in the harpoons of red-deer horn, the real significance of which had then, for the first time, been recognised, although a few specimens had already been found in one or two other caves and rock-shelters of the district. Since then, however, they have been discovered in considerable numbers, and are regarded as typical relics of the transition period. In face of these, and other rapidly accumulating facts, proving the existence of deposits of human débris containing relics stratigraphically proved to be later than those of the Reindeer period, but older than those of the polished Stone Age, de Mortillet abandoned the hiatus theory and filled up the gap by adding a new epoch to his previous classification of the Palæolithic Age, which he called Tourassien, after the Grotte de la Tourasse (Haute Garonne). This station was explored in 1891 by MM. Chamaison and Darbas, and yielded, along with a few other bone relics of intermediary forms, no less than a dozen harpoons of the kind now under review. The deposits in which they were found lay beneath a series of Neolithic burials. The flatness which is characteristic of these harpoons is due to the fact that the texture of red deer horn is spongy in the interior, and consequently it is only the outside of the horn that is used in their manufacture. On the other hand, the harpoons of the Palæolithic period have round stems and conical butts, with two projecting knobs close to the butt-end, instead of a hole, for keeping the string from falling away when the harpoon becomes eliminated from the handle. Altogether, they have a finer finish than the former, and show, either two rows of barbs, one on each side, or one row, having the barbs more closely set. The larger specimens are invariably made of reindeer-horn, but the smaller ones are sometimes made of bone.
(2) La Grotte de Reilhac.
Another station which has yielded flat harpoons is the Grotte de Reilhac (Lot), described by MM. Cartailhac and Boule (1889). Although the circumstances in which the Reilhac specimens have been found are not so definite as to give them a precise chronological value, the facts are by no
