138 ON VARIABILITY AND ADAPTATION 



atavism occurring after centuries of domestic breeding and loss 

 of the features in question. So in such a case of atavism we 

 must, in the terms of this theory, suppose that in the act of 

 fertilization there is a summation and preponderance of just 

 that series of unchanged ids which now in development lead to 

 the bringing into evidence of the atavistic ancestral features. 

 In other words, the theory demands that the ids must be singularly 

 stable in constitution — that they grow and multiply, but retain 

 the same structure. 



But now Weismann has to admit that, under certain condi- 

 tions, the ids are modified in their structure. This admission 

 indeed is contained in the idea that the individual hypothetical 

 ids vary in their properties ; and as, if we trace back these 

 ancestral ids to their common source, they must all originally 

 have been identical in structure, we conclude that at the same 

 time they are both stable and capable of change in constitution. 

 Here indeed is the crux of the theory. How are we to define 

 and realize for ourselves the limits of alteration ? Natural 

 selection cannot explain the alteration, unless we fall back upon 

 the far-away hypothesis of multitudinous separate acts of crea- 

 tion in the beginning of things — affording a large number of 

 distinct idioplasms, — and even this hypothesis does not work 

 out satisfactorily. 



In the example already given of crossing of the old-estab- 

 lished breeds of barbs, fantails, and spots, we must imagine 

 that all the ids of each breed have been, in the germ cells of 

 successive generations, exposed to almost identical conditions, 

 and, as a consequence, modified along the same lines. Exposed 

 to the same influences in the course of many generations, it is 

 almost inevitable that all must become modified, for if there 

 were any large number of unaltered ancestral ids contained in 

 the germ cells it would inevitably occur that spots and atavistic 

 forms would frequently present themselves. But this does not 

 happen. Bach of these varieties of the pigeon breeds singularly 

 true. How, in short, are we to picture some of them passing 

 from germ cell to germ cell through all the long years in an 

 ancestral condition ? Put to this test, the theory breaks 

 down ; we cannot picture the necessary conditions. It is, in 

 short, an absurdity to regard the nucleus of the germ cell as 

 containing a colony of what are, to all intents and purposes, 



