488 THE GRAMMAR OF SCIENCE 



(iv.) Cases of essentially normal regression, misinter- 

 preted as reversion. For example, a hyper-brachycephalic 

 man will have children regressing towards the general 

 population mean, and thus very probably towards their 

 grandparents, but this is not a true case of reversion at 

 all, it is the every-day phenomenon of regression. 



(v.) Cases of changed environment. The direct action 

 of the environment (or even natural selection due to 

 change of environment) may curiously modify the type, 

 and the result may often be loosely described as atavistic. 



A Skye terrier pup, one of a litter whose members 

 were quite normal Skye terriers, was taken when six 

 months old to the Straits Settlements and knocked about 

 two years on a coasting steamer. He returned with short 

 hair and long legs, and if his history had not been known 

 would doubtless have been described as a case of atavism, 

 and compared with the old English terrier ! 



(vi.) Cases of hybrids, when two distinct races are 

 crossed, and from physiological and mechanical reasons 

 the gametes produce a zygote which does not give an 

 individual blending the ancestry. Here any singularity 

 almost may be expected, and it will be perfectly easy 

 from some real or apparent resemblance in the cross to 

 other species to assert atavistic tendencies. 



In conclusion, it will appear difficult, I think, to 

 separate cases of true reversion or atavism from cases of 

 the above kind, and whenever we have a character which 

 is normally blended {e.g. stature in man, coat-colour in 

 horses) we shall find it impossible to assert reversion. It 

 is in the field of abnormal variation that we must look for 

 instances of atavism, and in that of exclusive inheritance 

 that we must seek for reversion.^ 



on " Regression, Heredity, and Panmixia," Philosophical Transactions, vol. 

 clxxxvii. , A., 1896. 



' Considering current biological and medical use, it may be difficult now 

 to attach definite meanings to the terms regression, reversion, and atavism, 

 but I have endeavoured in this book to keep them clearly apart. By regression 

 I mean the phenomenon described on p. 456, which occurs universally 

 in the inheritance of blended characters. Darwin includes it under reversion, 

 thus when in the Origin of Species, 4th ed. p. 1 1 5, he speaks of the tendency 

 of reversion to check natural selection being " greatly exaggerated by some 



