January 20, 1922] 



SCIENCE 



59 



a hybrid is pa-rtially sterile, and from it any 

 fertile offspring can be obtained, the sterility, 

 once lost, disappears. This has been the his- 

 tory of many, perhaps most of our cultivated 

 plants of hybrid origin. 



The production of an indubitably sterile hy- 

 brid from completely fertile parents which 

 have arisen under critical observation from a 

 single common origin is the event for which 

 we wait. Until this event is witnessed, our 

 knowledge of evolution is incomplete in a 

 vital respect. From time to time a record of 

 such an observation is published, but none has 

 yet survived criticism. Meanwhile, though our 

 faith in evolution stands unshaken, we have no 

 acceptable account of the origin of "species." 



Curiously enough, it is at the same point 

 that the validity of the claim of natural se- 

 lection as the main directing force was most 

 questionable. The survival of the fittest was 

 a plausible account of evolution in broad out- 

 line, but failed in aioplication to specific dif- 

 ference. The Darwinian philosophy convinced 

 us that every species must "make good" in 

 nature' if it is to survive, but no one could 

 tell how the differences — often very sharply 

 fixed — which we recognize as specific, do in 

 fact enable the species to make good. The 

 claims of natural selection as the chief factor 

 in the determination of species have conse- 

 quently been discredited. 



I pass to another part of the problem, where 

 again, though extraordinary progress in 

 knowledge has been made, a new and formid- 

 able diffl-culty has been encountered. Of varia- 

 tions we know a great deal more than we did. 

 Almost all that we have seen are variations in 

 which we recognize that elements have been 

 lost. In addressing the British Association in 

 1914 I dwelt on evidence of this class. The 

 developments of the last seven years, which 

 are memorable as having provided in regard 

 to one animal, the fly Drosophila, the most 

 comprehensive mass of genetic observation yet 

 collected, serve rather to emphasize than to 

 weaken the considerations which I then re- 

 ferred. Even in Drosophila, where hundreds 

 of genetically distinct factors have been iden- 

 tified, very few new dominants, that is to say 



positive additions, have been seen, and I am 

 assured that none of them are of a class which 

 could be expected to be viable under natural 

 conditions. I understand even that none are 

 certainly viable in the homozygous state. 



If we try to trace back the origin of our 

 domesticated animals and plants, we can 

 scarcely ever point to a single wild species as 

 the probable progenitor. Almost every natu- 

 ralist who has dealt with these questions in 

 recent years has had recourse to theories of 

 multiple origin, because our modern races have 

 positive characteristics which we cannot find 

 in any existing species, and which combina- 

 tion of the existing species seem unable to 

 provide. To produce oiu' domesticated races 

 it seems that ingredients must have been added. 

 To invoke the hypothetical existence of lost 

 species provides a poor escape from this diffi- 

 culty, and we are left with the conviction that 

 some part of the chain of reasoning is miss- 

 ing. The weight of this objection will be most 

 felt by those who have most experience in prac- 

 tical breeding. I can not, for instance, imagine 

 a round seed being found on a wrinkled vari- 

 ety of pea except by crossing. Such seeds, 

 which look round, sometimes appear, but this 

 is a superficial appearance, and either these 

 seeds are seen to have the starch of wrinkled 

 seeds or can be proved to be the produce of 

 stray pollen. Nor can I imagine a fem- 

 leaved Primula producing a palm-leaf, or a 

 star-shaped flower producing the old type of 

 sinensis flower. And so on through long 

 series of forms which we have watched for 

 twenty years. 



Analysis has I'evealed hosts of transferable 

 characters. Their combinations suffice to sup- 

 ply in abundance serie's of types which might 

 laass for new species, and certainly would be 

 so classed if they were met with in nature. 

 Yet critically tested, we find that they are not 

 distinct species and we have no reason to sup- 

 pose that any accumulations of characters of 

 the same order would culminate in the pro- 

 duction of distinct species. Specific difference 

 therefore must be regarded as probably attach- 

 ing to the base upon which these transferables 

 are implanted, of which we know absolutely 



