As truly observed by M. Agassiz, these moulins occur only at those parts of the glacier which are not much rent by fissures, for only at such portions can the little rills produced by superficial melting collect to form streams of any magnitude. The valley of unbroken ice formed in the Mer de Glace near Trelaporte is peculiarly favourable for the collection of such streams; we see the little rills commencing, and enlarging by the contributions of others, the trunk-rill pouring its contents into a little stream which stretches out a hundred similar arms over the surface of the glacier. Several such streams join, and finally a considerable brook, which receives the superficial drainage of a large area, cuts its way through the ice. [Sidenote: MOULINS EXPLAINED.] But although this portion of the glacier is free from those long-continued and permanent strains which, having once rent the ice, tend subsequently to widen the rent and produce yawning crevasses, it is not free from local strains sufficient to produce _cracks_ which penetrate the glacier to a great depth. Imagine such a crack intersecting such a glacier-rivulet as we have described. The water rushes down it, and soon scoops a funnel large enough to engulf the entire stream. The moulin is thus formed, and, as the ice moves downward, the sides of the crack are squeezed together and regelated, the seam which marks the line of junction being in most cases distinctly visible. But as the motion continues, other portions of the glacier come into the same state of strain as that which produced the first crack; a second one is formed across the stream, the old shaft is forsaken, and a new one is hollowed out, in which for a season the cataract plays the thunderer. I have in some cases counted the forsaken shafts of six old moulins in advance of an active one. Not far from the Grand Moulin of the Mer de Glace in 1857 there was a second empty shaft, which evidently communicated by a subglacial duct with that into which the torrent was precipitated. Out of the old orifice issued a strong cold blast, the air being manifestly impelled through the duct by the falling water of the adjacent moulin. These shafts are always found in the same locality; the portion of the Mer de Glace to which I have referred is never without them. Some of the guides affirm that they are motionless; and a statement of Prof. Forbes has led to the belief that this was also his opinion.[A] M. Agassiz, however, observed the motion of some of these shafts upon the glacier of the Aar; and when on the spot in 1857, I was anxious to decide the point by accurate measurements with the theodolite.