M. Troyon[138] considered that the blade of flint was used in connection with sand as well as water. This latter view appears, at first sight, far more probable, as the sawing instrument has in some instances cut nearly 3∕4 of an inch into the stone, which, it would seem, could hardly have been accomplished with a simple flint saw; and the sides of the saw-kerf or notch show, moreover, parallel striæ, as if resulting from the use of sand. The objection that at first occurred to my mind against regarding the sawing instrument as having been of flint was of a negative character only, and arose from my not having seen in any of the Swiss collections any flint flakes that had indisputably been used for sawing by means of sand. At one time I fancied, from the character of the bottom and sides of the notches, that a string stretched like that of a bow might have been used with sand in the manner in which, according to Oviedo,[139] the American Indians sawed in two their iron fetters, and I succeeded in cutting off the |45| end of an ancient Swiss hatchet of hard steatite by this means. I found, however, that the bottom of the kerf thus formed was convex longitudinally, whereas in the ancient examples it was slightly concave. It is therefore evident that whatever was used as the saw must have been of a comparatively unyielding nature, and probably shorter than the pebble or block of stone it was used to saw, for even the iron blades used in conjunction with sand and water by modern masons become concave by wear, and, therefore, the bottom of the kerf they produce is convex longitudinally. I accordingly made some further experiments, and this time upon a fragment of a greenstone celt of such hardness that it would readily scratch window-glass. I found, however, that with a flint flake I was able to work a groove along it, and that whether I used sand or no, my progress was equally certain, though it must be confessed, very slow. I am indeed doubtful whether the flint did not produce most effect without the sand, as the latter to become effective requires a softer body in which it may become embedded; while by working with the points and projections in the slightly notched edge of the flake, its scratching action soon discoloured the water in the notch. What was most remarkable, and served in a great measure to discredit the negative evidence to which I before referred, was that the edges of the flake when not used with sand showed but slight traces of wear or polish. On the whole, I am inclined to think that both the Swiss antiquaries are in the right, and that the blocks of stone were sawn both with and without sand, by means of flint flakes, but principally of strips of wood and bone used in conjunction with sand.[140] The reader may consult Munro’s Lake-Dwellings, 1890, p. 505.