18 BULLETIX 1090, U. S. DEPAETMEZ^TT OF AGRICULTUEE. 



for the correlation of family means is doubtless wholly due to the 

 comnion factor, heredity of size. The correlation within families is 

 probably due in part to genetic differentiation within the families and 

 in part to environmental causes. 



SEX RATIO. 



Sex follows closely the laws of random sampling in guinea pigs. 

 The sexes have been produced in nearly equal numbers in all experi- 

 ments. Thus, in the tabulation of the controls bom in 1906 to 1920 

 there were 2,051 males and 2,007 females. Among the inbreds born 

 in the same years were 12,831 males and 12,529 females. Exten- 

 sive tabulations have given no indications of any connection between 

 sex ratio and size of litter or season or year of birth. In every other 

 character considered, by far the highest records were obtained in the 

 year 1910. The percentage of males in this year was 50. Again, if 

 there were any important extraneous causes determining sex, litters 

 exclusively of males and exclusively of females should be more 

 numerous than expected by the laws of random sampling. The 

 actual numbers, however, have been very close to those expected; 

 (580 males and females were born in litters of 2 to 5 containing only 

 one sex, during 1916 and 1917, where 590 was the expected number.) 

 The only contrary indication was in an apparent differentiation of the 

 inbred families with respect to sex ratio, which was somewhat greater 

 than would be expected by random sampling. 



INDEXES FOR GROWTH AND MORTALITY. 



The purpose in going into the causes which affect the various 

 characteristics of guinea pigs has been to bring out the precautions 

 necessary in studying the effects of inbreeding and crossbreeding 

 upon them. 



Size of litter, for example, has such important effects on the rate 

 of growth and the mortality among the young that it would be unfair 

 to compare such characters as the average weight at 33 days and the 

 percentage raised, in stocks with different average sizes of litter. It 

 is, however, desirable to obtain a single figure to express the record of 

 each experiment with respect to each character. Accordingly, 

 indexes have been calculated for the weights and mortality percent- 

 ages for each stock, based on a fixed number of litters of each size. 

 The averages of litters of 1, 2, 3 and 4 have been assigned weights of 

 1, 3, 4, and 2, respectively, the resulting sum being divided by 10. 

 An index in a given stock is thus the average which would be obtained 

 if 100 guinea pigs were picked at random from litters of 1, 300 from 

 Utters of 2, 400 from litters of 3, and 200 from litters of 4. The 

 different weights assigned to the different sizes of Utter measure 



