This box was rectangular, and about three feet long, and was placed upon a table, and covered with a cloth. When any particular word was mentioned by the company, M. Kempelen caused the machine to pronounce it, by introducing his hands beneath the cloth, and apparently giving motion to some parts of the apparatus. Mr. Thomas Collinson, who had seen this machine in London, mentions, in a letter to Dr. Hutton, that he afterwards saw it at M. Kempelen’s own house in Vienna, and that he then gave it the same word to be pronounced which he gave it in London, viz. the word _Exploitation_, which, he assures us, it again distinctly pronounced with the French accent. M. Kratzenstein seems to have been equally unsuccessful; for though he assured M. de Lalande, when he saw him in Paris, in 1786, that he had made a machine which could speak pretty well, and though he showed him some of the apparatus by which it could sound the vowels, and even such syllables as _papa_ and _mama_, yet there is no reason to believe that he had accomplished more than this. The labours of Kratzenstein and Kempelen have been recently pursued with great success by our ingenious countryman, Mr. Willis, of Cambridge. In repeating Kempelen’s experiment, shown in Fig. 49, he used a shallower cavity, such as that in Fig. 50, and found that he could entirely dispense with the introduction of the hand, and could obtain the whole series of vowels by sliding a flat board C D over the mouth of the cavity. Mr. Willis then conceived the idea of adapting to the reed cylindrical tubes, whose length could be varied by sliding joints. When the tube was greatly less than the length of a stopped pipe in unison with the reed, it sounded I, and by increasing the length of the tube, it gave E, A, O, and U, in succession. But what was very unexpected, when the tube was so much lengthened as to be 1½ times the length of a stopped pipe in unison with the reed, the vowels began to be again sounded in an inverted order, viz. U, O, A, E, and then again in a direct order, I, E, A, O, U, when the length of the tube was equal to twice that of a stopped pipe, in unison with the reed. [Illustration: _Fig. 50._] Some important discoveries have been recently made by M. Savart respecting the mechanism of the human voice;[23] and we have no doubt that, before another century is completed, a _Talking_ and a _Singing machine_ will be numbered among the conquests of Science. [23] See _Edinburgh Journal of Science_, No. viii., p. 200. LETTER IX.