If the jet is left burning the bell will gradually sink again as the gas is used up, and should thereafter maintain an automatic balance without attention.
It can be turned off at any moment by simply closing the taps at the jet or, better, the lower tap at the generator, when the bell rises sufficiently to take the carbide out of the water; but if it is required to leave the generator unlit for a considerable time, it is better to turn off the tap on the top first. This causes the inner portion of the bell to fill with gas which cannot escape, and as that in the outer part burns out, the bell sinks to the bottom and remains there, the gas itself imprisoned in the inner chamber excluding the water from the carbide. The exact arrangement varies in different patterns of generator, but the above may be taken as roughly indicating the action, and further information may always be obtained from the maker or dealer.
Emptying should always be done out of doors, as the odour of acetylene gas is most objectionable, and for the same reason rubber tubes, &c., should be securely tied on, so that the slightest escape may be avoided.
If the exhibition has been a short one it will often be found that the upper cells have not been affected by the water, in which case they may be put back in the tin and used again, but it is not generally advisable to put in less than the full charge to begin with as the weight of the carbide plays a definite part in securing the smooth action of the apparatus. The sludge should be thrown away (it forms a good manure for the garden) and the entire generator thoroughly dried, otherwise rust will quickly appear.
Theoretically one of these generators may be filled and left standing indefinitely, but in practice it is not advisable, as the damp in the atmosphere is apt to produce a very slow generation of gas, sufficient often to cause a decided smell.
Of generators which act by admitting water to the carbide perhaps the best known is the A.L. or 'Popular' Model (Fig. 6), this being, in fact, a pattern designed for motor-car head-lights, but which answers well for lantern work.
Its exact operation need hardly be described here in full detail. It will suffice to say that the water gains access to the carbide by 'creeping' up between two concentric copper cones, and in the event of over-generation the pressure of the gas automatically checks the flow.
This generator is smaller than the gasometer pattern, and hence can be recommended for portability; but in my experience the light is not quite so steady, and the control rather less delicate, thereby causing on occasions a perceptible smell, especially if left standing for a considerable time.