the larvæ. She at once picked one up, and, with some little guidance from me, carried it off to the nest, returning at once for another, bringing some friends with her to help. When she knew her way, I gradually moved the cup across the table along the paper path

to M, placing it on a column five inches high. After a while the ants came to know the way quite well, and passed straight along the path from the nest to the larvæ at M. Having thus established a service of ants, I tried the following experiments :

1. I removed the piece of paper G F.

This dis

turbed them; but they very soon re-established the chain.

2. I turned round the central piece of the table G F, so that the paper G F was reversed, G being where F had been, and vice versa. This did not seem to diconcert the ants at all. They went straight over the paper as before, without a moment's hesitation.

3. When some ants were between 1 and D, I rotated the outer circle of the table halfway round, which of course carried the cup containing the larvæ from L to B. The ants took no notice of this, but went straight to L.

4. When some ants were between I and D, I rotated the table several times, bringing it finally to its original position. This disturbed them a good deal; but eventually they all continued their course to L.

5. When some ants were between I and D, I half rotated the two centre parts of the table, the result of which, of course, was that the ant was moving towards, instead of away from, the nest. In every case the ants turned round too, so as duly to reach L. So also those which were on their way back from the larvæ to the nest turned in the same manner.

6. When the ants were between 1 and D, I half rotated the whole table. Again the ants turned round too, though of course in this case, when they reached the place where L had been, the cup with the larvæ was behind them at B.

The two latter experiments, though quite in accordance with those previously made, puzzled me a good

deal. Experiment 3, as well as some of those recorded previously, seemed to show that ants were little guided in such cases by the position of surrounding objects. However, I was anxious to test this.

7. Accordingly I took a round box and placed it upside down on the table, having cut two niches, one at each side, where it lay on the paper path, so as to afford a passage for the ants, as in the experiments recorded in my previous paper; but on this occasion I left the lid on, cutting, however, a hole through which I could watch the result. In this case, therefore, the surrounding objects, i.e. the walls of the box, turned round with the table. Then, as before, when the ants were between 1 and D, I turned the table half round. The results were as follows:

In this case, then, only 11 ants turned; and as 4 of them were together, it is possible that 3 simply

followed the first. Moreover, the ants which turned did so with much more hesitation and less immediately. 8. For comparison, I then again tried the same experiment, but without the box. The results were as

Under these circumstances, therefore, all the ants but one certainly turned, and her movements were undecided.

From these last two experiments it is obvious that the presence of the box greatly affected the result, and yet the previous results made it difficult to suppose that the ants noticed any objects so distant as the walls of the rooms, or even as I was myself. The result surprised me considerably; but I think the explanation is given by the following experiments.

I again put some larvæ in a cup, which I placed in the centre of the table; and I let out an ant which I had imprisoned after the previous experiments, placing her in the cup; she carried off a larva to the nest and soon returned. When she was again in the cup I half rotated the table: when she came out she seemed a

little surprised; but after walking once round the cup, started off along the paper bridge straight home. When she returned to the cup I again half rotated the table. This time she went back quite straight. When she had come again, I once more half rotated the table; she returned quite straight. Again the same happened. A second ant then came: I half rotated the table as before. She went wrong for about an inch and a half, but then turned round and went straight home.

I was working by the light of two candles which were on the side of the table towards the nest. The next time the two ants came I half rotated the table as before, and moved the candles to the far side. This time the ants were deceived, and followed the paper bridge to the end of the table furthest from the nest. This I repeated a second time, with the same result. I then turned the table as before without altering the lights, and the ants (four of them) went back all right. I then again turned the table, altering the lights, and the ant went wrong.

I then altered the lights without rotating the table: the first ant went wrong; the second right; the third wrong; the fourth wrong; the fifth hesitated some seconds, and then went wrong; the sixth right; the seventh went all but to the edge the wrong way, but, after various wanderings, at last went right. When, therefore, the direction of the light was changed, but everything else left as before, out of seven ants, five were deceived and went in the wrong direction.

After an interval of a week, on March 25, I arranged