276 



Gates. 



its two parents, so that, as with various otlier characters, 

 fractionation and blending occurs. Such, behaviour slrows 

 that instead of one or a hundred "units" for time of blooming, 

 we have a blended condition which is fractionated again by 

 crossing back. It is probable that this process of fractionation 

 or dilution can go on indefinitely by continued back-crossing. 



15. The double-reciprocal liydrids from grandiflora , rubricalyx 

 and ritbn'calvx grandiflora give data which are different from 

 the results of the corresponding crosses made by de Vries 

 between 0. biennis L. and 0. )nuricata L. de Vries obtained 

 reversion to biennis or iiviiricata according to tlie way the 

 cross was made. Thus (biennis ■ murieata) ■ fmuricata x 

 biennis) gave biennis, and fmuricata biennis) ■ (biennis x 

 murieata) gave murieata. In tlie corresponding grandiflora- 

 rubriealyx crosses, the double-reciprocal hybrids were very 

 variable in foliage characters, and though decidedly nearer 

 the "outside grandparents", they certainly did not agree with 

 them, tlie results being much more like the F.2 crosses. These 

 crosses were not precisely reciprocal, because the races of grandi- 

 flora used were different in the two cases, but the difference 

 was probably not great enough to be significant, for the Fg 

 families from the original reciprocal crosses gave very similar 

 series of forms. The difference between my results and those 

 of DE Vries may possibly depend upon the fact that, as Gold- 

 SCHMIDT has suggested, the biennis-muricata series of hybrids 

 shows merogonv; while the grandiflora-riibricalvx hvbrids almost 

 certainly do not^). 



16. One of tlie important, though incidental, results of these crosses 

 is the proof that mutation and hybridization are separate 

 phenomena, and that the cause of some at least of the muta- 

 tions in Oenothera is independent of the combination of hybrid 

 characters. The present series of hybrids has furnished the 

 crucial proof of this statement, and fully confirms the con- 

 clusions I h.ad reached as a result of prolonged cytological 

 investigations. Of the various mutants and aberrant forms 

 which occurred among these hybrids, four are of special interest. 

 Two of these were seuiilafa grandiflora, combining the characters 

 of the de Vriesian mutant scniilata with certain characters of 

 grandiflora obtained by inheritance. Two others were lata 



1) But see footnote p. 264. 



