FERTILIZATION IN PIMA COTTON. 15 



but the difference between the increases in the two cases was less 

 than twice its probable error. 



The converse proposition, that the ovules of flowers with short 

 stigmas are less likely to be cross-fertilized than those of flowers 

 having long stigmas, would seem to be self-evident. It was not borne 

 out, however, by the results of an experiment performed by Balls 

 (#, pp. 118, 119) who compared two strains derived from an Egyp- 

 tian-upland cross, one of which had the stigmas so short as to be 

 surpassed by the uppermost anthers, while the other had stigmas 

 which greatly surpassed the anthers. No difference was found be- 

 tween the two strains in the percentage of hybrids resulting from 

 natural cross-pollination. Apparently in this case short stigmas 

 offered no effective obstacle to the access of foreign pollen. It is of 

 interest in this connection to note that the Lone Star variety of 

 upland cotton, in which the stigmas normally are exceeded by the 

 upper stamens, produced at Sacaton, Ariz., 5 per cent of vicinists 

 when grown in a row adjacent to a row of Pima (see Table 1, p. 5). 

 The probable explanation is that both in the hybrids compared by 

 Balls and in the Lone Star variety the density of the screen formed 

 by the stamens was not sufficient to protect the short stigmas from 

 access of foreign pollen. 



FLOWER STRUCTURE IN RELATION TO CROSS-FERTILIZATION. 



In Pima cotton the deposition of foreign pollen upon the ba^al 

 portion of the stigmas presumably is prevented by the density of the 

 surrounding girdle of stamens (PL I, Fig. 1; PL II, Fig. 1). An ex- 

 periment was performed in 1919 with the object of determining the 

 effectiveness of this protection. The material consisted of a row of 

 Pima plants having on one side a row of the Holdon variety and on 

 the other side a row of the Lone Star variety, the populations being 

 the same as in the vicinism experiment (Table 1). Flower buds of 

 all three varieties were opened before anthesis, and the extra- 

 staminal portion of the stigmas, if any, was excised, after which the 

 flowers were left exposed to the visits of insects. In the Lone Star 

 variety the stigmas usually are exceeded by the stamens ; hence little 

 or no excision was necessary in this case. In the Holdon variety the 

 portion excised was much shorter than in the case of Pima. 



The seed produced by the treated flowers of the three varieties 

 was planted in 1920. The population from seed borne by the Pima 

 plants was much larger than the upland populations, for the reason 

 that the proportion of the treated flowers which failed to set bolls 

 was much larger in the upland cottons 12 than in Pima, and the quan- 

 tity of seed produced was consequently much greater in the latter 

 case. Early in July, when the plants were well enough developed to 

 show their characters clearly, counts were made of the number of 

 hybrids in the three populations. The rows were not thinned, so that 

 all plants which survived the germination and seedling stages were 

 counted. 13 The results are given in Table 8. 



12 This does not indicate that the excision of the extrastaminal part of the stigmas 

 had been more injurious to the upland than to the Egyptian flowers, as the rate of boll 

 shedding at Sacaton, Ariz., is always much higher with upland than with Egyptian 

 cotton. 



13 Data given on page 5 indicate that there had been no natural selection in the earlier 

 stages of growth which would affect the percentages of hybrids. 



