6 INHERITANCE OP COAT-PIGMENTS AND COAT-PATTERNS 



nance, but because the majority of the Irish rats used happen to have been 

 heterozygous, bearing the hooded pattern as a recessive character; but, as 

 we have seen, this is not a necessary relationship. Hooded rats, however, 

 bear no other color-pattern, so that any pair of hooded rats of whatever 

 pedigree will produce only hooded pigmented offspring. 



To sum up in a brief statement the principal facts presented concerning 

 the coat-patterns of rats, the following series of conditions may be recog- 

 nized, each with pigmentation less extensive than the preceding: (i) Self, 

 whole body pigmented; (2) Irish, whole body pigmented except more or 

 less of the ventral surface ; (3) Hooded, only the head, shoulders, and usually 

 a median dorsal stripe pigmented; (4) Albino, no pigmentation. 



Each condition behaves as a dominant toward those which follow it in 

 the series, and as a recessive toward those which precede it. Bateson (:o3, 

 p. 81) refers to Crampe's failure to obtain the self or the albino conditions 

 by selection from the parti-colored, and adds : "The types are in fact definite, 

 and can not be built up by cumulative selection." The statement applies 

 strictly only to the two extremes of the series, viz, self and albino, but by 

 implication to the others also. Our experiments, however, indicate that it 

 is possible to modify by selection and cross-breeding both the Irish and the 

 hooded conditions, leading to the production of intermediate conditions. 

 We suspect that the same may be true of the self condition also. It has 

 been pointed out elsewhere (Castle and Forbes, :o6; Castle, :o6), that the 

 theoretical absolute gametic purity, in Mendelian inheritance, probably does 

 not exist in any case. Heterozygosis leads inevitably to modification in 

 character of the conjugating gametes, not simply as regards the entire 

 assemblage of characters, but as regards each character considered sepa- 

 rately. The degree of modification is probably indicated roughly by the 

 imperfection of dominance in the heterozygote. Thus, when a perfect 

 blend is obtained, as in the inheritance of ear-length in rabbits, the gametes 

 transmit that intermediate character. But when dominance is very com- 

 plete, as in a cross between a self and an albino mammal, the segregation of 

 self and albino conditions is very complete among the gametes formed by 

 the cross-breeds. Yet when spotted individuals result from cross-breeding 

 between self and albinos, these spotted individuals form gametes which 

 transmit that same mosaic character. Now, the cross between self and 

 albino rats gives rats of Irish pattern, but quite variable. But our experi- 

 ments indicate that from variable Irish rats one can by selection obtain 

 rats transmitting no pattern but Irish. The theoretical importance of this 

 is obvious. Cross-breeding and selection combined are means by which 

 we may not only modify existing Mendelian characters, but may even 

 create new ones. They are, then, factors of prime importance in evolution, 

 even in the case of characters which vary discontinuously. 



