iQiS] 



BA RT LETT— MA SS M UTATION 



453 



implicitly that he has made crosses only one way. If he had 

 studied the reciprocals of his crosses it is safe to assume that he 

 would never have ad- 

 vanced his Mendelian 

 explanation of muta- 

 bility. As far as his 

 results extend, his deriv- 

 atives of 0. Lamarcki- 

 ana fall, for the most 

 part, into two classes, 

 which conform in heredi- 

 tary behavior to the two 

 main classes of muta- 

 tions which have been 

 obtained from 0. pra- 

 tincola. 



Class I. — The muta- 

 tion breeds true, in the 

 sense that it gives no 

 reversions to the parent 

 form. The reciprocal 

 crosses with the parent 

 species are matroclinic. 

 The progeny conforms 

 to the type which sup- 

 pUes the female gamete. 



Class II. — The mu- 

 tation gives a progeny 

 consisting of the pa- 

 rental and mutational 

 types in greatly varying 

 proportions. The pro- 

 genies from reciprocal 

 crosses are mixed if the 

 mutation supplies the 

 female gametes, but consist of the parental type only if the muta- 

 tion supplies the male gamete. 



'. /Oc/p ~ 



Fig. 15. — Mut. gigas (above) and f. typica (be- 

 low) ; rosettes from the F3 progeny of f . lypica; the 

 rosette of mut. gigas, Lexington E-5-238, had a 

 darker color and more conspicuous pubescence than 

 the sister plant of f. lypica, but the difference does 

 not appear in the photograph; both plants are 

 sho"wn in fig. 5. 



