10 PROCEEDINGS OF THE NATIONAL MUSEUM VOL. 120 
length of cornea from anteroventral view (looking at cornea ‘‘end 
on’’), and breadth of cornea in similar view (see figs. 3B, C). Choice 
of the portions of the eye to be measured was conditioned partly by 
ease of precise recognition on successive occasions. Thus minimal 
corneal breadth was measured because recognition is aided by slight 
concave curvatures in the centers of the corneas, by the presence of 
larger facets in this area, and by transverse pigment lines in pigmented 
specimens. In all measurements there are subjective errors, and the 
accuracy of results decreases, possibly to +0.2 mm. From these 
measurements L/B cornea and L eye/B cornea were computed as the 
most accurate and meaningful of the possible ratios. 
Histograms of deviations from means give normal frequency distri- 
butions in each case. Values of standard deviation/mean (table 1) 
indicate that for L/B cornea data only the American material varies 
more than the Australasian. For both L/B cornea and L eye/B cornea 
data American specimens give higher values than Australasian, with 
the differences significant at values of p<0.001. 
Inspection of data indicated possible sexual dimorphism and possible 
size effects. Relevant data on sexual differences are shown below. 
L/B cornea 
no. mean weighted mean 
Australasia 382010! 1.2028 1.2013 
PRONG. 1.1991 
American 250'o' 1.3092 1.3090 
162 2 1.3088 
L eye/B cornea 
no. mean weighted mean 
Australasia 3ldéd UPL 1.3230 
232 2 1.3143 
American 2530'S 1.4772 1.4679 
U7 ere 1.4512 
For L/B cornea data, differences between the means of males and 
females are not significant either upon Australian material (p ca. 0.6) 
or American (p>0.9 level). For L eye/B cornea Australian data 
significance is at the 0.1 level but American data are not significantly 
different (p>0.9). 
Size effects were investigated only on L/B cornea data. The 
correlation coefficient between carapace length and L/B cornea for 
Australasian specimens is 0.296 (p ca. 0.02) and for American 0.476 
(p<0.001). Regression coefficients are 0.00145 and 0.00562, respec- 
tively. 
Summarizing these analyses, the American and Australasian 
specimens clearly belong to different populations as indicated by the 
