232 
Psyche 
[Vol. 88 
surrendering the polymorphism and polyethism of Daceton (Wilson 
1971). 
Orectognathus versicolor , as the sole polymorphic intermediate 
dacetine, is of special interest for polyethism analysis. The species is 
also an easy one to study, its slow-moving habits and small colony 
size making possible the recording of nearly every behavioral act 
performed by each individual worker. The minor workers possess 
the same long, slender mandibles, with pointed apical teeth, that 
their congeners bear. Majors, however, have massive, relatively 
short mandibles, with apical teeth thick, blunt and recessed; their 
large occipital lobes contain disproportionately developed mandible 
adductor muscles (figs. 1 and 2). In mandible allometry, at least, this 
species may be the most exaggeratedly polymorphic of all dacetines. 
The division of labor by which such morphologically divergent 
forms are utilized, particularly since the major caste is a secondary 
development, may shed light on the advantages of specialized castes 
in the context of dacetine evolution. To what use are the singular 
majors put? Does the polyethism of O. versicolor in any way 
resemble that of Daceton , or is it entirely independent? Has the 
return to polymorphism been accompanied by a return to the 
polyphagy of Daceton , or is O. versicolor a collembolan specialist, 
as the rest of its genus is thought to be (Brown 1953)? An 
opportunity to address these questions in the laboratory arose when 
Bert Holldobler brought a live queenright colony of these ants from 
North Queensland, Australia; the results of observation of this 
colony are reported below. 
Materials and Methods 
The O. versicolor colony was settled in a glass test tube (2 cm in 
diameter), with water trapped at its end behind a tight cotton plug. 
The tube was placed in a plaster-floored clear plastic container (18 
cm by 12 cm by 6 cm), and a dissecting microscope was set over it on 
a moveable mount to permit viewing of ants both inside the nest 
tube and out on the container floor. A total of 45 hours of 
observation were made over a period of five weeks, during which 
7,891 separate behavioral acts were recorded. Estimation of the 
completeness of caste behavior repertories was made by fitting the 
data to a lognormal Poisson distribution, following the method of 
Fagen and Goldman (1977). The ants were offered various food 
