No. 636] GUINEA PIGS AND TUBERCULOSIS 33 



more the growth of those guinea pigs which had been 

 gaining rapidly than of those which had more growth to 

 make up. The small positive correlation between weight 

 at 33 days and age at inoculation reflects a decline in the 

 condition of the stock, due probably to the use of an in- 

 ferior quality of hay, which took place during the sum- 











1 































































































V 







































































































































\ 























































































c 



HAH 



























































mer of 1919, especially during the time in which lots 10- 

 14 were being raised. 



With the help of these correlations, we can calculate 

 the degree to which variation in length of life is deter- 

 mined by each factor separately and by all of them com- 

 bined. For the last purpose, Pearson's coefficient of 

 multiple correlation can be used. This coefficient comes 

 out + .136 in lots 7-9 and + .251 in lots 10-14 for the cor- 

 relation between length of life and rate of gain after 33 

 days, age and weight at inoculation combined. The 

 degree of determination is measured by the square of 

 this coefficient. These three most important factors 

 combined, therefore, determine less than 2 per cent, of 



