granted that the generator is efficiently constructed, decomposition of the carbide is absolutely complete, so that no loss of gas occurs in this fashion; (2) the gas is evolved at a low temperature, so that it is unaccompanied, by products of polymerisation, which may block the leading pipes and must reduce the illuminating power; (3) the acetylene is not mixed with air (as always happens at the first charging of a water-to- carbide apparatus), which also lowers the illuminating power; and (4) the gas is freed from two of its three chief impurities, viz., ammonia and sulphuretted hydrogen, in the generating chamber itself. To prevent the loss of acetylene by dissolution, carbide-to-water generators are occasionally fitted with a reaction grid placed only just below the water-level, so that the acetylene has no more than 1 inch or so of liquid to bubble through. The principle is wrong, because hot water being lighter than cold, the upper layers may be raised to the boiling-point, and even converted into steam, while the bulk of the liquid still remains cold; and if the water actually surrounding the carbide is changed into vapour, nearly all control over the temperature attending the reaction is lost. The hand-fed carbide-to-water generator is very simple and, as already indicated, has proved itself perhaps the best type of all for the construction of very large installations; but the very simplicity of the generator has caused it more than once to be built in a manner that has not given entire satisfaction. As shown at L in Fig. 6, p. 84, the generator essentially consists of a closed cylindrical vessel communicating at its top with a separate rising holder. At one side as drawn, or disposed concentrically if so preferred, is an open-mouthed pipe or shoot (American "shute") having its lower open extremity below the water-level. Into this shoot are dropped by hand or shovel lumps of carbide, which fall into the water and there suffer decomposition. As the bottom of the shoot is covered with water, which, owing to the small effective gas pressure in the generator given by the holder, stands a few inches higher in the shoot than in the generator, gas cannot escape from the shoot; because before it could do so the water in the generator would have to fall below the level of the point _a_, being either driven out through the shoot or otherwise. Since the point _b_ of the shoot extends further into the generator than _a_, the carbide drops centrally, and as the bubbles of gas rise vertically, they have no opportunity of ascending into the shoot. In practice, the generator is fitted with a conical bottom for the collection of the lime sludge and with a cock or other aperture at the apex of the cone for the removal of the waste product. As it is not desirable that the carbide should be allowed to fall directly from the shoot into the thicker portion of the sludge within the conical part of the generator, one or more grids is