J. Arthur Harris 
209 
the trees with respect to seed production, but the bulk of the interdependence 
for seeds seems to be merely the resultant of r^'o' and r^s*- 
Relative net individuality of the trees with respect to capacity for seed pro- 
duction is highest in the Ohio series. This is due to the fact that both /o,,,,, and 
r^s are lower for this series than for any other. 
The frequency distribution of length of pod measured to the nearest millimetre 
in 50 pods each from 60 trees is shown in Table XII. The summed length for 
the individual trees, i.e. mean length x 50, is given in Table XIII. By the direct 
intra-class formulae I find 
fjj,, = -4784 ± -0095, 
a homotypic value distinctly higher than those found for the fertility characters. 
V. Discussion of Results. 
The only homotypic constants available for fertility characters in plants are 
those furnished for several species of-Leguminosae by Pearson and his associates f. 
They give the direct correlations as shown in Table C. 
TABLE C. 
Homolypic Correlations for Fertility Characters. 
Species of Legume 
Cytisus Scoparius 
Lotus Corniculatus 
j» >> 
Lathyrus odoratus 
Lathyrus Sylvestris 
Vicia Faba 
Vicia Hirsuta 
Ovules 
Abortive 
Ovules 
Ripe Seeds 
•4155 
•2354 
•1884 
•2182 
•2679 
•0830 
•3658 
•1759 
•2091 
■1695 
•1376 
■2184 
•1877 
•1724 
•1493 
•1243 
•2315 
•1827 
■2077 
In addition to these values I have found in a short series of only 12 trees 
of the arborescent legume Robinia Pseud-acacia worked out as an illustration of 
method J the values r^^g^ = -452, r^^g^ = -449, rg^,., = -383. For a short series 
(23 plants) of Cytisus Scoparius^ I have found r^^g.^ = •198. 
The average homotypic values for ovules in Cercis are distinctly higher than 
the comparable values for other species hitherto adequately investigated. The 
* In Sanr/uinaria (Biometrika, Vol. vn. p. 328) the correlation for the number of seeds on the two 
placentae of the same fruit seems to be chiefly due to physiological factors. In this case an organic 
correlation is superimposed upon a homotypic. The point may profitably be discussed comparatively 
when other data now in hand are completely analyzed. 
t K. Pearson and others, Phil. Trans. A, Vol. cxcvii. pp. 364-379, 1901. 
% BiomelriJca, Vol. ix. pp. 456-i58, 463, 1913. 
§ Amcr. Nat. Vol. xlv. pp. 566-571, 1911. Possibly this material is of closely selected ancestry. 
Biometrika xi 14 
