the nest and the rotating table as before, and let out three ants which I had imprisoned on the 19th, and which knew their way. I put them on the larvæ at M as before. The paper pathway had been left untouched. The ants examined the larvæ and then went straight home along the paper path; but, to my surprise, only one of them carried off a larva. Nevertheless they had evidently taken the news to the nest, for the ants at once began coming to the cup in considerable numbers and carrying off the larvæ. I do not altogether understand this proceeding, and unluckily had not marked the first three ants; so that I cannot tell whether they brought or sent their friends. It seems possible that they felt unequal to the exertion of carrying a burthen to the nest until they had had some food.

When the ants were fairly at work I turned the table 90 degrees. In this case eight ants which were on their way to the larvæ continued their march along the paper, while two turned back; but none left the paper and went across the table straight for the larvæ.

I then stopped the experiment for a while, so that the excitement might subside; as when the ants become too numerous it is not so easy to watch them.

When all was quiet I put the cup with the larvæ on the middle of the table, and covered the greater part of the table with the box as before. In a short time some ants again came to the larvæ, and then, just as they were leaving the cup on their way home, I turned the table, as before, half round.

Under these circumstances, however, instead of

turning as in the previous experiment, ten ants, one after another, continued their course, thus coming out of the box at the end furthest from the nest. When ten ants successively had, under these circumstances, gone wrong, to make the experiment complete, I tried in again, everything being the same, except that there was no box. Under these circumstances five ants, one after the other, turned directly the table was rotated.

From these experiments, therefore, it seems clear that in determining their course the ants are greatly influenced by the direction of the light.

March 27.-I let out two ants imprisoned on the 25th, and placed them on the larvæ, which I put on a column 7 inches high, covered with blue paper, and communicating with the nest by the paper path (A, Fig. 29) arranged as usual, but supported on pins. At first I arranged it as shown below, placing the larvæ at M, on a table 18 inches in diameter, so that the ants, on arriving at the larvæ, made nearly a semicircle round the edge of the table. I then gradually moved the larvæ to M' and afterwards to M". The ants, however, obviously knew that they were going unnecessarily round. They

M

Fig. 29.

M

ran along the paper bridge in a very undecided manner,

continually turning round and often coming down the

pins; while in returning to the nest they persistently came down the side of the pillar nearest to the nest, though I repeatedly attempted to guide them the other way. Even when placed on the paper bridge between M and M', they were very dissatisfied. In fact, it was obvious that they knew they were being sent a long way round, and were attempting to make a shorter cut.

Fig. 30.

I then again placed the larvæ on the column at M, and when the ants were once more going to and fro regularly along the paper path, I altered the position of the column and larvæ to M', placing the edge of the pillar, which the ants had been accustomed to ascend, towards the paper bridge, connecting it with the original bridge by a sidebridge a, M being an inch from the original bridge. Under these circumstances three ants ran on to M; then two found their way over the bridge a to M'. Of the next ten ants, five went to M and five over a to M'. The next ten all went over the paper bridge a to M'.

ΣΕ

M'

I then put the pillar and the larvæ on the other side of the original paper path at M', connected with the main path by a short bridge a', taking for a' a new piece of paper, so that scent would be no guide. I left the little bridge a in its place. The ants went as

follows:

It seems clear, therefore, that though the ants did not trust so much to their eyes as a man would have done under similar circumstances, yet that they were to some extent guided by sight.

I then removed all the paper

Fig. 31.

M"

pathways and put the pillar to M. Of the first two ants which came to the table, the first found the pillar in five minutes, the second, after wandering about for a quarter of an hour, gave the search up in despair, and went home. I then moved the pillar to M', and watched the next ant that came on to the table; she found it in a minute or

ΕΣ

M'

two. I then moved it to M". Two ants came together. One found the pillar in 7 minutes; the other took no less than 25, although, as already mentioned, the table was only 18 inches in diameter. Obviously, therefore, though it seems clear that they are helped by sight, still these last observations support those previously recorded, and show that in finding their way they do not derive by any means so much assistance from their eyes as we should under corresponding circumstances.

T