J. A. Harris 
491 
Allowing full weight to the fact that the frequencies in series A have their 
probable errors, I think we can maintain that series C and series A are differ- 
entiated from each other with respect to the number of '' odd " and " even " locules 
in their ovaries. 
Problem 5. This problem cannot be discussed, since no data were collected 
for other than 3-merous fruits. 
VI. Interrelationship of Characters Considered. 
For convenience and clearness our characters have been treated in the fore- 
going discussion of problems as though they were quite independent. 
Analysis can be carried somewhat further. If a character is quite uncorrelated 
with others, we shall be confident in regarding any selective elimination which we 
find associated with this character as arising through some unfitness for continued 
development associated with it. If, on the contrary, it appears that a character is 
correlated with some other, we cannot know without further evidence which of a 
pair is potent in producing a selective elimination. 
These points will best be made clear by an examination of the three relation- 
ships which seem worth consideration for our material. 
1. The CorrelatAon betiueen the Goeficient of Asymmetry and the Number of 
" odd " Locules per Ovary. 
The fact that both radial asymmetry as measured by the deviation of the 
number of ovules per locule from their mean and the composition of the ovary 
with regard to the " even " and " odd " nature of the individual locules seem to be 
of significance in determining whether it shall develop into a mature fruit, does 
not necessarily prove that with both of these characteristics there is associated 
some functional unfitness for development. 
There is necessarily some correlation between the two characteristics in any 
fruit. If the three locules have all the same number of ovules and they be even, 
say 8 — 8 — 8, the asymmetry will be "OOOO. If it be "2 even, 1 odd," say 8 — 8 — 7 or 
9 — 8 — 8, the asymmetry will be "4714. Thus a fruit composed of "odd" and "even" 
must necessarily be somewhat asymmetrical while one of "all even " or "all odd" 
may be perfectly asymmetrical. If an " all even " or " all odd " fruit is radially 
asymmetrical with respect to number of ovules per locule, it may be more asym- 
metrical than a fruit of locules of both types, since there must be a difference of 
two ovules between two different " even " locules, while an " even " and an ' odd " 
may differ by only a single ovule. For instance, 8 — 8 — 6 is more asymmetrical 
than 8—8—7. 
