MEANS AND METHODS OF EVOLUTIONARY CHANGE 343 



the next oeneration. In each feneration some of the rats in the control line 

 remained untrained and were mated to produce the next generation in that 

 line, while other rats were trained, to provide a measure of learning ability 

 in the control line in the veneration concerned. In the control line trained 

 rats were never used as parents. 



The investigators found that during the first fifteen or sixteen generations 

 the number of errors in both trained and control lines decreased progres- 

 sively. The causes of this decrease are obscure, but it was certainly not due 

 to inheritance of the effects of training, since parents in successive genera- 

 tions of the control line were not trained at all. The greatest deficiency of 

 McDougall's experiments was lack of a control line. In the light of this re- 

 cent investigation, therefore, there remains no reason to consider that his 

 results afford evidence of inheritance of the effects of training. 



The experiments of Agar and his colleagues continued beyond the six- 

 teenth generation. In later generations the number of errors fluctuated; 

 generations in which few errors were made were followed by generations in 

 which the number of errors was larger, and vice versa. The noteworthy fact 

 is that both trained and control lines fluctuated similarly. Factors affecting 

 both lines must have been operative. The experimenters concluded that en- 

 vironmental factors (such as seasonal fluctuations in temperature) and 

 fluctuations in the health and vigor of the rat colony from year to year 

 were reflected in the observed variations in speed of learning. At any rate, 

 clear evidence was obtained against the hypothesis that the effects of 

 training are inherited. 



Through the years many experiments of varying kinds have been per- 

 formed as attempts to demonstrate inheritance of acquired characters, and 

 results of some of them have been interpreted as affording evidence of it. 

 Nevertheless, deficiencies in planning or technique, overlooked sources of 

 error, possibility of interpreting experimental data in more than one way 

 have invalidated all experiments known to the author. 



The line of attack which came nearest to yielding an exception to this last 

 statement deserves special attention. From an a priori standpoint it seems 

 difficult to visualize a mechanism by which changes in the body can be 

 transferred to the germ cells and thus become inheritable. The germ cells 

 are shut away in one small organ of the body (ovary in females; testis in 

 males). How can changes in the body get to them? Since they are supplied 

 with blood as are the other parts of the body, it is conceivable that sub- 

 stances carried by the blood might serve as "go-betweens," conveying to 

 the germ cells the effects of changes in the body. Years ago it occurred to 



