128 
Miscellanea 
In the case of the late flowers, Dr Verschafielt's measure of correlation gives a value 
greater than unity; and Prof. MacLeod seems to have taken this as a sign of imperfect 
correlation : so that he describes the change in correlation as the reverse of that which 
actually occurs, and to speak of stamens and pistils as less perfectly correlated in late flowers 
than in early ones. 
Dr Verschafielt's method of attempting to measure correlation is so extensively used by 
botanists in his own country that it has seemed worth while to point out in some detail the 
divergence between his method and that based by Mr Galton on the theory of chance, and at 
the same time to indicate how it may lead to erroneous conclusions. 
Professor MacLeod's investigations on the difterences in mean, variability and correlation of 
the same plants during the flowering season i>rovide a most valuable lesson as to the possible 
danger of asserting that such differences are significant of local races. They may be due solely 
to the local environment, or to the period at which the individuals were collected ; the seasonal 
changes in these characters may be accelerated or delayed by local conditions. 
W. F. R. WELDON. 
II. Statoblasts of Pectinatella Magnifica. 
In an interesting paper on the variation of the statoblasts of this freshwater Bryozoa in 
The American Naturalist, Vol. xxxiv. p. 964, Professor Davenport has determined (a) the 
standard deviation, o-, of the number of hooks of the whole population of 635 statoblasts. He 
finds it to be 1'326. He has also found (6) the average standard deviation, 2, of eleven colonies 
of statoblasts. Each such colony is derived from a single statoblast. He finds the average 
value ri97. We thus have : 
o- ~ 1-326- 
Now a statoblast colony is one the ancestry of which has been selected back indefinitely. 
Hence if we accept the "Law of Ancestral Heredity"* we should expect that 2/o-='8944, a 
result in good accordance with the above. Thus heredity in this simple organism seems quite 
comparable with its value as found for man and the higher mammals, an exceedingly important 
result. We may hope that measurements like those of Professor Davenport will be carried out 
more extensively and on other species, for there is no more interesting problem than that 
involved in the diversity or constancy of heredity throughout living forms. 
K. P. 
* R. S. Proc. Vol. 62, p. 399, Mr Yule points out to me that a deduction of the fraternal correlation 
from this result — such as I have given in my paper " On the Principle of Homotyposis and its relation 
to Heredity" (Phil Trans, a. 1901, see Table XXXI.) — is open to questiou. 
