The train of wheels also set into motion a cylinder twenty inches in diameter and two feet six inches long; on this were fixed a number of brass bars of different lengths and thicknesses which in their revolution acted upon a row of fifteen keys or levers; three of these corresponded to the three little wind chambers containing air at different pressures, and, by means of little chains, operated their respective valves. There were seven levers set apart for operating the fingers, their respective chains making bends at the shoulders and elbows of the automaton, and terminated at the wrist in the ends of what I may call metacarpal levers attached to the fingers which were armed at their tips with leather to imitate the flesh of the natural hand. The motion of the mouth was controlled by four of the levers, one to open the lips so as to give to the wind a greater issue, one to bring them closer together, and so contract the passage, a third to draw the lips backward and away from the flute, and the fourth to push them forward over the edge of the embouchure. The last of the fifteen levers is the cleverest of all, for it has the power of controlling the tongue, an accomplishment which I think everyone (unless he be an Odd Volume) will agree with me is a very difficult one to acquire. The barrel worked upon a screwed bearing (similar to that of the cylinder of a phonograph), so that in its revolution all the levers described a spiral line sixty-four inches long, and, as the barrel during the performance made twelve revolutions it followed that the levers passed over a distance of no less than 768 inches in going through its performance of twelve tunes. In a Memoir read before the Académie des Sciences, M. Vaucanson described the very beautiful methods by which the barrel was set out, and by which the positions of the bars were determined on its surface so as to regulate the supply of air and to control the actions of the fingers, the motion of the lips and the movements of the tongue; and he gave a most interesting analysis of the acoustics of wind instruments; but time will not permit me to make more than this passing reference to them. The picture on the screen (Fig. 21) is a photographic reproduction of the plate attached to M. Vaucanson’s Memoir (a somewhat rare little tract published in 1738) in which his three automata are shown, and I hold in my hand a copy of the translation by Dr. Desaguliers, published in London in 1742, which, the imprint tells us, was “_sold at the long room at the Opera House in the Haymarket, where the mechanical figures are to be seen at 1, 2, 5, and 7 o’clock in the afternoon_.” [Illustration: Fig. 21.]