EVOLUTIONARY MECHANISMS 



305 



substantiated phenomena (pp. 381, 382), but these agents do not produce 

 specific changes in the way suggested by Harrison. The actual facts of 

 the case, however, are still not entirely beyond doubt, since other 

 workers^ have failed to repeat the results and have suggested that the 

 original stock may have contained the melanic gene; Harrison has 

 criticized their work as involving too high a mortality, so that if any of 

 the weaker melanics were induced they would not have survived. Per- 

 haps it is necessary that the work should be repeated once more. 



Macdougall's^ experiments on learning in rats are the most careful 

 and complete of any purporting to give positive evidence of the inheri- 

 tance of acquired characters. He found that if rats are taught a task (to 

 choose a dark platform when presented with both a dark and a light 

 one) their offspring tend to inherit the capacity to learn and can be 

 more easily trained to choose a given one of the two alternatives. The 

 experiments were made with an inbred strain, and great care was taken 

 to avoid selection. So long as these experiments stand alone, as reason- 

 ably convincing evidence of the inheritance of an acquired character, 

 they seem scarcely strong enough to support a theory of evolution of 

 such enormous scope. Moreover, they have been repeated without 

 success by Crew,^ who, while not able to offer any complete explanation 

 of Macdougall's findings, failed to obtain a similar result and drew 

 attention to certain disturbing facts, such as definite evidence of genetic 

 heterogeneity for learning ability in the inbred stock he used, which 

 would make it possible for selection to be effective. 



^ Hughes 1932. - Macdougall 1927, 1930. ' Crew 1936 



