These results are of very great interest, as they not only show clearly the facts already pointed out, but indicate that in any automatic apparatus on this principle the cut-off should be so arranged that at least one-fourth of the total holder capacity is still available to store the slowly generated gas.

Another very important deduction to be derived from the figures is the large excess of water over and above the theoretical quantity required to ensure complete decomposition of the carbide by this process, this being to a certain extent dependent upon the form of the generator.

According to theory, 64 parts by weight of carbide require only 36 parts by weight of water to completely decompose them and convert the lime into calcic hydrate. This would mean that each pound of calcic carbide needs a little under half a pint of water to complete the decomposition, whilst owing to evaporation due to the heat produced, half the added water is driven off as steam with the acetylene or left mechanically adhering to the lime, and the smallest quantity likely to complete the action would be a pint to a pound of carbide ; in reality the only way is to add sufficient water to drown the residue.

If this is not done the lime forms so protective a coating to the carbide that small quantities often remain undecomposed, and if the residues are thrown into a drain or cesspool, the evolution of acetylene would give an explosive mixture, which, on account of its low point of ignition, would be a serious danger.

Points of considerable interest to the generator maker are the space occupied by a given weight of carbide, the volume of the lime formed from it on decomposition, and the volume of gas that can be evolved from a given space filled with carbide.

The density of calcic carbide is 2.2, and therefore a cubic foot of solid carbide would weigh 137 lbs. In practice, however, the weight of carbide which can be got into a cubic foot space depends upon the size commercially sent out. A fair average would be 80 lbs. per cubic foot of carbide space, and this weight of carbide at 5 cubic feet per lb. would yield 400 cubic feet of acetylene.

One pound of pure calcic carbide yields 1*15 lbs. of slaked lime (one kg. of carbide yields 1.156 gr. of slaked lime), and the volume this will occupy depends entirely upon the way in which the water is brought in contact with it.

In an automatic apparatus of the first class, where water drips slowly upon the carbide in sufficient quantity to decompose it but not to flood it, the lime swells up and occupies 2 to 2.5 times the bulk of the original carbide; when, however, the water flows in more rapidly, the impact of the water beats down the lime and the space occupied is not so large.

In generators of the second class, in which water rises from below, the weight of the undecomposed carbide above it presses down the lime below and keeps it in a compact mass occupying about half more space than the carbide from which it was formed.

With the third type of generator it really becomes a question of the rate at which the excess of undissolved calcic hydrate settles, and this will be discussed later on.

The large proportion of water vaporised during the evolution of acetylene at once draws attention to the necessity of arranging all the generator connections in such a way that condensation shall not lead to stoppage of the delivery pipes.

It must also be clearly borne in mind that the liquid products condensible from the gas are of the most corrosive character both to paint and metal.

The moment that acetylene is subjected to the action of high temperatures, changes of great complexity at once commence, causing a great deal of impurities, and the tar is likely to cause considerable trouble, as it is of very viscous character, and, if it condenses in the delivery tubes, causes the lime-dust and carbon particles to collect and bring about stoppage.

A still more important evil, however, is to be found in the alteration which takes place in the composition of the gas, and which reduces the illuminating value of the gas to a serious extent.

A very considerable proportion of the generation takes place at a temperature above 600° C, about which point polymerisation commences. As benzene forms a large proportion of it, it is carried forward as vapours and remains suspended even in its passage through the gas-holder and ordinary pipes. Benzene requires three times the volume of air for combustion that acetylene does, and the result is that the most perfect acetylene burner shows a tendency to smoke directly any quantity of benzene is formed.

When acetylene has been made in a generator at an undue temperature, it carries with it benzene vapour, which as it commences to condense assumes a vesicular form, and on coming to the extremely minute holes which form the apertures of the burner the mechanical scrubbing which it encounters causes the breaking up of the vesicles and the deposition of the benzene and other hydrocarbons held in suspension by benzene, which soak into the steatite and carbonise. The pressure of finely-divided carbon has a great effect in determining the decomposition of acetylene itself, so that a rapid growth of carbon takes place at the burner, and no ordinary clearing of the deposited carbon from the exterior will ever make the nipple fit for constant use again.

