38 



sess in the highest degree the tendency to proliferation, and to se- 

 cure seeds from them. The task, however, is peculiar, and more 

 difficult than such experiments usually are, because there is little or 

 nothing in the way of an external clue to the desired character. 

 It may be necessary to cut open each infested square in succession 

 to make sure that the plant is allowing no weevil larvae to develop. 

 And after the most promising plants have been located, it may be 

 possible to obtain seed from them only by artificially protecting 

 them from the weevils. Otherwise the best stock might be lost if 

 the weevils were very abundant. Indeed, this suggests a reason 

 why ' gelatinization' has not become a fixed character already. 

 Selection thus far has only been in the direction of proliferation in 

 the bolls, since the proliferation of tissue in the buds would give a 

 particular plant no advantage over its neighbours in the matter of 

 seed production. It would enjoy no immunity from subsequent at- 

 tack because it had not allowed any weevils to reach maturity. 

 Weevils from other plants would continue to come to it, and the 

 chances of ripening seeds would not be increased. There has been, 

 in other words, no selective inducement for * gelatinized' buds to be- 

 come a uniform character except as they might be correlated with 

 'gelatinized' bolls, in spite of the fact that for killing the weevil 

 proliferation in the buds is more important than that in the bolls. 



These considerations reveal still another episode of evolutionary 

 history, and may explain why it is that the variety protected by 

 the ants, and the other ' upland' types which have originated in 

 the same region, have the additional protective adaptations. It 

 was only where the ants protected the cotton and thus perpetuated 

 it as a field crop that these other considerations could have a 

 cumulative effect. The other adaptations by which the tree cottons 

 have maintained a desultory existence are of suggestive interest, 

 but of apparently little practical importance, since no field culture 

 of a perennial cotton seems to be maintained in any weevil-infested 

 district. 



In eastern Guatemala the cultivation of cotton as a field crop is 

 strictly limited to localities suited to the ants, where they exist in 

 such numbers as to give practical protection. In Texas, however, 

 cotton is grown under a great variety of conditions. The climatic 

 vicissitudes of heat and cold, drought and flood are many times as 

 great as in Guatemala, so that notwithstanding the unexpectedly 

 great adaptability of the kelep, it can not be expected to thrive 

 equally well in all parts of the state, any more than does the 

 weevil. Even if it be found that the ants can thrive, breed and 

 establish new colonies in Texas, they will probably require many 

 years to take full and effective possession even of the more 

 favourable localities of this vast agricultural empire. Such a 

 mitigation of the weevil's injuries would be, of course, of great 

 practical value, and the work of the ants in destroying the larvae 

 of boll worms and leaf-worms might be only slightly less im- 

 portant in some districts. If, however, the hope of exterminating 



