336 THE EVOLUTION THEORY 



Engadine, where the nut-ja}^ is abundant, I have often seen the 

 ground underneath the eembra-pines covered with the debris of 

 its meal. In addition to these differences between the two races, 

 the Alpine form is stronger in build, the Siberian form is daintier ; 

 in the former the white terminal band on the tail is narroAV (about 

 1 8 mm.), in the Siberian form it is broader (about 27 mm.). 



Such cases of variation of individual parts in different areas 

 seem to me very important theoretically, because they furnish us 

 with an answer to the view which represents the species as a ' life- 

 crystal,' which must be as it is or not at all, and which therefore 

 cannot vary as ]'egards its individual parts. The case of the nut-jay 

 has the further interest that it is one of the few in which we find 

 the new adaptation of a single character without variation of most 

 of the other characters. 



It is only in an essentially different sense that we can compare 

 the species, like any other vital unit, to a crystal, in so far as its 

 parts are harmoniously related one to another, or, as I expressed 

 myself years ago, are in a state of equilibrium, which must be 

 brought about by means of intra-selection. This analogy, however, 

 only applies to the actual adjustment of the parts to a whole, and 

 not to their casual adjustment. Species are variable crystals ; the 

 constancy of a species in all its parts must be regarded as something 

 quite relative, which may vary at any time, and which is sure to 

 vary at some time in the course of a long period. But the longer 

 the adaptation of a species to new conditions persists the more 

 constant, ceteris 'paribus, and the more slowly variable will it become, 

 and this for two reasons : first, because the determinants which are 

 varying in a suitable direction are being more and more strictly 

 selected, more and more precisely adapted, and are thus becoming 

 more like each other ; and secondly, because, according to our theory, 

 the homologous determinants of all the ids do not vary in the 

 required direction, and a portion of the unvaried ancestral ids is 

 always carried on through the course of the phylogeny, and only 

 gradually set aside by the chances of reducing division. But the 

 more completely these unvaried ids are eliminated from the germ- 

 plasm the less likely will they be to find expression in reversions 

 or in impurities of the new specific characters. I may recall the 

 reversions of the various breeds of pigeons to the rock-dove, those 

 of the white species of Datura to the blue form, and the Hipparion- 

 like three-toed horse of Julius Csesar, and so on. The unvaried 

 ancestral ids, which in these cases find only quite exceptional 

 expression, will make the new 'specific character' fluctuating, as 



