Suppose the conditions of combustion are purposely violated; we at once have a gas producer. That is all gas producers are, extra bad stoves or furnaces, not always much worse than things which pretend to serve for combustion. Consider how ordinary gas is made. There is a red-hot retort or cylinder plunged in a furnace. Into this tube you shovel a quantity of coal, which flames vigorously as long as the door is open, but when it is full you shut the door, thus cutting off the supply of air and extinguishing the flame. Gas is now simply distilled, and passes along pipes to be purified and stored. You perceive at once that the difference between a gas retort and an ordinary furnace with closed doors and half choked fire bars is not very great. Consumption of smoke! It is not smoke consumers you really want, it is fuel consumers. You distill your fuel instead of burning it, in fully one-half, might I not say nine-tenths, of existing furnaces and close stoves. But in an ordinary gas retort the heat required to distill the gas is furnished by an outside fire; this is only necessary when you require lighting gas, with no admixture of carbonic acid and as little carbonic oxide as possible.

If you wish for heating gas, you need no outside fire; a small fire at the bottom of a mass of coal will serve to distill it, and you will have most of the carbon also converted into gas. Here, for instance, is Siemens' gas producer. The mass of coal is burning at the bottom, with a very limited supply of air. The carbonic acid formed rises over the glowing coke, and takes up another atom of carbon to form the combustible gas carbonic oxide. This and the hot nitrogen passing over and through the coal above distill away its volatile constituents, and the whole mass of gas leaves by the exit pipe. Some art is needed in adjusting the path of the gases distilled from the fresh coal with reference to the hot mass below. If they pass too readily, and at too low a temperature, to the exit pipe, this is apt to get choked with tar and dense hydrocarbons. If it is carried down near or through the hot fuel below, the hydrocarbons are decomposed over much, and the quality of the gas becomes poor. Moreover, it is not possible to make the gases pass freely through a mass of hot coke; it is apt to get clogged.

The best plan is to make the hydrocarbon gas pass over and near a red-hot surface, so as to have its heaviest hydrocarbons decomposed, but so as to leave all those which are able to pass away as gas uninjured, for it is to the presence of these that the gas will owe its richness as a combustible material, especially when radiant heat is made use of.

The only inert and useless gas in an arrangement like this is the nitrogen of the air, which being in large quantities does act as a serious diluent. To diminish the proportion of nitrogen, steam is often injected as well as air. The glowing coke can decompose the steam, forming carbonic oxide and hydrogen, both combustible. But of course no extra energy can be gained by the use of steam in this way; all the energy must come from the coke, the steam being already a perfectly burned product; the use of steam is merely to serve as a vehicle for converting the carbon into a convenient gaseous equivalent. Moreover, steam injected into coke cannot keep up the combustion; it would soon put the fire out unless air is introduced too. Some air is necessary to keep up the combustion, and therefore some nitrogen is unavoidable. But some steam is advisable in every gas producer, unless pure oxygen could be used instead of air; or unless some substance like quicklime, which holds its oxygen with less vigor than carbon does, were mixed with the coke and used to maintain the heat necessary for distillation. A well known gas producer for small scale use is Dowson's. Steam is superheated in a coil of pipe, and blown through glowing anthracite along with air.

The gas which comes off consists of 20 per cent. hydrogen, 30 per cent. carbonic oxide, 3 per cent. carbonic acid, and 47 per cent. nitrogen. It is a weak gas, but it serves for gas engines, and is used, I believe, by Thompson, of Leeds, for firing glass and pottery in a gas kiln. It is said to cost 4d. per 1,000 ft., and to be half as good as coal gas.

For furnace work, where gas is needed in large quantities, it must be made on the spot. And what I want to insist upon is this, that all well-regulated furnaces are gas retorts and combustion chambers combined. You may talk of burning coal, but you can't do it; you must distill it first, and you may either waste the gas so formed or you may burn it properly. The thing is to let in not too much air, but just air enough. Look, for instance, at Minton's oven for firing pottery. Round the central chamber are the coal hoppers, and from each of these gas is distilled, passes into the central chamber, where the ware is stacked, and meeting with an adjusted supply of air as it rises, it burns in a large flame, which extends through the whole space and swathes the material to be heated. It makes its exit by a central hole in the floor, and thence rises by flues to a common opening above. When these ovens are in thorough action, nothing visible escapes. The smoke from ordinary potters' ovens is in Staffordshire a familiar nuisance.

