24 VARIATION, DISTRIBUTION, AND EVOLUTION OF THE GENUS PARTULA. 
Despite the small number of areas under investigation, colonial variation is well 
displayed. As the Saipan material comprises the varieties of but a single species, to 
be dealt with separately in Chapter VII, the description of the areas of collection will 
be deferred to the later juncture. 
METHODS OF TREATMENT AND ANALYSIS. 
As the volume on the Tahitian species comprises a full description of the requi- 
site routine methods, only a brief digest is given here for the sake of unity and 
completeness. 
The animals were preserved in formalin, varying in strength from 2 to 5 per 
cent. They were first expanded by immersion in fresh water for a period of several 
hours, crowded closely in a closed glass jar; this treatment expands the bodies and 
renders them insensible. After they have been immersed in the preservative for 
several days, the greater part of the fluid may be poured off, thus lightening the 
jars for transportation. 
Brought to the laboratory table, the material of a given locality is assorted 
into its component species, if more than one exists, and into the subordinate 
varieties. The qualitative and quantitative characters of the shells are then re- 
corded, together with the data relating to the embryonic contents if any. The 
measured characters of the shells are (1) greatest length, including the lip, (2) 
greatest transverse dimension, also including the lip, (3) length of aperture, outside 
measure, and (4) width of aperture, outside measure. From these figures, the 
relative characters are determined, namely, (5) proportions of the whole shell, (6) 
proportions of the aperture, and (7) proportion of aperture length to the total 
shell length. Where a columellar tooth was present, its degree of development was 
noted according to an arbitrary scale. The colors of the outer surface of the shell, 
of the interior surface, of the apex, and lip were also precisely recorded. 
The statistical methods are those of standard biometry. The average value of 
a character and the standard deviation are the essential figures for the collective 
description of a group of individuals. In comparing two associations on the basis 
of their respective constants, the probable errors must be taken into full account. 
A given difference must be at least three times its probable error in order to be fully 
indicative; if it is between two and three times its error, it is only probably signifi- 
cant, and if it is less than twice its error it is only possibly an indication of a real 
difference. 
It remains to state that the author is personally responsible for the original 
measurements and for their mathematical reduction; therefore the personal 
coefficient is as uniform as it can be. 