It will be found with experience that the smoking of a burner will be overcome quite as much by attention to the temperature in the generator as to the burner itself, and where a generator is in use which gives overheating, a well-arranged scrubbing apparatus that would get rid of the benzene from the gas would be found a distinct advantage in stopping burner troubles.

At first sight these results seem an absolute condemnation of the second type of generators, but the fact remains that they constitute a very large percentage of those on the market, and that the best of them show no signs of overheating.

The reason of this apparent anomaly is that under certain conditions, which can be clearly defined, excessive heating is avoided.

The raising bell which draws a mass of wet carbide above the surface of the water is bad from every point of view.

But generators in which water rises from below and so attacks the carbide can be made safe if the arrangements are such that the water is never driven back from the carbide and the bulk of carbide is sufficiently subdivided. Under these conditions the slowly rising water is always in excess at the point where it decomposes the carbide, so that the evaporation by rendering heat latent keeps down the temperature, and although the steam so formed partly decomposes the carbide in the upper portion of the charge, the action is never sufficiently rapid to give anything approaching a red heat. In order to fulfil these conditions it is necessary that there should be a holder of considerable capacity, and that the leading tube conducting the gas from the generator to the holder should be of sufficient diameter to freely conduct away the gas, the water at the same time being allowed to rise in the generator so slowly as to do away with any risk of over-generation.

In the best generators of this class these conditions are more or less approached, and it is unusual to find that the melting point of tin, 228° C, has been reached in the charge of carbide during decomposition.

Where generators of this class are automatic and have no rising holder to take the gas, it is found that they work satisfactorily when supplying the number of lights for which they were designed, but if they are over-driven and the action becomes too violent, excessive heating takes place, whilst the turning off of the gas, and consequent driving back of the water from the carbide, also has a tendency to cause it. If however, the water has risen sufficiently slowly, the carbide below the surface has been practically all decomposed, so that the heating only takes place over a limited zone.

The makers of generators that are liable to give rise to excessive heating invariably deny the possibility of such an action taking place with their generator, and, if it is proved, fall back upon the defence that, even if the mass does become red hot, there is no particular danger.

In such generators the active danger of explosion only exists whilst any air is left mixed with the acetylene, and in those which have holders to take the gas as it is formed, the air remaining in the generator is swept rapidly over into the holder and out of the range of the source of heat; but with automatic generators this is not always the case, and the air space in the generators should always be made as small as possible, and some arrangement should be adopted, if possible, by which the air in the generator could be rinsed out by a little of the previously produced acetylene before decomposition of the carbide by water commences.

Under these circumstances danger from explosion during generation would disappear, but the drawbacks of smoky flames, reduced illuminating power, and choking tubes would still remain.

The generators of the third class are those in which carbide is allowed to fall into an excess of water, and these have many advantages. In such generators, as long as there is water present, and lime sludge is not allowed to accumulate, it is impossible to get above a temperature of 100° C, whilst with a properly arranged tank the temperature never exceeds the air temperature by more than a few degrees. Under these conditions the absence of polymerisation and the washing of the nascent and finely-divided bubbles of gas by the lime water in the generator yields acetylene of a degree of purity unapproached by any other form of apparatus.

This form of generator, however, although exhibiting the great advantages mentioned above, has the drawback of being one of the least economical in the output of acetylene per pound of carbide used, as the gas having to bubble through the water is rapidly dissolved by it, whilst in an apparatus in which only the surface of the water touches the gas the amount dissolved is comparatively small. The result is that with generators of this class the generation rarely exceeds 4.2 cubic feet of acetylene per pound of carbide instead of 5 cubic feet per pound.

Probably, therefore, from a practical point of view, the generators which are the best for general working are those of the second class, which are used in connection with a holder of sufficient size to take the gas evolved from the full charge of carbide employed.

Approximately, after an hour's standing each kg. of calcic carbide will give ten litres of lime sludge, or 1 lb. of carbide will yield eight pints, which can be got rid of by a sludge cock at the bottom of the apparatus.