1912] HARRIS—STAPHYLEA 207 
seems unreasonable to ask a publisher to print such a series. I 
retain my original tables, which are available to anyone desiring to 
consult them, and have deposited a duplicate set in the Library of 
the Missouri Botanical Garden. ae 
TABLE I 
CORRELATIONS FOR FRUIT LENGTH AND FERTILITY; 1906 SERIES 
B Length 5, Length f, Length 
ucbes sone ‘ovules per | ’/Er eh ee per t/Er ie ovules failing |. /Er 
locule locule per locule 
a veep aa — .246 .035 7.02 . 376+ .032 11.75 | —.399+.031 | 12.87 
1 Saar es - 201 .034 | 5.91 - 347 .031 11.18 | —.003* .035 2.66 
saa eae 300% .035 | 8.56 | .537%.027 | 19.87 | —.258%.036 | 7.17 
RA ee. 103.039 | 2.63 | .328.035 9.36 | —.185+.038 | 4.87 
1 ey 124.037 | 3.34] .247.035 7.05 | —.042.038 | 1. 
es ae 180+ .038 | 4.74] .347#.034 | 10.19 | —.083+.039 | 2.13 
Cee | 198.037 | 5.3§ | .311=.035 8.87 | —.o50+.038 | 1.30 
Sec ewiecs. 014+ .039 30 .343% 034 | 10.07 | —.264*.036 | 7.30 
ate eae re 068.039 | 1.75 320+ .035 9.14 129* .038 | 3.32 
ee ee 117.037. | 3.15 533.027 | 19.75 325.033 | 9-84 
Ae Taw oe 026+ .039 67 439 .031 14.15 | —.382 .034 | 11.22 
Be, 333.034 | 9.78 343.034 | 10.10 166+ .037 | 4.48 
Zo SA 0960 .039 | 2 315+ .035 8.98 | —.153*.039 | 3-91 
te 244.037 | 6.58 418 .032 | 13.05 178+ .038 | 4.67 
es hie 175.038 | 4.59 475.030 | 15.83 | —.300+.035 | 8.56 
ee 167+ .038 | 4.40 | .450%.031 | 14.70 | —.225+.037 | 6.08 
Re oe. 247.037 | 6.68 421+ .032 | 13.15 | —-117%.038 | 3.67 
ee 082+ .038 2.14 3604+ .033 11.02 | —.227%.036 | 6.31 
Me hy 093.039 | 2.37 415*.032 | 12.95 | —.270™.93 7-59 
ee es 043.039 | 1.08 402.033 | 12.18 | —.297#.036 | 8.25 
The physical constants (means, standard deviations, and coeffi- 
cients of variation) for the characters dealt with need not detain 
us here. The correlations for the 1906 series are given in table I, 
and for the 1907 series in table II. Assuming, as is generally done, 
that a correlation coefficient of 2.5 times its probable error may be 
regarded significant, we may examine the results for the 1906 series. 
In one case 7, is significantly negative; the remaining 19 con- 
stants are all positive, and 12 are greater than 2.5 &,. There can 
be no doubt of the significantly positive correlation between length 
and number of seeds per locule. Comparing 7, and 7, by means of 
6In calculating the probable errors for these individuals, NV was taken as the 
~ number of locules, not as the actual number of fruits; perhaps the latter course would 
have been better. There is always some question as to what should be done in cases 
of this kind. See Pearson, Phil. Trans. Roy. Soc. A 197: 295, footnote; Harris, 
Biometrika 7: 308-309. 1910. 
