238 Miscdlanea 



thcrc aro threo clianges. I further cliusscd ProfcssDr Pearson's constants for paretital aiid 

 grandiwirental data* in thc sainc nianner, takiiig tlie average of the coeflicients obtaiucd hy hiiu 

 für each class. 



The resiüts are aa foUows: 



PareiUal Inheritance. 



No change of sex. 



•4594 ... ... ... ... One change of sex. 



Grand parental Inheritance. 



Mean /-„.j, (•,.;= SfiÖT ... ... No change of sc.\. 



Mcan (-„.ä, r,_,, ;■,,„, r^ 7 = ■3004 ... ... ... ... ... ... One change of sex. 



Mean r„.u, r,,5=-2961 Two changes of se.\. 



Great-grand parental Inheritance. 



tGrouixs 8 and 0, 1.') and 1 give )' = -3471 ... ... Xo change of sex. 



Groni» 9 and 0, 11 and 0, lö and 0, 8 and 1, 12 and 1, 14 and 1 give 



r='2217 ... ... One change of sex. 



Croups 10 and 0, 12 and 0, 14 and 0, 9 and 1,11 and 1, 13 and 1 give 



»•=•1452 ... ... ... ... ... Two changes of sex. 



Groiips 13 and 1, 10 and give r=0375 Three change-s of sex. 



The mean result of the foui- gre;it-grandparental cocfficients here found is •1879, or if all thc 

 niaterial \x chibbed together and onc conelation table fornicd for it, it is ■1824, the two 

 resiüts bcing practically identical, considering the probable error of their diÖerence. At the 

 end of this note I give the actual correlation tables, remarking that the classes are thase of 

 Mr Galton's eye-colour Classification, i.e. 



1 = Light bhie. 5 = Light browu. 



2 = Blue, dark bhie. 6 = Brown. 



3 = Grey, blue-green. 7 = Dark brown. 



4 = Dark grcy, hazel. 8 = \'cry dark brown, black. 



For purj)o.scs of calculation thc fourfold table wa.s foriued by cla.s,siiig togother 1-3, and 4-8. 



Now, although tho probable error of tlicse rcsults iinis froni '033 to •O.')."), their invariable 

 dcci-ease with an increasiug niimber of change-s of sex is highly significant. It can hartlly 

 be Said tliat there is a constant factor of reduttion for e;ich change of sex ; but, if it cxisted, a 

 tuore claborate .System of nie;isureincnts WDuld probably Ix! i-equired to satisfactorily deteruiine 

 it. Still we can conclude with absolute .safety for eye-colour in man and probably therefore for 

 many other charactei^s in other forms of life that : 



Every change of sex in the line of ancestrij sensibly wcakem the intemity of inheritance. 



It is desirable accordingly in future heredity experinients to record carcfully such changes 

 for they ovn obviously be of inuch significance. We niay even see in them some justification 

 for refusing to admit inheritance of title through the female line, although thc objection would 

 apply equally to the transfer froni thc fenialc through the male line. 



* Phil. Trans. A, Vol. 195, p. lOfl. 



t The rcsults hcre are obtained, not by tnkiog the mean.s of correlation coefEcients, but by cinbbiug 

 groupg together. 



