No. 629] SHORTER ARTICLES AND CORRESPONDENCE 565 



In connection with his selection experiments, the author has 

 im fortunately adopted the theory of ''trerminal variations" (?. e., 

 modifiability of the frcnos) since abandoned by its priiicii)a] pro- 

 ponent. Tower's evidciiiM'. like all 11k" oWwr evidence for this 

 theory, is based on a vfr\ cvidciit con I'lisidn of liiiitvp.'^ with 

 pheilotypes, and an appaffnt failurt' td rccdunizc th, rlhil im- 

 portance of rigid brol h > r-si.<:l i r mntiini in sncli ('.\p<Tinicnt.s. In 

 the same experiments, thi' antlidi' tincU a vci-y conipN'i.' fi'fntaf ion 



inflnenee of the ,.nviir,nnirnt. In r.Mitnist to these indecisive or 

 nc-ative results, thcv is at least one very beautiful and uncpies- 

 tionahlc dcnioMsi ral iuii of the effect of modifying factors (here 

 call.d ■ ocnctic impuriiics--). Of the three methods by which he 

 attempt.',! to limit the raimv of variation in the form-index of a 

 certain • • hiot y i)c ' " i a^ he calls it) of />. ))nilt ita)vat(v—\\Si\\\Q\y, 

 control of the cu viroiinicntal complex, ordinaiy selection of ex- 



