86 



Mr. Hainan and Dr. Marshall. 



(6) There is no evidence of the existence of a compensatory mechanism 

 between the testes and the thymus. 



The work was carried on at the Field Laboratories, Cambridge. The 

 operations were done by F. H. A. Marshall ; the weighings and the chief 

 part of the other work by E. T. Hainan. The expenses were defrayed by 

 a grant made by the Board of Agriculture and Fisheries out of funds placed 

 at their disposal by the Development Commission. 



Note hy G. Udny Yule. 



In view of the disagreement with Prof. Paton's conclusions, Dr. Marshall 

 asked me to investigate the probable errors of some of the comparisons made, 

 with especial reference to the alleged effect of extirpation of the thymus on 

 the growth of the testes. 



The problem was not an easy one. A glance at Prof. Paton's figures, or at 

 the corresponding data given by Hainan and Marshall, will show how 

 exceedingly variable are the weights of the testes and how much caution 

 must consequently be used before basing any conclusion on a small 

 difference between the average weights for two groups of some 20 to 

 30 animals. Considerable differences might be shown even by the averages 

 of groups treated in precisely the same way. Were the animals adult, the 

 " probable error " of the difference between any two observed averages — the 

 amount which it would be as likely as not to exceed owing to mere 

 fluctuations of sampling- — might be readily obtained in the ordinary way. 

 But the animals are not adult ; the weight of the testes increases very 

 rapidly with the weight of the animal, and the weights of the different 

 individuals themselves vary greatly, so that the two groups of operated and 

 controls are not strictly comparable as a whole. 



What I finally decided to do, therefore, was this : To obtain, by known 

 methods, equations expressing as closely as possible the relation between 

 mean weight of the testes and body-weight, for operated and for normal 

 animals, and to see whether the constants in these equations differed more 

 than could be expected owing to the chances of sampling alone. As in 

 Prof. Paton's data the weight of testes did not seem to be a linear function 

 of the body- weight, and it was these data that I first investigated, the 

 logarithm of the testes-weight was substituted for the actual value, and 

 this seemed to give an approximately linear relation, judging from the 

 diagram (fig. 6). The two equations, with the probable errors of the constants 

 which I finally obtained from Prof. Paton's data, including all the 23 normal 



