes on the Marlborough School- Statistics" by Francis Galton himself follow this paper, pp. 130-5.