26 HEREDITY 



conditions more or less abnormal, such as deficient 

 food, exceptional exercise of a particular member, dis- 

 ease, mutilation, &c., acquired characters would be 

 produced. 



If we could take a pair of identical t\\'ins, and separate 

 them at birth, giving the one fresh air, abundant food 

 and exercise and education, and rearing the other amid 

 the unhealthy and degrading conditions of our slums, 

 the differences, wliich would certainly be enormous, 

 would be altogether modification differences. If, on the 

 other hand, we could take two children of different 

 parentage, and rear them under exactly the same 

 conditions, the differences between the two M'ould be 

 altogether inborn. 



In practice it is not always possible to distinguish 

 inborn from acquired characters, even when we know 

 fully the conditions under which the organism has Hved. 

 But theoretically the distinction between what is " bom " 

 and what " made," is absolute. 



We may now turn to the inborn variations and en- 

 deavour to see how they arise. How does an organism 

 come to differ from its parents, or from the mean of 

 its parents, if it has two, -with regard to its inborn char- 

 acters ? 



It seems certain that the great majority of inborn 

 differences between parent and offspring are due simply 

 to new combinations of previously existing characters. 

 It is easy to see that with sexual reproduction, where 

 two parents are involved in the production of the 

 offspring, there is a continual mixing of different germ 

 plasms, and thus almost infinite chances of new com- 

 binations. 



By way of illustration, we may suppose four sacks 

 of marbles of different colours, say white, black, red, 

 and A-ellow respectively, to represent the respective 

 germ plasms of four individuals. By mixing the black 

 with the white ar:d the yellow wdth the red, we get 

 something corresponding to two new germ plasms. If 

 now we take random samples from each lot, say half-a- 

 dozen at a time, and mix the sample from the one lot 



