122 A First Study of Natural Selection in Clausilia laminata 
If the result indicated by the observations we have discussed were true, 
without any qualification, then we might expect the variability of the very last- 
formed portion of a growing shell to be greater than that of any portion which 
precedes it ; but the last line of Table IV. does not give the impression that this 
is the case. Of the young shells measured, 48 were so young that the last 
measurable part of their spire occurred above the point at which the columellar 
radius is 5 mm. long; but in 62 shells it was possible to measure a radius below 
this point : and the variability of these 62 radii appears, from the last line of 
Table IV., to be hardly, if at all, greater than that of the corresponding radii 
of adult shells. 
The evidence of this small group of measures does not seem to me sufficient to 
outweigh that of the rest, because on the one hand the probable error of the 
determination is relatively large and on the other hand, an explanation of the 
apparent contradiction between the two sets of evidence may very possibly be 
found in some property of the last whorl of a growing shell. There are two 
obvious differences between the last-formed whorl of such a shell, and those which 
precede it. The last whorl is quite sensibly thinner than the others, because the 
inner, "nacreous" layer is not yet secreted in such quantity as in the higher 
parts of the shell ; and the last whorl has nothing attached to its lower surface. 
Now, in the formation of a further whorl, the lower surface of the last-formed 
whorl will have a new layer of shell closely appressed against it : and my friend 
Professor Pearson has suggested to me that the strain on the shell, during this 
process, may cause a sensible increase in its variability. I think it very likely 
that some explanation of this kind will be found after further study of the last- 
formed whorl, which I hope shortly to undertake. In the meantime I publish 
the observations as they stand. They seem to me to show clearly that the 
mean character of the spiral radius is stable from generation to generation in 
Gremsmiihlen, and that it is not being changed by selection during the growth 
of a young generation : and they seem to me to show further that the variability 
of young shells is diminished in every generation by a process of " periodic 
selection." Such selection is, of course, " indirect " : that is to say, the life or 
death of the individual is determined in each case by the value of a (probably 
large) number of correlated characters, of which the length of the peripheral radius 
is only one. 
I am aware that to many people it will seem absurd that the life or death 
of an individual should be supposed to depend upon differences of structural 
character so slight as those just described. The species of Glamilia show, however, 
if possible more clearly than other species, the way in which small and apparently 
unimportant differences of structure are associated with the difference between 
the survival and the total extinction of a race in a particular locality: and if 
so great a difference in death-rate as this is correlated with the apparently 
trifling differences which separate species, the hypothesis suggested by measure- 
ment of the spiral does not seem to me ridiculous. 
