256 D. F. JONES. 



whether or not any selective action was shown by the plant's 

 own pollen as compared to that from a plant of somewhat 

 different type. If one member of the pair of plants which 

 furnished the pollen for the mixture is designated A and the 

 other B, the two kinds of seeds grown on A plants are A X A, 

 self- fertilized, and A X B, cross-fertilized; on the B plants 

 B X A cross-fertilized and B X B self- fertilized. Since the same 

 pollen mixture is applied to both, the ratio of the seeds resulting 

 from A pollen to the seeds resulting from B pollen, on A plants, 

 should be the same as the ratio of the seeds resulting from A and 

 from B pollen on B plants. In other words, the numbers form 

 a proportion which, irrespective of the relative amounts of A 

 and B pollen in the mixture, should be a perfect proportion 

 within the limits of random sampling if fertilization takes place 

 at random. The end terms of the proportion comprise the 

 self-pollinated seeds and the middle terms the reciprocally cross- 

 pollinated seeds. If a true proportion is obtained the products 

 of the end terms should naturally equal the products of the middle 

 terms. If they do not the deviation is either in an excess of 

 cross-fertilized or of self-fertilized seeds indicating a selective 

 action in one or the other direction. 



The advantages of this method of attacking the problem are 

 readily apparent. It is practically impossible to make up a 

 mixture of large amounts of pollen in which the proportion of 

 each is known. Either measuring or weighing the pollen is 

 out of the question because maize pollen takes up moisture from 

 the air rapidly and when any quantity is brought together it 

 becomes aggregated into a flocculent mass. Moreover this pollen 

 soon loses its viability so that even in case equal numbers of 

 pollen grains could be had there would be no proof that there 

 were equal numbers of functional pollen grains in the mixture. 



The method of reciprocal application and arranging the results 

 in the form of a proportion automatically overcomes all these 

 difficulties and the experiment is as simple as could be devised. 

 An attempt was made to have as nearly equal quantities of pollen 

 as possible by measuring out the two kinds roughly. But in 

 many cases the results showed that one kind of pollen was far 

 more effective than the other. This, however, does not destroy 



