650 



SCIENCE 



[N. S. Vol. XXXIX. No. 1009 



reciprocal crosses, like the reciprocals, also 

 breed true. 



Modifications of the double reciprocal 

 crosses, termed by De Vries sesquireciprocals, 

 may be made by combinations of the recipro- 

 cals with the parents in such a manner that 

 similar parent types occupy the periphery of 

 the formuli©— & X (« X &) =• &. (& X "») X 

 l = h, mXiiXm)=m, and (m X &) X 

 m^m. Other arrangements with the periph- 

 eral positions of the formulae occupied by 

 difPerent parent types give iterative hybrids — 

 hX(iX'>n)=m, {mX'b)X'b = l, mX 

 (m X ^) ^ ^^ ^iid (b X ™) X ™ = ">■• 



The explanation of this remarkable behavior 

 is not as yet determined. An attractive hy- 

 pothesis postulates the differentiation of 

 classes of gametes carrying the characters of 

 the parents in pure form and their appropriate 

 combinations either by selective fertilization 

 or through the elimination of such gametes 

 as do not fit into the assumed schemes of com- 

 bination. But these and other speculations 

 must await the results of cytological studies as 

 well as further experimentation. An interest- 

 ing peculiarity of these crosses is their very 

 high degree of sterility and it remains to be 

 seen whether the same phenomena will be 

 found in other QHnothera species crosses that 

 are more fertile. Thus it is possible that 

 numerous and varied types of gametes may be 

 developed by the hybrids, as theoretically 

 would be expected, but that physiological con- 

 ditions will allow only certain types to ma- 

 ture or function. 



A long and detailed account of twin hybrids 

 constitutes the third part. These classes of 

 hybrids were first discovered by De Vries 

 among hybrids of Lamarckiana with other 

 species of (Enothera. He has since greatly 

 extended his observations and finds twin hy- 

 brids also differentiated when certain " mu- 

 tants " from Lamarckiana are similarly 

 crossed (e. g., nanella, lata, scintillans, oblonga 

 and also Icevifolia). Brevistylis in such 

 crosses follows a Mendelian ratio and giga^ 

 gives intermediate hybrids. The twin hybrids 

 appear in the first hybrid generation which 

 consists of two sharply contrasted groups. In 



one group, termed loeta, the characters of La- 

 marclciana are strongly dominant over those 

 of the other parent. In the other group, 

 termed velutina, the characters of the other 

 parent dominate those of Lamarckiana. The 

 proportions of the Imta and velutina types ap- 

 pear to vary greatly in different cultures. 

 The velutina types breed true in the second 

 and later generations, but the Iceta forms were 

 found in certain cultures to split off new lines 

 of velutina in successive generations. 



Another pair of twin hybrids, very differ- 

 ent from Iwta and velutina, are distinguished 

 as densa and laxa. They appear in crosses be- 

 tween certain broad-leaved forms of American 

 biennis and cniciata with Lamarckiana or its 

 " mutants." The distinctions here concern 

 chiefly the form of the foliage and the number 

 of capsules over a given length of stem ; densa 

 is broad-leaved with thickly crowded capsules. 

 laxa has smaller leaves and capsules less nu- 

 merous. The densa type breeds true, the laxa 

 throws off in successive generations still an- 

 other form, atra, distinguished by dark green 

 leaves. 



With the twin hybrids may be found a 

 class of delicate and dwarfed plants, the seed- 

 lings of which are etiolated. These are named 

 gracilis and they are present in very diverse 

 proportions. Very many of the gracilis types 

 die as seedlings, a few with care may be 

 brought to maturity as narrow-leaved plants 

 mostly sterile. The classes of dwarfs that the 

 writer has reported for a number of crosses 

 between grandiflora and American hiennis 

 may correspond to this group. Finally among 

 the twin hybrids are occasionally found plants 

 of marked size or luxuriance which appear to 

 hold to the cultures as a whole a relation some- 

 what similar to that of gigas or semi-gigas to 

 Lamarckiana. This type is called hero. 



De Vries offers an explanation of twin hy- 

 brids, and of the gracilis and hero types 

 through his theory of mutation and intra- 

 cellular pangenesis. Mendelians may attempt 

 to find in the phenomena the results of recom- 

 binations of multiple factors, although they 

 will have difficulties in working out consistent 



