We are now in a position to judge what is the practical efficiency of the gas-engine. Some years since, in a letter which I addressed to _Engineering_, and which also appeared in the _Journal of Gas Lighting_,[2] I showed (I believe for the first time) that, in the Otto-Crossley engine, 18 per cent. of the total heat was converted into power, as against the 8 per cent. given by a very good steam-engine. About the end of 1883 a very elaborate essay, by M. Witz, appeared in the _Annales de Chimie et de Physique_, reporting experiments on a similar engine, which gave an efficiency somewhat lower. Early in 1884 there appeared in _Van Nostrand's Engineering Magazine_ a most valuable paper, by Messrs. Brooks and Steward, with a preface by Professor Thurston,[3] in which the efficiency was estimated at 17 to 18 per cent. of the total heat of combustion. Both these papers show what I had no opportunity of ascertaining, that is, what becomes of the 82 per cent. of heat which is not utilized--information of the greatest importance, as it indicates in what direction improvement may be sought for, and how loss may be avoided. But, short as is the time that has elapsed since the appearance of these papers, you will find that progress has been made, and that a still higher efficiency is now claimed. [Footnote 2: See _Journal_, vol. xxxv, pp. 91, 133.] [Footnote 3: Ibid., vol. xliii., pp. 703, 744.] When I first wrote on this subject, I relied upon some data which led me to suppose that the heating power of ordinary coal gas was higher than it really is. At our last meeting, Mr. Hartley proved, by experiments with his calorimeter, that gas of 16 or 17 candles gave only about 630 units of heat per cubic foot. Now, if all this heat could be converted into power, it would yield 630 x 772, or 486,360 f.p.; and it would require only 1,980,000 / 486,360 = 4.07 cubic feet to produce 1 indicated horse power. Some recent tests have shown that, with gas of similar heating power, 18 cubic feet have given 1 indicated horse power, and therefore 4.07 / 18 = 22.6 of the whole heat has been converted--a truly wonderful proportion when compared with steam-engines of a similar power, showing only an efficiency of 2 to 4 per cent.