308 Morphology of the Fruit of Sanguinaria Canadensis 
To reduce the distributions to somewhat more workable limits the fruits have 
been combined in groups of 3 mm. range. These have proved very satisfactory 
working units. Constants calculated from them are essentially the same as those 
calculated from the ungrouped distributions, as I have assured myself by actually 
carrying out the arithmetic. 
3 — 11. Meristic Characters of the Fruit. The counting of the number of 
ovules formed, the number of seeds developing, and the number of abortive ovules 
is not at all difficult in this species. The fruits were of slightly different ages in 
the two collections, but in both cases they were old enough for the determination 
of the number of seeds developing to be entirely trustworthy. 
In recording these characters the numbers on each placenta were noted 
separately. In the following pages the placentae are referred to as " first " and 
" second." The two are not differentiated at all, the " first " placenta being 
merely the first one counted when the fruit was opened. The constants for 
the distribution of ovules, seeds and aborted ovules should be the same for the 
two placentae within the limits of the probable error of random sampling. In 
investigating many relationships the tables may be made symmetrical, that is to 
say each placenta may be used once as the "first" and once as the "second" 
number of a pair. The advantage of using these symmetrical tables has been 
several times pointed out in biometric literature*. 
I have found it rather an advantage to work with ovules, seeds and aborted 
ovules. Naturally, the seeds developing and the aborted ovules per fruit equal 
the number of ovules formed, and the tables may easily be verified if all three 
characters be used. Characters 9, 10 and 11 are obtained by adding 3 and 5, 
4 and 6, 7 and 8. 
(c) Method of Reduction of Data. 
Sheppard's correction was used throughout in the calculation of the standard 
deviations. 
In calculating the probable errors where more than one organ from each 
individual is measured, there is always some question as to what value of N 
shall be used. I have taken N as the actual number of flowering stalks in all 
cases in which one of the characters involved in the constant was measured only 
a single time for each inflorescence. For instance in correlating between length 
of fruit and number of seeds per placenta, or between the fertility characters of 
the two placentae of the same fruit, N was used as 1000 and 400 in the two 
* For instance, by Pearson and his assistants in their memoir on homotyposis in the vegetable 
kingdom [Phil. Trans. Boy. Soc. Lond. A, Vol. cxcvii. pp. 285 — 379) ; by Pearl on Paramecium 
(Biometrika, Vol. v. pp. 249 — 251, 1907) ; by Pearson and Barrington in their memoir on the inheritance 
in vision {Francis Gallon Lab. for Nat. Eugenics, 5, 1909) ; by Harris (Biometrika, Vol. vii. pp. 214 — 
218, 1909). 
