THE ASSOCIATION OF CHARACTERS 265 



and therefore more slowly and more irregularly germinating 

 kernels. 



As we have already seen, correlations may be divided 

 into those, which are easily traceable to a common cause, 

 and those in which the causal relation of the single phenom- 

 ena remains obscure. In the adduced example of the seeds 

 this relation is evident, at least to some extent, and therefore 

 I will now choose a case in which there is no obvious reason 

 for the connected qualities to be so connected. Our appre- 

 ciation of the correlation can in such cases only be based 

 on direct observation. This however, requires a most 

 minute study, on one side of seemingly insignificant botani- 

 cal characters, and on the other of the industrial qualities 

 of the plant. 



Among the grains the panicles of the oats give the best 

 evidence. Their shape and mode of branching are variable 

 to a high degree. The branches are combined into whorls. 

 The number and relative development of these branches 

 proves to be constant for each pure type. The apparent 

 high variabihty is to be reduced to a large group of single 

 forms, which when considered for themselves are narrowly 

 circumscribed and do not transgress their differentiating 

 limits. Some panicles have their largest branches longer 

 than the main axis, in others they are shorter. Those of 

 the first group are ordinarily stiff and dense, those of the 

 second being more flaccid and more widely spread. The 

 biggest kernels may be found in the lower spikelets, or they 

 may prefer the middle and upper ones, and experience has 

 shown that only such sorts arc commendable which have 

 their biggest seeds above the middle. 



Most reliable marks are given by the color of the seeds. 

 The sHghtest differences in tint are wholly constant and 

 indicative of corresponding practical qualities of the sorts. 

 Yellow and yellowish, straw and reddish-yellow, whitish and 



