successive journeys from the nest to the base of the pencil. I then moved the pencil 6 inches to D, and the two following routes are marked 5 and 6. In one of them, 5 (thick white line), the ant found a stray

Routes followed in experiment No. 3, as described in text.

The line at the six arrows represents a paper bridge going to nest. C, china cup on top of pencil. D, pencil moved. E, place where a stray larva was found. 1, 2, 3, 4, dotted lines show the nearly airect journeys. 5, thick white line (crossing C in black) of route returning to nest, the ant having picked up a stray larva at E. 6, very circuitous thin white line of track from nest to pencil D.

larva at E, with which she returned to the nest, without finding the pencil at all. On the following journey, shown in the fine white zigzag line (6), she found the pencil at last, but, as will be seen, only after many meanderings.

Diagram of complex path traversed in experiment 4.

A, first position of pencil. B, second position of pencil. 1, 2, straight lines of two tracks of the observed ants. 3, winding narrow white line, showing course pursued by the same ant before arriving at B, when the position of the pencil was unchanged.

Experiment 4.-I then repeated the observation

on three other ants (see Figs. 15-17) with the same result the second was 7 minutes before she found the pencil, and at last seemed to do so accidentally; the third actually wandered about for no less than half an hour (Fig. 15), returning up the paper bridge several

Diagram representing three tracks of an ant in another experiment,

A, the first position of pencil and the food, towards which and from the base-line of nest 1 and 2 lead by nearly direct broadish white lines to A. When the latter was removed to B the ant, in its effort to reach this, pursued the narrow white winding line ending in 3 →

ceding, the results of which are shown in the figures 16 and 17, seem to prove that this species of ant, at any

rate, guides itself but little by sight. This, which I had not at all anticipated, seems to follow from the fact that after the pencil and tray of larvæ had been removed but a short distance to the right or left, the

Fig. 17.

Another tracing showing a similar experiment. 1, 2, 3, the direct broad lines towards A; and 4, the complicated track made when reservoir of larvæ was removed to B.

ants on their journey to the shifted object travelled very often backwards and forwards and around the spot where the coveted object first stood. Then they would retrace their steps towards the nest, wander hither and

S

thither from side to side between the nest and the point A, and only after very repeated efforts around the original site of the larvæ reach, as it were accidentally, the object desired at B.

Another evidence of this consists in the fact that if when ants (L. niger) were carrying off larvæ placed in a cup on a piece of board, I turned the board round so that the side which had been turned towards the nest was away from it, and vice versâ, the ants always returned over the same track on the board, and, in consequence, directly away from home.

If I moved the board to the other side of my artificial nest, the result was the same. Evidently they followed the road, not the direction.

In order further to test how far ants are guided by sight and how much by scent, I tried the following experiment with Lasius niger. Some food was put out at the point a on a board measuring 20 inches by 12 (Fig. Fig. 18.

18), and so arranged that the ants in going straight to it from the nest would reach the board at the point b, and after passing under a paper tunnel, c, would proceed between five inches in length and 12

in height. When they got to know their way, they went quite straight along the line de to a.

The board was