Miscellanea 
109 
PAPERS REFERRED TO. 
1. Castle, W. E., and Faiiabek, AV. C. Notes on Negro Albinism. Science, 1903, xvii. 
2. Batesox, W. The Present State of Knowledge of Colour- heredity in Mice and Rats. 
Proc. Zoolng. Soc. 1903. Vol. 11. 
3. Carkud, a. Ueber chcniisehe Individualitiit und cheniische Misshildungen. PJtigcrs 
Arch. f. d. gcsammte Plii/siologic, xcvil (1903). 
4. Auc'OLEo, O. Suir Albinismo in Sicilia. Arcliivio per I' Aiitnijitilagia, i (1871). 
5. Darbishire, A. D. On the Result of Cro.ssing Japanese Waltzing with AlI)ino Mice. 
Blo'inetrika, Vol. Ill, pp. 1—51. 
C. CuENOT. L'Heredite de la Pigmentation chcz les .Souris. Arch. Zoul. Exp. (Notes et 
Revue), 1902. Ser. 3, T. x. No. 2. 
7. Livi, R. Antropometria inilitare. 4to. Roma, 1896. 
IV. A Mendelian's View of the Law of Ancestral Inheritance. 
It is a very thankless task to try and correct every mis-statement that untrained minds make 
when they attempt to deal with a statistical problem. But one can only hope that the persistent 
exposure of the blunders made by non-statistically trained biologists, when they treat problems 
of heredity, may ultimately j^roduce some effect. To the statistician nothing is more obvious and 
intelligible than the independence of the correlations, variations and means of characters. 
A knowledge of any one of the three involves no knowledge of the other two. Again, equally 
distinct in his mind are the multi})le regression coefficients and the correlations. No practised 
bionietrician could for a moment confuse with the ancestral correlations the multiple regression 
coefficients, which appear when we calculate the most probable deviation of an individual from 
his own type with the known deviations of his ancestors from their types ; still less would he 
confuse a correlation coefficient with the number of offspring that take after a parent or grand- 
parent 1 It is difficult to understand at all the attitude with which a biologist like Professor 
Castle* approaches a problem in inheritance, but it is summed up in the words that he has not 
the least idea that there is any distinction between a correlation, a multiple regression coefficient, 
and the number of individuals who may take after an individual ancestor ! 
Mr Francis Galton published in his Natural Inheritance (p. 136), " with hesitation " the 
following .statement "consequently the influence of the individual parent would be |, and 
of the individual grandparent and so on." 
In his work on Bassett Hounds, Mr Galton subsequently extended this principle by supi^osing 
that of the " heritage " as represented by all the offspring together, ^ of the total would on the 
average have the character of each parent, that of each grandparent, that of each great 
grandparent, and so on. This statement is obviously difterent from that in Natural Inheritance, 
where Mr Galton is treating of a blending character. It deals with alternative inheritance. 
From these numbers Professor Castle deduces " Galton's Series." He calls the parental influence 
•50, the grandparental influence '25, the great grandparental -125 and so on. I ^jresume that 
he understands by this the proportions in the total offspring who will be like each individual 
ancestor. Against the.se proportions Professor Castle puts a series which he calls " Pearson's 
* "The Laws of Heredity of Galton and Mendel, and some Laws governing Eace Improvement 
by Selection," by W. E. Castle. Proceedings of the American Academy of Arts and Sciences, Vol. xxxix, 
pp. 223—242. 
