Nature of Inherited Variations 215 



facts leave readily open the possibility of the building up of 

 a character by minute graded changes. 



In essentials, therefore, the study of mutations, when 

 carried so far as in Drosophila, is not in disagreement with 

 our observations of gradual variation in the Protozoa nor 

 with the conclusions of palaeontologists as to the gradual de- 

 velopment of the characteristics of organisms in past ages. 



These conclusions of the palaeonologists are well stated 

 in the recent work of Osborn (1917). He sets forth that in 

 following given stocks from earlier to later ages, characters 

 arise from minutest beginnings and pass by continuous 

 gradations to a highly developed condition. This seems 

 in agreement with the experimental results on both higher 

 and lower organisms, as I have tried to set them forth. The 

 palasontogolical evidence, he holds further, indicates that the 

 hereditary changes as one passes from age to age do not oc- 

 cur in random directions, but follow a definite course, which 

 might seem to have been predetermined in the constitution 

 of the organisms, or otherwise. In the experimental work 

 on the lower organisms little that indicates this has thus far 

 been observed. By selection we can move in more than one 

 direction ; though it is also true, of course, that the varia- 

 tions possible are limited by the constitution of the organ- 

 ism. The experimental work has hardly gone far enough 

 to offer important evidence on this problem. 



There is one other point in the work on higher organisms 

 that we may briefly consider. This is the point made by 

 Bateson (1914) in his Presidential Address before the Brit- 

 ish Association, and further developed in a recent paper by 

 Davenport (1916). It is the paradoxical proposition that 

 since practically all observed variations are cases of loss 

 and disintegration, we are driven to suppose that evolution 

 has occurred by loss and disintegration. Davenport com- 



