94 EXPERIMENT STATION. [Jan. 



correlation. The natnre of the gronps in tlio first part of the 

 table lias already been explained. The remaining varieties have 

 been grown from commercial seeds bonght in the open market. 

 The table shows that the correlation coefficient is closely related 

 to the mean viiie length in different varieties. The rule that is 

 generally, thongh not invariably, followed is that the longer 

 the vine the lower the correlation l)ctween vine length and pods 

 per vine. This is reversed in the case of the different years 

 shown in the first part of the table. This is dne to the fact 

 that the vines branched more freely in 1911, owing presnmably 

 to weather conditions. Vine length is taken as the length of 

 the main stem, and when there are one or more branches bearing 

 pods it is obvious that the correlation is lessened. 



To this same factor is due in part, but not wholly, the smaller 

 correlation in the groups having longer vines. 



One or two pods may l)c borne at each node of the plant, but 

 at many nodes no pod is produced. Whether or not a pod is 

 produced depends presumably on environmental conditions at 

 the time when the early development of the node has reached a 

 certain stage. The period of growth and node production is 

 much longer with long-vined plants, and the plant is subjected 

 to a greater variation of conditions. As successive nodes de- 

 velop some will experience favorable and some unfavorable 

 conditions, and this will not be in very close correlation with 

 vine growth. In some varieties and in some seasons the produc- 

 tion of doubled podded peduncles is more common, and this 

 operates to disturb the correlation. This question will l)e fur- 

 ther dealt with in connection with the later discussion of varia- 

 tion in productiveness. 



VAKIATIOISr. 



We may now proceed to a discussion of the amount and na- 

 ture of the variation that has been in evidence in the different 

 groups from season to season. 



In Table X. are given the standard deviations and coefficients 

 of variability that are available, and the means ai'e inserted for 

 convenience of innncMliate com]>arison, though most of them 

 have already been given in another connection. 



