The conclusions deduced from a large number of experiments were that when, as in type 1, water is allowed to drip or flow in a fine stream upon a mass of carbide, the temperature rapidly rises until after eighteen to twenty-five minutes the maximum is reached, which varies from 400° to 700° C. (720° to 1120° Fahrenheit), and it is probable that in some of the mass the higher limit is always reached, as traces of tar are usually found in the residual lime, in some cases in sufficient quantity to make the lime yellow and pasty, while vapors of benzene and other polymerization products pass off with the gas. Leaving the question of temperature in this type of generator, another important question is the length of time during which the generation of gas continues after the water supply is automatically cut off. It is found that gas is evolved with increasing slowness sometimes for an hour and three quarters after the water supply has ceased, the total volume of gas so evolved being large. The experiments showed that in any automatic generator of this type the cut-off should be so arranged that one quarter of the total capacity of the gas holder is still available to store the slowly generating gas. The second class of generators bring about contact either by water rising from below to the carbide suspended in the cage (II, _A_), or by a cage of carbide suspended in a movable bell which, as it falls, dips the carbide into water, withdrawing the carbide from the water as the excessive generation of gas lifts the bell (II, _B_). Lewes found that under certain conditions generators of the type II, _B_ were far worse than those of type I. The trials were made with a movable glass bell, with counterweights, containing a half-pound of carbide. The maximum temperatures reached in four trials were 703°, 734°, 754°, and 807° C. Excessive heating took place in every case; in the last mentioned the temperature was far above the point at which acetylene is decomposed into carbon and hydrogen, a thin black smoke being formed immediately around the carbide while tar vapor poured out. On removing the residue after cooling it was found to be coated with soot and loaded with tar. On several occasions the charge was removed from the generator just after the maximum temperature was reached, and was found to be at a bright red heat.