562 THE AMERICAN NATURALIST [Vol. LIII 



Tlie.so experimonts, to this point, had been previously pub- 

 lished l)y the author (Tower, 1910). Further work on this 



the ciii-licr coMcliisions as to its composition and behavior. Care- 

 ful nii iiMiirniniis showed that this apparently pure signaticoUis 

 raee ])cli;ivr(l nofiiially as rcfjards the form-index (relation of 

 width tn Icntitli ), a specific eharaeter never dissociated, in other 

 expcMiiiciiis. h-diii the conspicuous pattern difference involved. 

 XumriDUs c\[)cnniciits with this peculiarly behavino' hybrid 

 faihd Willi oMc .x^.ption to find ain evidence of the clivn-^a 

 pattern factor {)resetit. Individuals of this race having the 

 (Jnursa (broad) form, crossed with one particular strain of sig- 

 vatu'oUis, grave in F.. four homoz.yfrous forms : pure signaticallis, 

 the peculiar true-breedmor hybrid {divfr^ia form with signati- 

 coJIis pattern), pure diversa, and a new form, with a different 

 pattern. Tliese appeared in the ratio of 4 : 2 : 1 : 1 (actual num- 

 bers not given), with the corresponding array of heterozygotes. 



At first sight, all this seems easy to account for: one or both 

 of the parents in the original cross were heterozygous for one or 

 more factors. The actual results, however, can not be accounted 

 for. according to any known genetic principles, on the basis of 

 the facts <invu. There are m.iin thinos at every stage of this 

 work, which we should like to know, about which we are told 

 norhin^j. Ttir aiiihor siaies rhar the original parents were *'not 

 heter(»z\ 'joiis. 1,111 no cvuhMice of any genetic analysis is given. 

 In tact, rlie aiitlior (i.-claivs nion- than once that he is not inter- 

 esleil 111 ordmai-N Meiideliaii analysis, implying that it is unim- 



The author liimvli shows \ei\ umipletely, in later chapters, that 

 'the>ie sj[)ecies, as found m nature, are genetically very hetero- 

 geneous. He also emphasizes that the results here obtained can 

 7)robably not be repeated with material from other locations or 

 even from the same location at other times. Again, we know 

 nothing of the genetic const if iil ion ..f the <^hinaticoUis strain 

 used in the "test reaction" wh- i. t^l.ha) (l>vrrm is recovered; 

 (\cept that onl> this parii. ular strain will give these results. 

 I r array in this case shows that tlwre are at least two factor- 

 !M cs involved, while there is only one in the ordinary cros.s 

 , '//^ ' diversa. Other crosses, which will immediately 

 _ ^ th. tiisehos as bearing on the problem, have not been 

 iiuhli' or are, at least, not mentioned. 



