82 



THE AMERICAN NATURALIST [Vol. XLIX 



pletely self-sterile, and apparently there was cross-steril- 

 ity in about 6 per cent, of the possible combinations. 



In the F 4 generation, 10 plants resulting from crossing 

 two sisters of the F 3 generation were selected for experi- 

 ment. Unfortunately, I was able to make only 58 inter- 

 crosses, 5 of which, almost 10 per cent., failed. 



Back crosses have furnished another line of experiment, 

 though they have not been carried on as systematically as 

 were those of Correns. Nearly 85 back-crosses using 

 plants from the progeny of four combinations which in- 

 cluded four individuals as parents, have been made. The 

 plants themselves all proved self-sterile, and in addition 

 5 of the back crosses failed. 



When these experiments were begun I expected to find 

 that the facts would accord with a simple dihybrid Men- 

 delian formula, similar to that which Correns later pro- 

 posed as an interpretation of his results, yet only by con- 

 siderable stretching and a vivid imagination will Cor- 

 rens 's data fit such an hypothesis, and my own data do 

 not fit at all. No self-fertile plants have been produced 

 by any combination, and cross- sterility is a possibility in 

 only from 1.5 to 10 per cent, of the combinations. Fur- 

 thermore, Correns 's idea of inhibitors appears unlikely 

 from some other data I have gathered with the help of 

 Mr. J. B. Park. Ten plants were involved in this experi- 

 ment. Pairs of plants were provided to furnish series 

 of selfed and crossed flowers. The pistils of these flowers 

 were fixed at regular periods after pollination, stained, 

 sectioned, and the pollen tubes examined. Fertilization 

 not later than the fourth day marked the end point of the 

 crossed series, the dropping of the flowers between the 

 eighth and the eleventh day ended the selfed series. As 

 the flowers on each plant had about the same length pistils, 

 curves of pollen tube development for both crossing and 

 selfing could be constructed. The pollen grains germi- 

 nated perfectly on stigmas from the same plant, from 

 1,200 to 2,000 tubes having been counted in sections of 

 single pistils. The difference between the development 



