218 
Miscellanea 
Ob\'iously this method of obtaining the totals of the clifFerence cohinins can be vised for the 
symmetrical tables as well, where only the diagonal cell and half of the remaining diagonal rows 
need be considered, thus doing away with the work required by the method of copying the 
frequencies from the correlation table into difi'erence columns as suggested by Pearson. 
Unfortunately these difference methods are limited in their applicability. When integral 
variates are under consideration, they give the same result as the product moment method 
when Sheppard's Correction is not applied in the calculation of the standard deviations. But 
difficulties arise when other than integral variates are considered. The divergence of the 
coefficients of correlation calculated by different methods is shown by Wright, Lee and Pearson*. 
But where the limits of their applicability are borne in mind, I think, these difference methods 
deserve more consideration in practical work than they have hitherto received. 
III. Note on Variation in Adoxa. 
By J. ARTHUR HARRIS, Ph.D., Cold Spring Harbor, L.I., U.S.A. 
In examining Whitehead's paper t' on the Moschatel, certain points came to my attention 
which seem to merit more detailed consideration than he has given them. What is really 
needed is a more extensive quantitative investigation than is possible ou the data hitherto 
published. Possibly the indication of certain points of interest not discussed by Whitehead 
may induce some one who has the opportunity of collecting fresh material to treat the problems 
in the detailed way which they deserve. 
Two questions occur to one examining Whitehead's second table : 
(a) Is there any relationship between the number of flowers which an inflorescence 
produces and the characteristics of these flowers ? 
{h) What is the similarity of the flowers of an inflorescence? 
Both of these, I take it, are of considerable interest to morphologists. 
Problem (a). The correlation between the number of flowers per inflorescence and the 
number of the divisions of the corolla. 
In our almost complete ignorance of factors influencing the number of parts of the flower, it 
seems important to investigate every possible interrelationship between the characteristics 
of the flower and the characteristics of the individual which produces it. Such investigations 
may not be expected to yield more than a small part of the information which we desire, 
but before experimental methods are extensively applied, it seems logical to determine whether 
there is any dependence of the number of floral parts developed, or any dependence of any 
characteristics of these floral parts, upon the vegetative organs of the individuals which produce 
them. 
In the present case only one characteristic other than those belonging to the flower itself 
is available for comparison, i.e. number of flowers per inflorescence. This character seems, 
however, a particularly interesting one for study. In the Moschatel there is one form of 
inflorescence which is conspicuously the modal one. This type is a very neat, compact, morpho- 
logical structure and any departure from it appears at once to be "abnormal." The flowers, 
too, have a normal form, from which a considerable number of deviations are to be found. 
Are abnormalities in the number of lobes of the corolla associated with abnormalities in the 
number of flowers borne on the inflorescence? The data for determining the coefficients of 
correlation are extracted from Whitehead's Table II and presented in the correlation tables 
* Biometrika, Vol. v. p. 410. 
+ Biometrika, Vol. ii. pp. 108—113, 1902. 
