SEED WEIGHT AND GERMINATION 



273 



The impression convej^ed by the graph or by a classification 

 and study of the constants is two-fold. First, that the correla- 

 tion between seed weight and time required for germination is 

 very low indeed — of such an order that many experiments involv- 

 ing large numbers of seeds are required to obtain a trustworthy 

 measure of its intensity. Second, that when the magnitude as 

 well as the numbers of individual constants is taken into account, 

 the evidence for a positive correlation between weight of seed 

 and time required for germination is much more convincing. 



Consider now the statistical tnistworthiness of the individual 

 constants as indicated by their probable errors. The ratio 

 r jE^ is shown in diagram 2. Appealing to the actual numerical 

 values for more precise information, it appears that of the 14 

 ratios for negative constants only 6 exceed 2.5, while only 1 is 

 over 4.0. Of the 36 ratios for positive correlations, on the other 

 hand, 23 are over 2.0, 19 are over 2.5, 12 are over 3.0, 8 are over 

 4.0 and 3 are from 7.0 to 33.0. The mean ratio for the negative 

 cases is only 2.15 while for the positive correlations it is 3.76. 

 Thus, as far as the statistical significance of the individual con- 

 stants is concerned, there can be no doubt that those indicating 

 a positive correlation are much more reliable than those which 

 indicate a negative. 



I have also classified the fifty experiments according to the 

 six distinct varieties of beans used. The number of experiments 

 for the individual strains, ranging from 1 to 19, is not large 

 enough to make it worth while to consider them in detail. In 

 all six cases, however, the correlations are either exclusively or pre- 

 ponderantly positive in sign, and the general average is positive. 

 Apparently, some varieties show higher correlations for the two 

 variables under consideration than others do, but the final answer 

 to this question depends upon the results of further studies. 



III. DISCUSSION OF RESULTS 



It has been shown with a high degree of probability that in 

 Phaseolus vulgaris there is a positive correlation between the 

 weight of the seed and the time required for its germination — 

 in other words, that as the weight of the seed increases from the 



