1899] EVOLUTION AND CHANCE 361 



upward from the lowest forms of life, we see this principle constantly 

 at work ; each species is guided by its own phylogeny, by the course 

 of its own development. Though it may deviate, it must not go 

 beyond certain limits. Thus primitive reptiles had before them at 

 least two possibilities — avian and specialised reptilian lines of develop- 

 ment. A species of bird, if it develops further, must modify the 

 avian type, or the type of its own particular order, but it cannot 

 become a reptile. The amphibian type is still further beyond the 

 region of possibility. As we ascend, it will be seen that the lines along 

 which species may develop, though certainly not less in number than 

 lower in the scale, are less divergent one from another. A primitive 

 bird was evidently capable of evolving an enormous number of 

 specialised forms, but the distance between these is as nothing when 

 compared with the distance between the descendants of primitive 

 Protozoa. The more elaborated and complex the species, the firmer 

 the hand with which its own phylogeny guides it. 



In describing this limiting principle I may have seemed to be 

 speaking of known forms as the only possible ones. I am far from 

 thinking that. The known species, living and fossil, may be far less- 

 in number than those of which we have neither record nor living 

 representatives. I only maintain that at any 'particular stage in the 

 evolution of a species the paths open to it are limited in number. This 

 sets no limit to the number of forms that may eventually arise. For 

 each advance not only leaves behind a number of possibilities, but 

 opens up a number of others. The particular number of variations 

 possible at any stage we cannot of course discover. But there are 

 the following reasons for believing that they are not very numerous. 

 We have seen that for one-celled organisms the only possible way of 

 advance in size is by the coherence of the resultant cells after fission. 

 When we come to the more complex animals we know their various 

 parts are tied together on the principle known as correlation. The 

 subject of correlation is very imperfectly understood. What is known 

 of it enables us to see that the organism is no mere aggregate of units, 

 but that a variation in one part involves a variation in several others. 

 Hence the organism's capacity for variation does not increase so fast as 

 its complexity increases. Of the limitation which we must attribute 

 to correlation there is abundant evidence. There is the experience of 

 breeders already mentioned. They have made many new breeds, and 

 often with only small numbers to select from. To pass on to other 

 evidence of limitation we find among evolutionists a growing tendency 

 towards the view that species, and even larger groups, not seldom tend 

 to converge. In the Arthropoda it is probable, or at anyrate possible, 

 that we have instances of convergent evolution. There is reason to 

 believe that the ostrich and its allies illustrate the same principle ; 

 they are probably polyphyletic in origin, and have converged towards 

 one another. The dentition of the marsupial animals, according to 



