1924] 
HARRIS—THE INFLORESCENCE OF MANFREDA VIRGINICA 447 
TABLE XIX 
PARTIAL CORRELATION BETWEEN FLOWERS PER INFLORESCENCE AND 
SEEDS PER LOCULE FOR CONSTANT NUMBER OF FRUITS AND BETWEEN 
FRUI 
WER 
PARTIAL CORRELATIONS WITH THE GROSS CORRELATIONS 
Partial Partial 
i dece и, 
i owers —r rults —7, 
сее and seeds LM and seeds PM 
frcs 
ме Highlands 1906 (I) | +.0829-+.0387} +.0508 | —.08752-.0836 | —.0446 
а рае А? 4-.1003--.0985 | +.0621 | —.10882-.0385 | —.0519 
г жада Barracks 1906 (П) -F.18852-.0411| +.0932 | — ii^ iei 0412| —.0977 
Jefferson Barracks 1907(VI) +.0565-&.0425 | +.0767 | —.1370+.0419 —.0105 
larger, as measured by the number of flowers which they produce, 
ripen slightly larger numbers of seeds per locule. 
The reader will have noted that if the physiological relationship 
between size of inflorescence and number of seeds per locule be of a 
positive kind, such that (other factors being eliminated) larger 
inflorescences tend to mature larger numbers of seeds in their 
locules, this will tend to screen the true physiological relationship 
between the number of fruits matured per inflorescence and the 
number of seeds ripened per locule. 
We therefore require the partial correlation between the number 
of capsules and number of seeds for constant number of flowers 
per inflorescence. The required formula is: 
Tes — Tycl fs 
CMT тут т 
The coefficients are given in table xix. These are negative 
throughout, and all have a higher negative value than that of the 
uneorrected correlation. 
Turning to the literature for comparable observations, we note 
that while data for size of inflorescence and number of seeds per 
fruit cannot be obtained in Cercis, we do have a series of ovaries 
(Harris, 719) in which the number of ovules was counted under 
the microscope. The correlations between the number of flowers 
per inflorescence and the number of ovules per ovary have been 
shown to be: For tree 1, r = — .007 + .023; for tree 2, r = 
+ .030 + .021; for tree 3,7 = + .134 + .024. These constants 
are so low that their significance is questionable. 
