BREEDING RESISTANT VARIETIES. 51 



much emphasis cannot be laid on the desirabihty of concentrating one's 

 energy on the careful comparison and selection of a few mother plants, 

 which are to be kept separate, rather than on the "rogueing" of large fields. 



The Factor of Cross-Pollination. 

 When the selection of wilt resistant strains of cotton was first undertaken 

 considerable importance was attached to the close fertilization of the flowers 

 on the selected plant, since it would seem that if they received pollen from 

 adjoining non-resistant plants the seed produced would be less resistant. Our 

 field experiments have not supported this view, however, and at present little 

 importance is attached to the covering of selected plants to protect them from 

 foreign pollen, though as a precautionary measure the plants are isolated as 

 much as possible, in order to reduce the amount of cross-pollination. In 

 our experiments a large number of flowers were covered with bags and al- 

 lowed to fertilize themselves. The seed thus obtained was planted on wilt 

 infected land besides some from open fertilized bolls. The results showed 

 no advantage from self-pollination either in uniformity of type or resistance 

 to disease. The plants produced from uncovered flowers were, if anything, 

 more vigorous than the others. It should be said in this connection that it 

 IS probable that a considerable percentage of cotton blossoms in the open 

 field are normally close fertilized. 



Fixing the Type. 

 It was thought at the beginning of the work that the selection for re- 

 sistance would have to be continued for two or three generations in order 

 to fix the type. E.xperience has shown, however, that with cotton our re- 

 sistant types are fully fixed from the beginning. In our present work, if the 

 offspring of any selected plant fail to show this uniform resistance, they are 

 all discarded. Such cases are few, and may be due to an error in selecting as 

 resistant a plant never exposed to infection, or the plants may have lacked 

 prepotency. 



Effect of Fertilization. 



In our work on the wilt diseases it has been found important to iiave the 

 land where the selections are to be made well fertilized, particularly with 

 stable manure or other organic substances. The resistant plants attain a 

 better development under such circumstances, and it is easier to distinguish 

 between resistant and partially diseased plants. 



Selection vs. Hybridization. 

 In breeding wilt resistant varieties of cotton it has been found possible 

 to attain the end desired by selection alone, without having recourse to 

 hybridization. Hybrids have been made in connection with the work, particu- 

 larly with Egyptian cotton, and have proved very resistant (PI. III.), 

 but the difficulty of securing a fixed type of commercial value is so great 

 that we have not preserved them, since selections from well fixed varieties 

 answered the purpose better. With watermelons the same objection to 

 hybridization holds with greater force on account of the difficulty usually- 

 experienced in fi.xing hybrid cucurbits. In working with other crops and 

 other diseases it may be desirable to use hybrids, but under our conditions 



4 



