ON USEFUL SUBSTANCES IN CACTUS 



of their combination in a predicted manner was 

 only one in something over a million. In the case 

 of the cactus the factors for spininess doubtless 

 number far more than ten; from which it follows 

 that the probability that any given seedling will 

 have germ plasm absolutely free from any of the 

 factors for spininess is much less than one in a 

 million. 



This explains why it was necessary, in our ex- 

 periments at Santa Rosa, to plant the seeds by lit- 

 eral millions, and to select persistently among un- 

 computed multitudes of seedlings. 



Fortunately the spiny condition reveals itself 

 almost from the outset, so that it was possible to 

 weed out the vast majority of all the seedlings, 

 retaining only, perhaps, a stray dozen or so from 

 among the legions. 



As the experiment proceeded, however, it was 

 gratifying to note that in succeeding generations 

 there was an ever-increasing proportion of spine- 

 less seedlings. This suggests that some of the fac- 

 tors for spininess were being dropped out of the 

 heredity of the selected plants. 



Obviously this seems to aUgur that should the 

 experiment be carried forward through a suffi- 

 cient number of generations, the time will prob- 

 ably come when all factors for spininess will have 

 been eliminated from the germ plasm of the 



[263] 



