256 
Miscellanea 
The " coefficient of fertility " is §a|g = -8438. 
The coefficient of fertility diflers according to the number of canals and to their regularity. 
Thus: 
Number 
of 
canals 
CoeflScient of Symmetry 
All individuals 
Irregular 
individuals 
3 
•75 
•555 
•782 
•69 
5 
■844 
■63 
6 
•814 
■69 
7 
•785 
■57 
It appears then that the less typical an individual the less its fertility, and irregular 
individuals are more sterile than those having some sort of symmetry. Thus the typical form 
and symmetry tend to be preserved. 
It v?ill be noted that this is a case of "reproductive selection," i.e. that there exists a 
correlation between fertility and the intensity of certain characters. 
C. B. DAVENPORT. 
II. Variation of the Egg of the Sparrow (Passer domesticus). 
I HAD recently occasion for a special purpose to measure 180 clutches, containing 687 eggs of 
the house-sparrow. They were collected from Essex, Hertfordshire, Buckinghamshire, Berkshire, 
Surrey, Cambridgeshire, and even as far north as Nottinghamshire and Yorkshire. They were 
thus a fairly representative English series. The bulk of them were collected early in the year, 
and were undoubtedly first-nesting eggs. Having calculated out the constants for my series, it 
may not be without interest to compare them with constants obtained from the measurements of 
Professor Hermon C. Bumpus on both the American and English races of Passer domesticus. 
Professor Bumpus in 1896 measured 868 American and 868 English eggs and published* 
diagrams giving the frequency in length and index — breadth/length — of these series. He did 
not publish, as far as I am aware, the means and standard deviations of his frequency distri- 
butions, but concluded from his diagrams that the egg of the American sparrow is shorter and 
rounder than that of the English sparrow and further that it is more variable in length and 
shape. The greater variability of length and shape was judged not by the standard deviation 
(S. D.), or the coefficient of variability (C. of V.), but by the extent of the ranges. This method 
is, I believe, open to criticism +, and it seemed desirable to work out the numerical constants 
of Professor Bumpus' series. The results I have found are given in the table below, absolute 
lengths being in mms. 
Sufficient data are not given for me to determine the S.D.'s of the breadths in Professor 
Bumpus' series, and as for the actual values of the breadths, I think, all we can say is that they 
must be very nearly equal and lie between 15'3 and 15-4. Now from these results we conclude : 
(i) That the American egg is not more but less variable than the English in length. 
* Biological Lectures delivered at the Marine Biological Laboratory of Wood's Holl. Summer 
Session, 1896. Boston, 1897. 
t The Chances of Death, Vol. i. Variation in Man and Woman, p. 275. 
