140 Doubles in Stocks 



2. In order to obtain direct proof that a successful result depends 

 upon the selection for planting of the most vigorous seedlings the 

 following method was pursued. The sowing was made in the autumn, 

 the seeds being sown, as usual, as evenly as possible in pots, and allowed 

 to grow undisturbed until it became necessary to give the young plants 

 more room. They were then repotted singly and the batch from each 

 pot sorted into two groups according to the surface character, the 

 one gi-oup consisting of the fully hoary, the other of the partially hoary 

 and the glabrous plants. Each group was then numbered off in order 

 of size, the largest being labelled 1, the next largest 2, and so on in 

 orders The repotting was carried out in February and March when 

 the young plants were at most 3 — 5 inches high, long before it was 

 possible to determine the character of the flower since the earliest 

 plants to flower only began to show buds in the latter part of May, 

 many not until June and July and a large number not until the 

 following spring. The comparison in regard to size, however carefully 

 made, will no doubt fail to be absolutely exact, since comparison of 

 an individual shorter and more compact with one a little taller and 

 more slender, or of an individual having several small leaves with 

 one having fewer larger ones is difficult to standardise. Furthermore 

 slight inevitable inequalities in the environment will no doubt have 

 some effect. But despite some possible errors in arranging the young 

 plants in their true sequence due to these causes, the marked difference 

 between the ratios obtained from the groups indicated by the lower 

 and higher numbers respectively is sufficiently striking as shown in 

 the accompanying table. The proportion of doubles occuiring in the 

 whole number of fiimilies taken together is very near expectation, being 

 220 singles to 307 doubles, or about 58 per cent, of doubles, but the 

 proportion occurring among the most vigorous plants (indicated by the 

 lower numbers 1, 2, 3, etc.) is very considerably in excess of this ratio, 

 this excess being counterbalanced by a corresponding deficiency among 

 the groups comprising the less vigorous members of the different 

 families (indicated by the higher numbers)^. We may then conclude 



' This pieliminary sorting was carried out in order to avoid any invalidation of the 

 results from unknown inter-relations between degree of vigour and surface character 

 or doubleness. This precaution appearing unnecessary was not adopted in numbering the 

 next generation, where the whole lot of plants in a pot were numbered in one series 

 irrespective of the surface character. 



- At what point it is first possible to detect this difference of vigour I have not 

 attempted to determine. For such a determination a more stringent comparison in 

 regard to many points would be necessary. 



