6 
G. H. Parker 
blood corpnscles. When the magnification was so arranged that each 
Square contained on an average about eight fibres the repeated enu- 
meration of any group of Squares gave nearly Constant results. Thus 
ten enumerations carried out on the same four Squares gave the 
following figures: 37, 36, 36, 36, 36, 35, 36, 35, 36, 36: average 
35.9. 
The number of ommatidia, or retinal elements, wa determined 
by counting the corneal facets. Since the facets were extremely 
uniform in size, proportional estimates could be made without se- 
riously impairing the final results. The method employed in this enu- 
meration consisted in comparing the area of a facet with the area 
of the whole corneal cuticula in terms of weight, and was carried 
out in the following way. From a number of small sheets of paper 
of equal size, those were selected that did not vary in weight from 
a given sheet assumed as a standard by more than a half per cent. 
A complete corneal cuticula was then carefully cleaned, cut into 
small squarish pieces, and the outlines of the pieces traced by means 
of a camera on the sheets of paper previously selected. The paper 
areas thus outlined were cut out and weighed, the total weight being 
taken to represent the total area of the corneal cuticula. In a similar 
way outlines of small parts of each piece of corneal cuticula were 
made, and the exact number of facets within each one of these was 
accurately counted. The total weight of these pieces could then be 
said to represent the number of facets which they collectively con- 
tained, and in this way the value of a facet in terms of weight 
was easily obtained. Having the whole corneal cuticula, as well as 
# a single facet, represented in weights, it was easy to estimate the 
total number of facets. 
Several precautions are necessary in using this method. First, 
the pieces of corneal cuticula, though assumed to be fiat; are in 
reality small parts of a nearly spherical surface, and thus must 
be somewhat greater in area than their outlines indicate. Further, 
it must be remembered that in drawing with a camera the image 
is somewhat distorted, its periphery being relatively larger than its 
centrai part. These sources of error partly correct each other; they 
can be largely eliminated, however, by judiciously selecting, in 
reference to the larger area, the position of the smaller one in which 
the facets are counted. If the smaller area Stretches from the centre 
to the periphery of the larger one ^nd has an appropriate form, both 
sources of error would be corrected; for both the unit of measure 
