clear that if she had been left alone, all, or at least most of, these ants would have accompanied her to the store of larvæ. Three of them Three of them were accidentally

allowed to do so; but of the remainder, only 5 found their way to the larvæ; all the others, after wandering about a while, returned hopelessly to the nest.

One of the ants which I employed in my experiments was under observation several days. I was, however, away from home most of the day, and when I left in the morning and went to bed at night I put her in a bottle; but the moment she was let out she began to work again. On one occasion I was away for a week, and on my return I let her out of the bottle, placing her on a little heap of larvæ about 3 feet from the nest. Under these circumstances I certainly did not expect her to return. However, though she had thus been six days in confinement, the brave little creature immediately picked up a larva, carried it off to the nest, and, after half an hour's rest, returned for another.

I conclude, then, that when large numbers of ants come to food they follow one another, being also to a certain extent guided by scent. The fact, therefore, does not imply any considerable power of intercommunication, There are, moreover, some other circumstances which seem to show that their powers in this respect are but limited. For instance, I have already mentioned that if a colony of Polyergus changes the situation of its nest, the mistresses are all carried to the new one by the slaves. Again, if a number of F.

fusca are put in a box, and in one corner a dark place of retreat is provided for them with some earth, one soon finds her way to it. She then comes out again, and going up to one of the others, takes her by the jaws. The second ant then rolls herself into a heap, and is carried off to the place of shelter. They then both repeat the same manœuvre with other ants, and so on until all their companions are collected together. Now it seems to me difficult to imagine that so slow a course would be adopted if they possessed any considerable power of descriptive communication.

On the other hand, there can, I think, be no doubt that they do possess some power of the kind.

This seems to me clearly shown by the following observations. In order, if possible, to determine whether the ants in question were brought to the larvæ, or whether they came casually, I tried (1875) the following experiments: I took three tapes, each about 2 feet 6 inches long, and arranged them parallel to one another and about 6 inches apart. One end of each I attached to one of my nests (L. niger), and at the other end I placed a small glass. In the glass at the end of one tape I placed a considerable number (300 to 600) of larvæ. In the. second I put two or three larvæ only; in the third none at all. The object of the last was to see whether many ants would come to the glasses under such circumstances by mere accident; and I may at once say that but few did so. I then took two ants and

placed one of them to the glass with many larvæ, the other to that with two or three. Each of them took a larva and carried it to the nest, returning for another, and so on. After each journey I put another larva in the glass with only two or three larvæ, to replace that which had been removed. Now, if other ants came under the above circumstances as a mere matter of accident, or accompanying one another by chance, or if they simply saw the larvæ which were brought and consequently concluded that they might themselves also find larvæ in the same place, then the numbers going to the two glasses ought to be approximately equal. In each case the number of journeys made by the ants would be nearly the same; consequently, if it was a matter of scent, the two glasses would be in the same position. It would be impossible for an ant, seeing another in the act of bringing a larva, to judge for itself whether there were few or many larvæ left behind. On the other hand, if the friends were brought, then it would be curious to see whether more were brought to the glass with many larvæ, than to that which only contained two or three. I should also mention that, excepting, of course, the marked specimens, every ant which came to the larvæ was imprisoned until the end of the experiment. I give the details in the Appendix.

The results of the above experiments are shown at a glance in the following Table :

is calculated to increase the number of friends brought by the ants to the glass with only 2 or 3 larvæ, for several reasons, but especially because in many cases an ant which had for some time had access to a glass with many larvæ was suddenly deprived of it, and it might well be that some time elapsed before the change was discovered. Some stray ants would, no doubt, in any case have found the larvæ; and we may probably allow for about 25 under this head. Again, some would, no doubt, casually accompany their friends; if we allow 25 also in this respect, we must deduct 50 from each side, and we shall have 254 against 54. Nevertheless, even without any allowances, the results seem to me very definite. Some of the individual cases, especially perhaps experiments 9, 10, 20, 21, and 22 (see Appendix), are very striking; and, taken as a whole, during 52 hours, the ants which had access to a glass containing numerous larvæ brought 304 friends; while during 59 hours those which were visiting a glass with only 2 or 3 larvæ brought only 104 to their assistance.

One case of apparent communication struck me very much. I had had an ant (L. niger) under observation one day, during which she was occupied in carrying off larvæ to her nest. At night I imprisoned her in a small bottle; in the morning I let her out at 6.15, when she immediately resumed her occupation. Having to go to London, I imprisoned her again at 9 o'clock. When I returned at 4.40, I put her again