1924] 
HARRIS—THE INFLORESCENCE OF MNAFREDA VIRGINICA 455 
TABLE XXII 
CORRELATION BETWEEN NUMBER OF SEEDS PER LOCULE IN THE SAME 
FRUIT AND REGRESSION EQUATIONS SHOWING THE RELATION 
BETWEEN THE NUMBER OF SEEDS IN THE LOCULE OF 
Series Weighted | Correlation Regression 
N equation 
Jefferson Barracks 1906 6720 .7901 + .0044 l = 2.1619 + .7901 I 
Meramec Highlands 1906 12102 |.7119 + .0043 1, = 3.9669 + .7119 lı 
Jefferson Barracks 1907 8274 | .7408 + .0047 l, = 2.8374 + .7408 1, 
Meramec Highlands 1907 10374 .7383 + .0042 1, = 3.3479 + .7883 lı 
have a value of r = .80 to r = .84. That the correlation for 
seed production is not due solely to the high correlation of the 
numbers of ovules on the 2 placentae is shown by the fact that 
the partial correlations between the numbers of seeds on the 2 
placentae for constant numbers of ovules have a material value. 
In Hibiscus (Harris, ’13) it has been shown that for 1000 fruits 
examined at the Missouri Botanical Garden in the fall of 1905 the 
intra-ovarial correlations were 
For ovules per locule, r = .3843 + .0081 
For seeds per locule, 7 = .5557 + .0066 
Excess for seeds .1714 + .0104 
Here again the results indicate distinct physiological factors 
influencing the capacities of the several locules of the fruit for seed 
production in such a way as to bring about a similarity between 
them. 
In Crinum longifolium a correlation of r = .676 + .008 has 
been demonstrated (Harris, 712 a) between the weight of the 
seeds from the same fruit. 
All these results agree in indicating that there are morpho- 
genetic or physiological factors tending to bring about a similarity 
in the seed production and in the seed weight of the locules of 
the fruit. 
IV. REGCAPITULATION AND DISCUSSION 
This paper has had for its purpose the consideration of various 
problems of fertility in Manfreda virginica (А gave virginica). The 
conclusions are based on the statistical analysis of extensive series 
