158 
Psyche 
[March 
Standard methods and tests (Hays 1963) were used for computa- 
tion and assessment of significance of regression slopes. Lewis (i960) 
presents techniques for the analysis of intersample differences between 
regression slopes; we used the following statistic to test the signifi- 
cance of intercaste regression slope differences. 
Let k(p = estimated regression slope for i*, h caste. 
<7(i> = sample variance in i th caste of In (tergite width) 
given In (alitrunk length). 
o-i(i) = sample standard deviation of In (alitrunk length) 
in i th caste. 
ni = number of individuals of i th caste sampled. 
i = 1 for workers and 2 for queens. 
Under standard assumptions of regression analysis (normality, 
homeoscedasticity, interindividual independence of measurements) 
the statistic 
k(i)q , Ki)\Avi ~ k(2)<ri (2) V ft 2 
V2 
/n <7(i) + n 2 o- ( 2 ) 
\ ru -f n 2 - 4 
has a t-distribution with n t + n 2 — 4 degrees of freedom. 
Results 
Coloration. Both castes are yellow with brown or black dark 
markings ( Fig. 1 ) . The yellow of workers, however, is a bright 
lemon yellow, while that of queens contains much more brown. The 
dark markings of workers are mostly black, while in queens they are 
dark brown, especially on the gaster. Thus, the dark and light colora- 
tion of workers is much more contrasting than that of queens. 
On average the dark markings are slightly more extensive on the 
head and thorax of workers than of queens. On the other hand, the 
dark bands on the gaster are proportionately wider in queens than 
in workers. The dark band of the second abdominal tergite in 
workers has a pointed median posterior extension that is lacking in 
queens. 
Body size. Results of the two measurements made are summarized 
in Table II. There is no overlap in size (length of alitrunk) of 
queens and workers in any of the colonies; thus the two castes are 
