Chappe, in 1793, devised an improvement on this in what was called the T telegraph. An upright post supported a cross-bar (the top of the T), at each end of which were the short dependent beams, making the figure a complete Roman capital T. The horizontal bar as first used could be worked by ropes within the telegraph-house, so as to be inclined either to right or left. It thus had three positions. Each dependent beam could be worked (also from within the house) so as to turn upwards, horizontally, or downwards (regarding the top bar of the T as horizontal), thus having also three positions. It is easily seen that, since each position of one short beam could be combined with each position of the other, the two together would present three times three arrangements, or nine in all; and as these nine could be given with the cross-bar in any one of its three positions, there were in all twenty-seven possible positions. M. Chappe used an alphabet of only sixteen letters, so that all messages could readily be communicated by this telegraph. For shorter distances, indeed, and in all later uses of Chappe’s telegraph, the short beams could be used in intermediate positions, by which 256 different signals could be formed. Such telegraphs were employed on a line beginning at the Louvre and proceeding by Montmartre to Lisle, by which communications were conveyed from the Committee of Public Welfare to the armies in the Low Countries. Telescopes were used at each station. Barrère stated, in an address to the Convention on August 17, 1794, that the news of the recapture of Lisle had been sent by this line of communication to Paris in one hour after the French troops had entered that city. Thus the message was conveyed at the rate of more than 120 miles per hour. Various other devices were suggested and employed during the first half of the present century. The semaphores still used in railway signalling illustrate the general form which most of these methods assumed. An upright, with two arms, each capable of assuming six distinct positions (excluding the upright position), would give forty-eight different signals; thus each would give six signals alone, or twelve for the pair, and each of the six signals of one combined with each of the six signals of the other, would give thirty-six signals, making forty-eight in all. This number suffices to express the letters of the alphabet (twenty-five only are needed), the Arabic numerals, and thirteen arbitrary signals.