293 ^- tiuggtes Gates. 



Lamarckiana and one which is furthermore present in varying 

 degrees in various Qinothera races. The flowers of the Lamarckiana 

 of De Vries are distinctly larger than those of the gnindiflora forms 

 from Alabama. The Fj^ hybrids of Davis were, however, (as would be 

 anticipated) intermediate in flower-size between grandiflora and the 

 small flowers of biennis, the intermediacy also extending to all the 

 other characters, though forms with larger flowers have been obtained 

 in later generations. The reciprocal crosses were made, biennis x 

 grandiflora yielding eight plants, seven of which resembled grandi- 

 flora and were therefore patroclinous while one exhibited a certain 

 amount of crinkling of the foliage. O. grandiflora x biennis, from 

 the same parent individuals, produced twenty plants which were in 

 two groups (twin hybrids), nine being grandiflora-like in foliage and 

 habit while eleven resembled biennis. 



Larger cultures of these F, and Fj hybrids from fresh crosses 

 were made in 1910 and 1911. The chief defect of these papers as 

 descriptions of the hybrids is that the search for Lai)iarckiana-like 

 plants among the offspring was allowed to overshadow the really 

 more important general results of the crosses, since it was not to be 

 expected in any case that Lamarckiana could have resulted from a 

 single simple cross. In his last paper on these hybrids Davis virtually 

 admits the correctness of this position. 



In 1907 De Vries first described twin hybrids {lata and velutina) 

 in the Fj^ when Lamarckiana is used to pollinate biennis, Hookeri 

 and other species. Notes on the anatomical features of these 

 hybrids have been published by Andrews (1). The hybrid types 

 both breed fairly true when self-pollinated. This hybrid behaviour 

 has been confirmed by Gates (21) in the case of biennis x 

 Lamarckiana, with full illustrations of the broad-leaved and narrow- 

 leaved forms. Corresponding results were also obtained with biennis 

 X Icevifolia. It therefore appears that in Lamarckiana and its 

 mutant derivatives, including Icevifolia, the pollen grains are of two 

 types, leading to a constant dimorphism in the Fi of the hybrids 

 when these forms are used as male parent. In the reciprocal 

 crosses, however, De Vries obtained a single constant race, 

 indicating an absence of this type of dimorphism in the egg-cells of 

 Lamarckiana, though these results have not been fully corroborated 

 by the later investigations. 



This short review of the more recent hybridization results is 

 necessarily very incomplete, but even a general summary of the 

 data involved would be much too lengthy for the present purpose. 

 The results show clearly, however, that several types of hereditary 

 behaviour exist, and that these types depend upon the manner of 

 origin and hence the relationship to each other of the various races 

 and species concerned. Certain crosses give blended inheritance, 

 others give alternation or segregation, others twin types unlike 

 either parent, etc. A knowledge of the cytological conditions is 

 necessary for an explanation of the hereditary behaviour of such 

 mutants as gigas and lata, and also apparently of the results of 

 certain double reciprocal crosses. In conjunction, these two classes 

 of data furnish a consistent and rational view of very complex and 

 sometimes apparently conflicting phenomena. 



In connection with the twin hybrids pvoducedhy Lamarckiana, 

 Honing (25) has made an interesting comparison of Lamarckiana 



