Miscellanea 
219 
I — III. Hero the inflorescences are necessarily weighted with the number of flowers which 
they produce. In all these cases the terminal flower seems to have been present. The number 
of lateral flowers, then, represents the real variable characteristic of the inflorescence, and has 
been taken as the first variable in preparing the correlation tables. The constants calculated 
are given in Table A. The weighting of the inflorescences must be borne in mind by anyone 
desiring to use these standard deviations or coefficients of variation for other purposes. 
The means and variabilities of the three collections show some differences which may be 
significant, but until further collections from a wider range of habitats are available, little can 
be profitably said concerning them. Like other kinds of biological work. Biometry cannot 
make great progress until enough series of data have been reduced to permit of extensive 
comparisons. 
The correlation coefficients are only about five to seven times their probable errors and 
so cannot be given too much significance*. Furthermore, two of the coefiicients are negative 
while the other is positive. In such cases as these too great significance must not be attached 
to the probable errors. The actual number of inflorescences is really very small in two of the 
series, and the tables appear to contain adequate data merely because the inflorescences are 
necessarily weighted with the number of flowers which they bear. In cases in which deviations 
from the typical condition are very few the collection of a slightly larger series may materially 
change the constants for a racet. The Kent series is the only adequately large one of the 
three, and the correlations derived from it may be considered significantly negative. The other 
two collections would be sufticieutly large to give fairly trustworthy results in characters with 
more normal variability. But where only 9 of the 163 inflorescences actually involved depart 
from the normal type — as is the case in the Essex series — it is dangerous to lay much stress 
upon the results from small samples. 
Considering all these difficulties I think we can draw no final conclusion from these data. 
But it appears from the results of this first examination of the relationship between the number 
of flowers per inflorescence and the number of divisions of the floral envelopes that the problem 
is worthy of detailed study by someone living in a region where considerable quantities of 
Adoxa may be obtained:];. 
Problem (b). The correlation between the number of corolla lobes of flowers of the same 
inflorescence. 
Teratologists have long known that when one organ of an individual is abnormal there 
is some probability that a second homologous organ of the same individual will be abnormal 
also. How great is this probability ? 
For the small deviations from the type of the race which have generally been called variations 
— or fluctuations by many modern writers — Pearson and others have determined quantitatively 
the degree of resemblance between the undifierentiated like organs of the individual for a 
considerable series of species. The interindividual correlations for a form like Ado.va will 
be interesting to compare with the coefficients obtained from other homotypic relationships. 
Adoxa produces ordinarily only a single inflorescence, and the interest of the comparison lies in 
the fact that all the horaotypes are so closely associated. A priori, 1 would have expected such 
organs to be more highly correlated than those more widely separated on the individual. Even 
* All the probable errors are calculated from the formulae commonly used, but it must be noted 
that for the number of flowers per inflorescence the percentage of deviations from the normal approaches 
the limits beyond which these formulae cannot be used with perfect confidence. 
t For another illustration see Report Missouri Botanical Garden, Vol. 20, 1909. 
X In this note I have not considered the number of lobes of the terminal flower because the data are 
too few. Tliis should be taken up when larger series of material are available. I note in passing that 
there is a slip in Whitehead's Table III, where the frequency of the 5-merous flowers should be given as 
2479 instead of 2494. 
28—2 
