K. Pkahson 221 



Heredity, aml I hrliuvc it. is a gciun-alisatioii, wliich iVlr (laltoii iiii;^'lit., pcrhaps, 

 he williii"- to acc('ii( in place ot' iiis iiiui-h inori' ilefiiiite se^ic^s of \, |, J, etc. 



(5) AssuiiiiiiL,^ \ve had thc data ti>r ipiaiititatively measurable charactcrs oiie 

 would prefer to dcal witli such, hiit, failiug this I have asked niyself liow far thu 

 Kuccessivc corrclation coetficicnts Cor anceslry in <jualitativo characters (whioh we 

 luay icasoiiahly suppose to havu a ([iiantitative scalc bchiud thoin) fall into a 

 geoiiictrical progrcssion. In eyo-colour for man I have previciusly givcn* the fi)nr 

 parental and the eight grandparental correlations. Mr F. l.ntz has conipletedf 

 the series, as far as it is possible to do on Mr (ialton's tnaterial, hy investigating 

 the corrclation of oftspring with the groat-grandpareuts, arrangod in fonr groujis. 

 Mr L. Bramley-Moore and I had given the four parental and two of the grand- 

 parental coat-colour correlations for horses in 189f)|. Mr N. Blanchard published 

 two further grandparental correlations recently§, and he has novv conipleted the 

 whole series of eight, and they will be found in a note attached to this paper. 

 Dr Alice Lee worked out the great-grandparental and the great-great-grand- 

 parental corrclation froni the data provided by Mr Blanohanl's MS. pedigree 

 books; thesc are given in a second note below. 



We have reached the foUowing results : 



Mean Correlatiou Eye-colour in Man Coat-colour in Horse 



Parental '4947 -5216 



Grandparental -3166 -2976 



Great-grandparental -1879 -1922 



Grcat-grcat-grandparental — "1469 



When we note that the probable error of the.se coefficients is about the order 

 ■0.3, one is at once Struck with tlieir substantial agreenient. We are forccd to the 

 conclusions : 



(i) That eye-colour in man and coat-colour in horses are inheritiMl iu 

 sensibly the same mannen 



(ii) That the original series proposed by Mr Galton \, ^ ... for these 

 correlations]! will not fit them at all. 



It might be supposed that pigmcntation foUowed sorae other rulc thaii 

 measurable characters. But in reccnt work on measurable characters for 12 

 series, each numbering upwards of 1000 cases, I have found for man the mean 

 parental value '46. This is the largest inheritance series for man yet worked 

 out, being based on raeasurements of upwards of 1100 families, so I am pretty 



* Phil. Tram. A, Vol. 19.5, p. 10(>. 



t See Note ou the Inflnence of Change in Sex on Inheritance. lUomi-trilii, Vol. ii. p. 287. 



* Phil. Trans. A, Vol. 1H5, p. 93. 

 § Biometrika, Vol. i. p. 3()1. 



il Natural Inheritance, p. 133. 



