I can easily see it glittering on the red ground of the pot.

The day is drawing to a close and no alteration has yet taken place in the position of the trail.

A striking proof confirms this.
The trajectory is not a plane curve, but one which, at a certain point, deviates and goes down a little way to the lower surface of the cornice, returning to the top some eight inches farther.

I marked these two points of deviation in pencil on the vase at the outset.

Well, all that afternoon and, more conclusive still, on the following days, right to the end of this mad dance, I see the string of caterpillars dip under the ledge at the first point and come to the top again at the second.