84 CARNEGIE INSTITUTION OF WASHINGTON. 



The seeds were then germinated between blotters at constant temper- 

 ature by iNIiss Anne M. Lutz in the United States seed laboratory at 

 Berkeley, CaUfornia. In this way the percentage of germination was 

 obtained in all cases. It varied all the way from to 100 per cent. 

 The seedlings were then planted in flats and the ungerminated seeds 

 examined for embryos. The results show that, except in a few cases, 

 no conclusions can be drawn from a knowledge of the percentage of 

 gennination and the number of empty seeds. There are also certain 

 sources of error. For example, the seeds sometimes vary in size so 

 enormously that it is difficult to detennine whether the smallest should 

 be counted as seeds. 



Among the plants grown this year were wild .species from California, 

 North Dakota, Nova Scotia, and elsewhere, whose characters and 

 variability were studied. A form belonging to Oenothera hooker i was 

 represented by 11 cultures, each from a capsule of seeds from a different 

 individual, collected in 1915 from the large population of Oenotheras at 

 Lake jMerced, near San Francisco. The percentage of germination in 

 these 11 cultures varied from 19.5 per cent to 100 per cent, and the 

 percentage of empty or nearly empty seeds from 54.6 per cent to 0. 

 The culture (No. 64) in which all seeds germinated contained originally 

 66 seedhngs, but of these only 26 survived until they were planted 

 out in May. The remainder died during the interval from January to 

 May, when they were kept in a greenhouse without artificial heat. 

 When the rosettes of the survivors developed, 20 were found to be of 

 the ordinary broad-leafed type and 6 were very narrow, linear-leafed 

 dwarfs. The original seeds were nearly all large and plump. It is 

 probable that those wliich died contained at least as many dwarfs as 

 the survivors, and some of the dwarfs at least must therefore have 

 come from large, plump seeds. 



The main results of these Oe. hookeri cultures were as follows: 9 of 

 the 11 famiUes were relatively unifonn, while 2 produced markedly 

 aberrant mutant types. One of the latter families was mentioned 

 above; the other (No. 63) contained 63 plants, representing 4 rather 

 well-marked types. Type (1) was the normal broad-leafed rosette 

 (37 plants) ; tj^ie (2) the linear-leafed dwarf found also in culture No. 

 64 (5 plants); type (3) narrow-leafed and clearlj^ marked (13 plants); 

 type (4) rather narrow-leafed, paler green, smaller and more compact 

 rosettes, these shading into the broad-leafed condition (8 plants). 

 The appearance of these strikingly aberrant types, many of which have 

 since come into bloom, in Oe. hookeri, was all the more unexpected, as 

 my previous cultures of this species had proved very unifonn. Some 

 of the remaining families contained plants, all of the broad-leafed type 

 above mentioned, although they mmibered hundreds of individuals. 

 This shows that individuals which were extei'nally alike may breed 

 true or may produce a large percentage of aberrants. The remaining 

 families, although by no means unifonn in all their characters, were in 



