36 THE EVOLUTION OF LIVING BEINGS. 



deviating from the normal mendelian numbers, by the 

 failure of certain kinds of gametes to produce viable 

 offspring with other kinds of gametes, while certain 

 other kinds, mating together, do produce viable off- 

 spring; in other words the normal mendeUan numbers 

 are disturbed by preferential mating, a process known 

 to occur in other cases (gametic coupling and gametic 

 repulsion). 



All this is furthermore complicated by the death of 

 some kinds of gametes before they are in a position to 

 mate, as is shown by the sometimes very large proporti- 

 ons of abortive pollengrains and ovules, and comphca- 

 ted once more by the death, as shown by Renner, of 

 certain kinds of embryo's at a very early stage in their 

 development, aU or part of this, perhaps either caused 

 by, or compUcated through, irregularities in the meiotic 

 divisions of these heterozygotes. 



No unprejudiced investigator will accept the beha- 

 viour of such an evidently impure form as evidence for 

 the existence of mutation. 



The question remains : is there sufficient evidence to 

 accept the existence of such a novel process as mutati- 

 on wovild be, in other cases? 



The difficulty is again to obtain material of unim- 

 peachable specific purity to experiment with. 



We have seen that we possess no certain means to 

 prove specific purity in any case, but we know that it 

 is far easier to obtain the recessives in a pure state than 

 the dominants. 



Now it is a fact, well worthy of our careful conside- 

 ration, that whenever so called mutants are described, 



