98 EXPERIMENT STATION. [Tan. 



Considering first the figures for vino length, we find that in 

 1908 the stan(hird deviation and coefficient of variability were 

 much lower than in any of the following three years. This is 

 due to two factors, the more potent of which, doubtless, was the 

 soil of the plot on which the plants were grown; this was grav- 

 elly and the ])lants suffered severely from drought. The other 

 was the small number of plants grown, the total of 1908 being 

 227, while in subsequent years the total of the same groups 

 has been more than 1,000. During the years 1909-11 there 

 seems not to have been in Excelsior I., or any of its sul)-grou])s, 

 any constant differences in variation that cannot be ascribed to 

 seasonal inlluonces. In Excelsior II. both constants are notably 

 low in 1911. This may be due to the fact that they were 

 planted later this year than previously and encountered the 

 unusually hot weather of July, 1911, at an earlier stage of 

 development than either the other lots of Excelsior, or this lot 

 in earlier years had encountered the less severe midsummer heat 

 of those years. A comparison of the two strains of Excelsior I. 

 shows that A, D, E, G, K has had uniformly greater standard 

 deviation than B, E, 11, J, luit this has not been in jiroportiou 

 to the higher mean, so the coefficient of variability is less in 

 the longer vined strain. This same tendency is seen in the dis- 

 tinct varieties, although it is not invariably the case. 



We may ask if the variation within the lines of the two 

 strains of Excelsior I. give evidence of individuality of these 

 several lines ? Is any line constantly more or less variable than 

 the others of the same strain ? With the possible exception of 

 line I), which has a standard deviation uniformly larger or at 

 least as large as its fellows, there seems to be no evidence of such 

 a condition of affairs. It appears that the diiferences in the 

 variability within the different lines is mostly, if not entirely, 

 environmental and due chiefly to varying soil conditions. 



We may now turn our attention to the figures for the nund)er 

 of pods per vine. We see first of all that the coefficient of va- 

 riability is nearly twice as large as that for vine length, and in 

 many cases the difference is even greater than this. In gen- 

 eral, a high variability in vine length is accompanied by a high 

 variability of pods per vine and vice versa, as would be expected 

 from the strong correlation already shown to exist between these 



