J. W. H. Harrison 81 



been predominatingly melanic, which is diametrically opposed to their 

 actual condition. 



Similarly the fourth supposition, although almost certainly the correct 

 explanation of the more extraordinary of the Hagedoorns' and Bonhote's 

 rats, appears in itself inadequate to account for every peculiarity of my 

 F^ and F^ broods, although doubtless to its action one must assign the 

 overweighting of the paler classes and the weird designs of some of 

 the insects. Therefore we must fall back on the third consideration to 

 account for the bulk of the facts. In my opinion, this contamination of 

 the gametes, or, to put it bluntly, the inconstancy during crossing of a 

 character demonstrated by other and intraspecific experiments to behave 

 as a unit character, operating in conjunction with the circumstances 

 referred to in (4), affords the only satisfactory interpretation of the 

 experimental data. To the same effect, and powerfully supporting the 

 views just enunciated, is the evidence supplied by the imperfect segre- 

 gation observed in the imagines developed parthenogenetically from the 

 heterozygous F^ melanic female. One warning must be given ; all this 

 necessarily demands that in species- and race-hybrids this gametic 

 blending or contamination is more prone to occur than in crosses 

 between varieties of the same species. That being so, the occurrence 

 must be intimately connected with the inability of the chromosomes to 

 play their normal parts owing to partial or total incompatibikty. 



Now if changes in gametic values such as described be deemed 

 proved — and I see no other possible explanation when due regard is 

 given to the perfect Mendelian behaviour of the melanism in the mating 

 of T. crepuscularia with its black form delamerensis — one must be 

 prepared to admit the truth of Castle's contention, backed up by the 

 evidence provided by his researches into the genetics of hooded rats 

 and English rabbits, that unit characters are inconstant and therefore 

 capable of response to selection. Even if they are but slight, physiological 

 divergencies of low grade, but of the same type as those distinctive of 

 recognised species and races, must exist between individuals of the same 

 species. No matter how unimportant, if of the same order, these differ- 

 ences must have similar effects, and thus determine a change in what is 

 customarily reckoned a unit character, just as Castle insisted. 



At any rate, granting that a definite change in a unit character, me- 

 lanism to wit, has not been proved absolutely within the bounds of a 

 species, it must, nevertheless, be admitted that conclusive proofs have 

 been supplied that the given character has been modified during an 

 interspecific cross. If this is so, it seems a very minor extension of our 



Journ. of Gen. x 6 



