DIVISION OF LABOUR. 23 
state of imperfection, that some neuters shall exhibit 
characters at variance with those of the common kind.’ 
This, indeed, credits them with a very remarkable 
instinct, and yet I see no more probable mode of ac- 
counting for the facts. Moreover, the exact mode ky 
which the differences are produced is still entirely 
unknown. 
M. Forel, in his excellent work on ants, has pointed 
out that very young ants devote themselves at first to 
the care of the larvee and pupz, and that they take no 
share in the defence of the nest or other out-of-door 
work until they are some days old. This seems natural, 
because at first their skin is comparatively soft; and it 
would clearly be undesirable for them to undertake rough 
work or run into danger until their armour had had 
time to harden. There are, however, reasons for think- 
ing that the division of labour is carried still further. I 
do not allude merely to those cases in which there are 
completely different kinds of workers, but even to the 
ordinary workers. In L. flavus, for instance, it seems 
probable that the duties of the small workers are 
somewhat different from those of the large ones, 
though no such division of labour has yet been detected. 
I shall have to record some further observations point- 
ing in the same direction. 
The nests of ants may be divided into several 
classes. Some species, such as our common Horse ant 
(Formica, rufa), collect large quantities of materials, 
such as bits of stick, fir leaves, &c., which they heap 
