W. F. R. Weldon 
iir 
can undergo are those due to erosion by weather, or to such absorption of the 
supporting inner layer as may cause some portions of the shell to be broken away 
from the rest. Neither of these disturbing influences affects Glaasilia laminata in 
Gremsmtihlen to any sensible extent : and we may therefore conclude that the 
upper whorls of an adult shell represent the condition of the young shell, from 
which this adult was formed by the addition of new material, in a practically 
unaltered form. This conclusion may not justify us in comparing the very last, 
newly formed portion of a growing shell with the corresponding part of an adult, 
but except at this point, there is no doubt of its validity. 
Such an unaltered record of the young condition of the shell, borne by every 
adult examined, is of great value as a means of estimating the incidence of natural 
selection : for by measuring the mean and the standard deviation of radii of the 
spiral in the upper part of a series of adult shells, we obtain a measure of the mean 
length and the variability of the radii exhibited by these shells when they were 
still young ; and in this way we can estimate the condition of the spiral in a series 
of young individuals, of which we can say with certainty that every one was 
capable of attaining the adult condition. In a sufficiently large sample of young 
shells, we are sure that we have a number of individuals which will die before they 
attain maturity ; and therefore a comparison between the mean character and 
variability of these young shells, and the mean character and variability of the 
upper whorls of adult shells, enables us to determine whether tliere is a correlation 
between the character of the spiral, and the death-rate, or not. For if the death- 
rate during growth affects individuals with all sorts of spirals to the same extent, 
then there will be no change in the mean character and variability of the spiral 
after the time of its first formation, whether we examine it in half-grown indi- 
viduals, soon after it has been formed, or in adults. If, on the other hand, 
individuals with different kinds of spirals die at different rates during growth, then 
either the mean length or the standard deviation of radii, or both, will change : 
so that we shall certainly not get the same value for both these characters in young 
and in adults, if the character of the spiral is being acted on by any process of 
selective elimination. 
In order to compare the radii of young shells with those of the upper whorls of 
adults, one hundred half-grown individuals from Gremsmtihlen were ground and 
measured, the measures being treated in the manner already described. The 
smallest of these individuals was 5"3 mm. long, the majority being from 6 to 7 mm. 
long. The length of the largest was 8 8 mm., but only three were more than 
8 mm., the greater number being less than 7 mm. long, the mean length of an 
adult shell being about 15 mm. 
The mean radius-length, with its mean angular distance from the plane of 
the standard columellar radius, is given for every group in Table II. The probable 
error of the mean radius-length has been determined for the six lowest groups, 
which contain the last- formed radii. These groups are those for which the 
probable error is given in adults also (see Table I.). 
