TEE AMERICAN NATURALIST [Vol. XLV 



thera, a blend of the parental characters, and at the same 

 time agree with 0. Lamarckiana. 



Similarly, in the description of the foliage of his hy- 

 brids, Davis fails to take into account the leaf -differences 

 involved. 0. grand if ora agrees with most American 

 races of 0. biennis, in having stem-leaves which are 

 lanceolate and tapering at both ends, forming a petiole. 

 In 0. Lamarckiana the stem-leaves are, on the contrary, 

 usually sessile and broad, almost clasping, at the base, in 

 addition to being crinkled. 



A form which would resemble 0. Lamarckiana much 

 more closely than the hybrids of Davis, could doubtless 

 be synthesized by using a race of 0. biennis which I have 

 cultivated from seeds obtained from the Chelsea Physic 

 Garden. This race differs strikingly from any other race 

 of 0. biennis I have seen, in having stem-leaves sessile, 

 very broad at base and very much crinkled. If such a 

 race were crossed with 0. grandiflora, hybrids might be 

 expected whose foliage at least would closely resemble 

 that of 0. Lamarckiana, though the flowers would prob- 

 ably be intermediate between 0. Lamarckiana and 0. 

 biennis. 



One feature which has always made it appear improb- 

 able that 0. Lamarckiana originated as the result of the 

 single cross 0. grandiflora X 0. biennis, is the way in 

 which the mutants of 0. Lamarckiana all agree with that 

 form in certain characters; and not one of them, so 

 far as I am aware, shows any indication of a reversion 

 towards either 0. biennis or 0. grandiflora. That the 

 mutants all have certain features in common, which they 

 do not share with 0. biennis or 0. grandiflora, and that 

 all may be grouped around O. Lamarckiana, was shown 

 in my analytical key to these forms (Gates, '09a, pp. 126, 

 127). While there are wide departures from 0. La- 

 marckiana in many characters, in no case does this ap- 

 pear to bring them nearer either 0. biennis or 0. grandi- 

 flora races. It is difficult to understand why tbis does not 

 occur, on the theory that 0. Lamarckiana is such a 



