Correlation and Application of Statistics to Problems of Heredity 59

various degrees (pp. 1-2). Such are the three fundamentally novel problems which Galton set himself in Natural Inheritance; we shall endeavour to show the extent to which he has solved them, or at least has suggested methods of solving them, in the following discussion of that work.

Chapters II and III are general in character, expressing Galton's own views on heredity, and erring, if at all, in rather too much appeal to analogy. In the first of these chapters Galton states his opinion as to " natural " and " acquired" characters, indicating that he considers the inheritance of the latter extremely doubtful; he emphasises the importance of closely criticising the evidence offered in each case to prove the transmission of acquired faculties, citing especially the possibilities of intra-uterine influence*. He refers to the difficulty of combining male and female measures, and states that:

"Fortunately we are able to evade it altogether by using an artifice at the outset, else, looking back as I now can, from the stage which the reader will reach when he finishes this book, I hardly know how we should have succeeded in making a fair start. The artifice is never to deal with female measures as they are observed, but always to employ their male equivalents in place of them. I transmute all the observations of females before taking them in band, and thenceforward am able to deal with them on equal terms with the observed male values." (p. 6.)

Galton for stature multiplied every female stature by 1.08 to reach its male equivalent, or added about one inch to every foot of female stature. He does not tell us how he demonstrated that equivalence, whether from the ratio of the mean values in men and women, or more adequately by finding it held (approximately) for all grades t. The true method is to reduce each deviation from the mean by dividing by its standard deviation, or other measure of variability, and it was an inspiration on Galton's part that led him to recognise that at any rate for the case of stature, the ratio of variabilities in male and female was close to the ratio of their mean values. See our p. 15 above.

On p. 7 Galton deals with what he terms Particulate Inheritan He recognises that an individual may possess characters, which a e kn wn to have existed in an ancestor, but were not in the immediate parents. From this idea of latent characteristics Galton reaches the conception of inheritance in the individual as a "mosaic" of ancestral factors, and illustrates his views by two analogies, that of a builder's yard, with fragments of old buildings ready to be used again (p. 8), and the vegetations on two islands which spread to adjacent islets (pp. 10-12). I think he would have done better to have retained his earlier conception of the "stirp" (see our Vol. ii,

* The complexity of this latter source must be borne in mind, if we can accept Galton's statements on pp. 15-16, that not a drop of blood passes from mother to child, and yet that a mother's system maybe "drenched with alcohol and the unborn infant alcoholised" during all its intrauterine existence.

t Probably in this latter way; see his p. 42, where he says we are to transmute female to male measures by comparing their respective "schemes, and devising a formula which will change one to the other. A "scheme," supposed normal, depends on two constants, the mean and the variability. Galton does not point this out, or state the inference which follows from his use of the factor 1.08.   -

8-2