70 



Life and Letters of Francis Galton 



(i) Mild, (ii) Docile, (iii) Fretful, (iv) Violent, and (v) Masterful. I liave 

 already discussed this classification in Vol. n, p. 271. I would only add that 

 if we follow Galton's classification of Good and Bad Temper, we find a slight 

 negative correlation between Husband and Wife. If it be considered signi- 

 ficant, the mild temper of one mate may be due to the experience that 

 control is needful or at least advisable in the environment of a violent 

 consort. On the other hand the fourfold table for siblings, i.e. offspring of 

 the same parents, is: 



Temper in Siblings. 



1st Sibling 



J 



3 



02 

 T3 



a 



<M 



The numbers in curled brackets give the frequencies which would occur in 

 each category on the basis of independent chance. It will be seen that 

 observation shows a heaping up in the like categories at the expense of the 

 unlike categories. The correlation coefficient is "3167 for this fourfold table; 

 there is thus a considerable degree of resemblance between the temper of 

 siblings, but I believe this measure would be considerably increased if sullen 

 and fiery tempers were not included in one group. 



Chapter X deals with the subject of Disease. This is a most interesting 

 and suggestive chapter, but the data were too sparse to provide definite con- 

 clusions of any kind. 



Galton states (p. 165) first (by again appealing to an analogy!) his 

 Preliminary Problem. We know, he tells us, the ages at death and the 

 causes of death of the population as a whole. We know the proportions at 

 each age of those who die of diseases A, B, C, etc. He would assume — 

 which I think is somewhat doubtful — that the proportions of persons dying 

 of these diseases at various ages in two successive generations are the same. 

 If now we limit ourselves to persons dying at a certain age of disease A, 

 how, if at all, does this affect the distribution of deaths from the diseases 

 A, B, C, etc., in the offspring generation ? 



The problem is an exceedingly difficult, if an extraordinarily important 

 one, for it requires an immense mass of data. In the first place the pro- 

 portional death distribution is a function of social class, and of occupation ; 

 it is as we have seen a function of age ; it influences fertility ; and in more 

 than one way is affected by sex*. Anyone who has seriously faced the 

 problem, and seen the number of groups into which the various affecting 

 factors compel him to sort the material, will recognise how hopelessly 



* The male in many cases, as by foreign travel or by military or naval service, runs greater 

 risks than the female. 



