40 



THE AMERICAN NATURALIST [Vol. LYI 



(a) The Quality and Quantitii of fhr Chanf/r 

 In spite of the fact that the drawing of inferences 

 coneerniug the gene is very much liindered, in tliis 

 method, on account of the remoteness of the gene-cause 

 from its character-effect, one saUent point stands out 

 ah-eadv. It is that the change is not always a mere loss 

 of material, because clear-cut reverse mutations liave 

 been obtained in corn, DrosopMla, Portnhtcn, and prob- 

 ably elsewhere. If the original mutation was a lo-s, tlu^ 

 reverse must be a gain. Secondly, the mutations in many 

 cases seem not to be quantitative at all, since the different 

 allelomorphs formed by mutations of one original gene 

 often fail to form a single linear series. One case, in fact, 

 is known in which the allelomorphs even affect totally 

 different characters : this is the case of the truncate series, 

 in which I have found that different mutant genes at tlie 

 same locus may cause either a shortening of the wing, an 

 eruption on the thorax, a lethal effect, or any roniiii na- 

 tion of two or three of these characters. In >uch a case 

 we may be dealing either with changes of different types 

 occurring in the same material or with changes (i)ossi))ly 

 quantitative changes, similar in type) occurring in dif- 

 ferent component parts of one gene. Owing to tlie uni- 

 versal applicability of the lattei- interi)retation. even 

 where allelomorphs do not form a linear -crir^. it can 

 not be categorically denied, in any indivldnal ca-c. that 

 the changes may be merely quantitative ehan-c- of M)nie 

 part of the gene. If all change^ were tlni^ .|"antital ive, 

 even in this limited sense of a lo-^ ni- -nin of part of the 

 gene, our problem of why the cli;in-,Ml ovj,,. stiil seems 

 to l.e antocatalytie would in the main di^app<.ar, but such 

 "I'i >"ice in that case the 



chid 



