EFFECTS OF INBREEDmG AND CROSSBREEDING. 



19 



roughly the relative reliability of the averages. The number of 

 litters of each size produced by the inbred families between 1906 and 

 1915 was as follows: 



Size of 

 Utter. 



Number. 



Size of 

 litter. 



Number. 



1 

 2 

 3 

 4 



1,187 

 2,230 

 2,253 

 1,111 



5 

 6 



7 

 8 



3«7 



100 



19 



2 



Data for later years are discussed in more detail in Bulletin 1121. 

 There was a considerable smaller average size of litter than in the 

 preceding years, which made a different index deshable. In this 

 bulletin the same index is used for 1916 to 1920 as in the earlier years. 



In the above data we see that litters of 1, 2, 3, and 4 occurred ap- 

 proximately in the ratio of 1 : 2 : 2 : 1, while large litters were much 

 less numerous. The young in these litters were in the ratio 1 : 4 : 6 : 4. 

 Use of the latter ratio would give too much importance to the larger 

 litters, because of the obvious tendency for litter mates to resemble 

 each other in birth weight, rate of gain, and fate. On the other hand, 

 averages based on a given number of litters of 4 are more reliable than 

 ones based on the same number of litters of 1. In the case of birth 

 weight it has been noted that variations within litters and variations 

 among the means of litters of a given size are about equally important. 

 It has accordingly seemed best to adopt a compromise between the 

 ratio based on litters and the ratio based on individuals. The ratio 

 1:3 : 4 : 2 has thus been arrived at. Where two stocks differ con- 

 sistently in aU sizes of litters it makes little difference what ratio is 

 used in calculating indexes. Where they do not differ consistently, 

 no index is of much value. 



ENVIRONMENTAL CONDITIONS. 



Tabulations of the various characters among the inbred families 

 were originally made according to generation. The results, however, 

 were irregular. Family 2, for example, reached its highest point in 

 size of litter in the fifth generation of mbreeding. The high pomt in 

 Family 23 was a generation earher. The variations in other characters 

 tended to agree with those for size of litter in a given family. A 

 study of the curves suggested that much of the variation was probably 

 due to variations in conditions from year to year. In order to inter- 

 pret the results with safety it is necessary to bring out as fully as possible 

 all causes of change other than mbreeding. Thus tabulations were 

 made with the year instead of the generation as a unit. As all 

 mbreeding began in 1906 and the origmal matmgs produced no young 

 after the sprmg of 1907, each year can safely be assumed to represent 

 a higher average degree of inbreeding than the preceding year. 



