SOMATIC CHARACTERS AND FERTILITY 
407 
coefficients 19 are over 2.5 times their probable error, as compared 
with the 5 possibly significant values noted for the negative. Further- 
more, the mean of the ratios of the 10 negative coefficients to their 
probable error is only 2.49 as compared with a mean ratio of 9.74 for 
the positive coefficients. 
Thus the evidence seems to justify fully the assertion that there is a 
slight physiological relationship between the number of pods which a 
plant produces and the number of seeds wJiich it is capable of maturing 
in these pods, and that this relationship is quantitatively independent 
of the more strictly morphogenetic factors linking together number of 
pods and number of ovules per pod. 
Further than this the analysis cannot safely be pushed on the basis - 
of available data. 
III. Comparison of the Foregoing Constants with other 
Correlations for Somatic Characters and fertility 
Since many additional series of data bearing on these problems are 
in hand and will eventually be published, I shall not compare in detail 
the results set forth here with others but I shall indicate merely the 
order of correlations which have been found. 
For length of flowering stalk and number of flowers in the umbel- 
like inflorescence of two liliaceous plants the correlations are roundly^ 
For Nothoscordium striatum, r = .500, 
For Allium stellatum, r = .620. 
For the relationship between the length of the flowering stalk 
(pedicel) and number of ovules and seeds per fruit the only available 
data are^ 
For Sanguinaria, rio = .323 dh .019, 
For Sanguinaria, ru = .363 zfc .019. 
Here ru > rio, but ru — rio = .040 i .027 only. Thus the difference 
is hardly significant with regard to its probable error; there is only a 
single series of material; further collections might show that here, as 
in garden beans, the correlation for degree of vegetative development 
and number of seeds is on the average lower than that for vegetative 
development and number of ovules per pod. The partial correlation 
in the case of Sanguinaria would of course have a significantly positive 
value. 
^Ann. Rept. Mo. Bot. Gard. 20: 105-115. 1909. 
Biometrika 7: 316. 1910. 
