1915.] PURE SPECIES OF CENOTHERA. 13 



ative hybrids are most surprising in the degree of sterility or de- 

 layed germination shown. They are given in Table II., where are 

 also presented the records of four cultures sown in Petri dishes in 

 which the germination was complete as proved by an examination 

 of the residue. 



A comparison in Table II. of the record for culture 15.31 with 

 *i5.3i^ 15-33 with *I5.33, and 15.35 with *I5.35 will illustrate the 

 gain in germination that may come through sowing seeds in Petri 

 dishes. The percentages of germination presented above for the 

 hybrids of biennis and muricata must not be regarded as expressing 

 exactly the degree of seed fertility under the conditions of the 

 experiments since with the harvests of seed are frequently found 

 very many structures too large to be abortive ovules and too small 

 to be counted as " seeds " in the sense of falling within the limits of 

 seed size. These structures are probably undeveloped seeds but 

 only a microscopical examination can determine this point; if so, 

 their presence of course always lowers the percentage of zygotes 

 capable of giving progeny. 



Bearing in mind the fact that pollen sterility in biennis and 

 muricata is 50 per cent., or more and that pollen abortion in the F^ 

 hybrids is very much higher (in fact very little good pollen is pro- 

 duced) the total amount of sterility both gametic and zygotic is 

 simply amazing. Under such conditions how can the behavior of 

 these hybrids be looked upon as indicative of anything but a most 

 unusual situation, in itself very interesting, but far beyond the ex- 

 pectations of normal hybrid behavior. This remarkable degree of 

 sterility among the hybrids of biennis and muricata is perhaps ex- 

 treme for the Oenotheras, but it serves to illustrate conditions ex- 

 tensively present in the writer's experience and doubtless also in 

 the experience of others. 



De Vries has described the hybrids between biennis and muri- 

 cata as breeding approximately true which in the main has also been 

 my observation. Apparently largely upon this behavior and that of 

 certain other crosses he has reached the conclusion that hybrids 

 between species of CEnoihera are stable. In this opinion of De 

 Vries I cannot agree for my crosses between grandiflora and certain 

 small-flowered American species (Davis, '12 and '13), and between 



