m 



THE AMERICAN NATURALIST [Vol.XLVII 



stood that when we speak of bicolored types we refer 

 simply to the somatic character, and, as stated provi- 

 sionally, we shall rank all of our bicolors, genetically, as 

 tricolors. 



Our chief problem resolves itself, therefore, into the 

 question of how the different types of tricolor behave 

 when mated to each other. 



Methods 



The following methods were used in these experiments : 

 Marking. — At first the guinea-pigs were marked by 

 means of a numbered aluminum disk attached to the ear 

 with wire staple. This method was unsatisfactory, as 

 the tags were frequently torn off and lost. A system of 

 ear holes was substituted, but this method had the disad- 

 vantage that the holes sometimes heal up in young ani- 

 mals. We know, however, of no better method. 



Records. — The young animals were each given a num- 

 ber taken consecutively, and opposite the first individual 

 of each litter the mother's and father's number was re- 

 corded together with the date of birth. A journal was 

 also kept in which the records of the various matings 

 were kept. 



Matings. — As a rule several females were mated simul- 

 taneously with a single male. In the early part of the 

 work the mothers were allowed to litter in the common 

 pen and the mother identified by the presence of milk in 

 her breasts. Sometimes two litters resulted at the same 

 time, in which case it was impossible to assign the young 

 to the proper mother. To avoid this, if more than one 

 female seemed likely to litter at the same time, the fe- 

 males were isolated until after they had littered. 



Charts.—A. young individual was killed and skinned 

 and the skin stretched just enough to hold it flat and 

 then dried. From this a cardboard pattern was prepared 

 and the outlines of all the sketches drawn from this. 

 The midline of the sketch was divided into six equal parts, 

 as an aid in locating the areas, and the various areas of 



