But Galton found 48 sons per 100 fathers! Now I have already referred (p. 96) to my doubts as to Galton's estimate of the number of relatives to be attributed in each grade to an eminent man. He was perhaps biased by the wickedness of Judges and the misogyny of Statesmen ! Anyhow I feel certain that the columns C of his tables and consequently the columns D are incorrect'. Had he attributed 200 or 250 sons to 100 eminent fathers or families, say, of 4 to 5, he would have found 19 to 24 eminent sons to 100 eminent fathers-still far too many-but approaching nearer our 13 with a
Eminence 1 in 1000
Father
Eminence 1 in 500
Father
Eminence 1 in 100
Father
Non-
eminent
Eminent
Totals
Non-
eminent
Eminent
Totals
Non-
eminent
Eminent
Totals
Son Non-eminent
{Eminent
998,054
946
999,000
996,139
1,861
998,000
581,298
8,702
990,000
946
54
1,000
1,861
139
2,000
8,702
1,298
10,000
Totals ... ...
999,000
1000
,000,000
998,000
2,000
1,000,000
990,000
10,000
1,000,000
No. of eminent
sons per 100 emi-
5.5
7
182
nent fathers
much lower degree, however, of eminence. An explanation of the remaining discrepancy may, however, be found in the hint' thrown out by Galton in this chapter, that "a large number of eminent men marry eminent women'." He had already emphasised this point of view when discussing Men of Science and Divines. But such a mating of `like with like' raises the correlation between offspring and parents slightly under 50 °/ab: Forming a table under these conditions we find for 1 in 100 degree of eminence
' Loc. cit. p. 317 for general table, and compare tables at end of each section. This is much of the order one finds for number of insane sons of insane fathers.
3 Loc. cit. p. 325.
' "The large number of eminent descendants from illustrious men must not be looked upon as expressing the results of their marriage with mediocre women, for the average ability of the wives of such men is above mediocrity. This is my strong conviction, after reading very many biographies, although it clashes with a commonly expressed opinion that clever men marry silly women. It is not easy to prove my point without a considerable mass of quotations to show the estimation in which the wives of a large body of illustrious men were held by their intimate friends, but the following two arguments are not without weight. First, the lady whom a man marries is very commonly one whom he has often met in the society of his own friends, and therefore not likely to be a silly woman. She is also usually related to some of them, and therefore has a probability of being hereditarily gifted." (p. 324.)
s The multiple correlation coefficient between parentage and offspring is now •7071.