a few days, dying between the 16th and 20th. These two ants, therefore, must have been five years old at least. It is certainly curious that they should, after living so long, have died within ten days of one another. There was nothing, as far as I could see, in the state of the nest or the weather to account for this, and they were well supplied with food; yet I hardly venture to suggest that the survivor pined away for the loss of her companion.

Some workers of F. cinerea lived in one of my nests from November 1875 to July 1881.

In a nest of F. fusca, which I brought in on June 6, 1875, and in one of Lasius niger brought in on July 25, 1875, there were no queens; and, as already mentioned, no workers have been produced. Those now living (December 1881) are therefore the original ones, and they must be more than six years old.

The duration of life in ants is therefore much greater than has been hitherto supposed.

Though I lose many ants from accidents, especially in summer, in winter there are very few deaths.

I have given the following figure (fig. 2), which represents a typical nest belonging to Lasius niger, because it is a good instance of the mode in which my ants excavated chambers and galleries for themselves, and seems to show some ideas of strategy. The nest is, as usual, between two plates of glass, the outer border is a framework of wood, and the shaded part

Ground-plan of a typical nest of Lasius niger, reduced. a, narrow doorway; b, hall; c, vestibule; d, main chamber; e, inner sanctum; f,f,f,f, narrow entrance passages to sanctum; g, g, spe cial pillars.

represents garden mould, which the ants have themselves excavated, as shown in the figure. For the small doorway (a), indeed, I am myself responsible. I generally made the doorways of my nests narrow, so as to check evaporation and keep the nests from becoming too dry. It will be observed, however, that behind the hall (b) the entrance contracts, and is still further protected by a pillar of earth, which leaves on either side a narrow passage which a single ant could easily guard, or which might be quickly blocked up. Behind this is an irregular vestibule (c), contracted again behind into a narrow passage, which is followed by another, this latter opening into the main chamber (d). In this chamber several pillars of earth are left, almost as if to support the roof. Behind the main chamber is an inner sanctum divided into three chambers, and to which access is obtained through narrow entrances (f, f,f,f). Most of the pillars in the main chamber are irregular in outline, but two of them (g, g) were regular ovals, and round each, for a distance about as long as the body of an ant, the glass had been most carefully cleaned. This was so marked, and the edge of the cleaned portion was so distinct, that it is impossible not to suppose that the ants must have had some object in this proceeding, though I am unable to suggest any explanation of it.

I have already mentioned (ante, p. 23), that there is evidence of some division of labour among ants. Where, indeed, there are different kinds of workers,

this is self-evident, but even in species where the workers are all of one type, something of the same kind appears to occur.

In the autumn of 1875 I noticed an ant belonging to one of my nests of F. fusca out feeding alone. The next day the same ant was again out by herself, and for some weeks no other ant, so far as I observed, came out to the food. I did not, however, watch her with sufficient regularity. In the winter of 1876, therefore, I kept two nests under close observation, having arranged with my daughters and their governess, Miss Wendland (most conscientious observers), that one of us should look at them once an hour during the day. One of the nests contained about 200 individuals of F. fusca, the other was a nest of P. rufescens with the usual slaves, about 400 in number. The mistresses themselves never came out for food, leaving all this to the slaves.

We began watching on November 1, but did not keep an hourly register till the 20th, after which date the results are given in the following tables (see Appendix). Table No. 1 relates to the nest of F. fusca, and the ants are denoted by numbers. The hours at which we omitted to record an observation are left blank; when no ant was at the honey, the square is marked with an 0. An ant, marked in my register as No. 3, was at the time when we began observing acting as feeder to the community.

The only cases in which other ants came to the honey were at 2 P.M. on November 22, when another ant came

out, whom we registered as No. 4, another on the 28th, registered as No. 5. Other ants came out occasionally, but not one came to the honey (except the above mentioned) from November 28 till January 3, when another (whom we registered as No. 6) began feeding. After this a friend visited the honey once on the 4th, once on the 11th, and again on the 15th, when she was registered as No. 7.

Table No. 2 is constructed in the same way, but refers to the nest of Polyergus. The feeders in this case were, at the beginning of the experiment, registered as Nos. 5, 6, and 7. On November 22 a friend, registered as No. 8, came to the honey, and again on December 11; but with these two exceptions the whole of the supplies were carried in by Nos. 5 and 6, with a little help from No. 7.

Thinking now it might be alleged that possibly these were merely unusually active or greedy individuals, I imprisoned No. 6 when she came out to feed on the 5th. As will be seen from the table, no other ant had been out to the honey for some days; and it could therefore hardly be accidental that on that very evening another ant (then registered as No. 9) came out for food. This ant, as will be seen from the table, then took the place of No. 6, and (No. 5 being imprisoned on January 11) took in all the supplies, again with a little help from No. 7. So matters continued till the 17th, when I imprisoned No. 9, and then again, i.e. on the 19th, another ant (No. 10) came out for the food,