58 CROSS-FKRTILISATION IN TIIF. INDIAN COTTONfi. 



colour introduced into the flower, the foUage shows the impure con- 

 dition with the red colour as far as the veins only. These two crosses 

 must have been produced by a cross from a yellow or impure red, 

 parent and the nearest plant of either kind is one with a yellow flower 

 situated at a distance of lOft. In the third case the full red foliage 

 colour is retained and the cross is indicated by the introduction of 

 the yellow only. This can only be effected by a pure, or by an im- 

 pure, red. The nearest plant of either kind is a pure red situated 

 at a distance of 13ft. 



In Diagram I are given a group of eight plants with their neigh- 

 bours ; though each of the eight is situated within 2ft. Gins, of at 

 least one plant the cross from which would be recognisable in the 

 offspring, yet in four out of the eight cases no such cross has occurred. 

 The facts here detailed are insufficient to justify any absolute 

 statement, but clearly point to the conclusion that, for practical 

 purposes, there is no objection to cultivating different races in ad- 

 joining plots provided the two lines bordering the second race are 

 discarded for the purposes of seed supply. Additional security, 

 and one which might render even this precaution unnecessary, may 

 perhaps be obtained by the separation of two kinds with one or two 

 lines of a crop which flowers fi-eely during the cotton season and will 

 grow under the same conditions as cotton. For this purpose the 

 Indian Sann {Hibiscus cannahinus) or H. Sahdariffa naturally sug- 

 gest themselves. 



The insects visiting the cotton flower. 

 It is not proposed to devote more than a cursory glance to this 

 aspect of the subject which, as far as India is concerned, has been 

 investigated in some detail by Burkill.* The diversity of the in- 

 sect flora in the various cotton tracts of the world prevents the 

 deduction of general conclusions from observations in one area. Such 

 conclusions must be, to a greater or less extent, of local application 

 oidy. Certain points in this local aspect may, however, be referred 

 to as having some bearing on the experiments above noted. 



* Loc. cil. 