In the Siemens gas producer and furnace, of which Mr. Frederick Siemens has been good enough to lend me this diagram, the gas is not made so closely on the spot, the gas retort and furnace being separated by a hundred yards or so in order to give the required propelling force. But the principle is the same; the coal is first distilled, then burnt. But to get high temperature, the air supply to the furnace must be heated, and there must be no excess. If this is carried on by means of otherwise waste heat we have the regenerative principle, so admirably applied by the Brothers Siemens, where the waste heat of the products of combustion is used to heat the incoming air and gas supply. The reversing arrangement by which the temperature of such a furnace can be gradually worked up from ordinary flame temperature to something near the dissociation point of gases, far above the melting point of steel, is well known, and has already been described in this place. Mr. Siemens has lent me this beautiful model of the most recent form of his furnace, showing its application to steel making and to glass working.

The most remarkable and, at first sight, astounding thing about this furnace is, however, that it works solely by radiation. The flames do not touch the material to be heated; they burn above it, and radiate their heat down to it. This I regard as one of the most important discoveries in the whole subject, viz., that to get the highest temperature and greatest economy out of the combustion of coal, one must work directly by radiant heat only, all other heat being utilized indirectly to warm the air and gas supply, and thus to raise the flame to an intensely high temperature.

It is easy to show the effect of supplying a common gas flame with warm air by holding it over a cylinder packed with wire gauze which has been made red hot. A common burner held over such a hot air shaft burns far more brightly and whitely. There is no question but that this is the plan to get good illumination out of gas combustion; and many regenerative burners are now in the market, all depending on this principle, and utilizing the waste heat to make a high temperature flame. But although it is evidently the right way to get light, it was by no means evidently the right way to get heat. Yet so it turns out, not by warming solid objects or by dull warm surfaces, but by the brilliant radiation of the hottest flame that can be procured, will rooms be warmed in the future. And if one wants to boil a kettle, it will be done, not by putting it into a non-luminous flame, and so interfering with the combustion, but by holding it near to a freely burning regenerated flame, and using the radiation only.

Making toast is the symbol of all the heating of the future, provided we regard Mr. Siemens' view as well established.

The ideas are founded on something like the following considerations: Flame cannot touch a cold surface, i.e., one below the temperature of combustion, because by the contact it would be put out. Hence, between a flame and the surface to be heated by it there always intervenes a comparatively cool space, across which heat must pass by radiation. It is by radiation ultimately, therefore, that all bodies get heated. This being so, it is well to increase the radiating power of flame as much as possible. Now, radiating power depends on two things: the presence of solid matter in the flame in a fine state of subdivision, and the temperature to which it is heated. Solid matter is most easily provided by burning a gas rich in dense hydrocarbons, not a poor and non-luminous gas. To mix the gas with air so as to destroy and burn up these hydrocarbons seems therefore to be a retrograde step, useful undoubtedly in certain cases, as in the Bunsen flame of the laboratory, but not the ideal method of combustion. The ideal method looks to the use of a very rich gas, and the burning of it with a maximum of luminosity. The hot products of combustion must give up their heat by contact. It is for them that cross tubes in boilers are useful. They have no combustion to be interfered with by cold contacts.

The flame only should be free.

The second condition of radiation was high temperature. What limits the temperature of a flame? Dissociation or splitting up of a compound by heat. So soon as the temperature reaches the dissociation point at which the compound can no longer exist, combustion ceases. Anything short of this may theoretically be obtained.

But Mr. Siemens believes, and adduces some evidence to prove, that the dissociation point is not a constant and definite temperature for a given compound; it depends entirely upon whether solid or foreign surfaces are present or not. These it is which appear to be an efficient cause of dissociation, and which, therefore, limit the temperature of flame. In the absence of all solid contact, Mr. Siemens believes that dissociation, if it occur at all, occurs at an enormously higher temperature, and that the temperature of free flame can be raised to almost any extent. Whether this be so or not, his radiating flames are most successful, and the fact that large quantities of steel are now melted by mere flame radiation speaks well for the correctness of the theory upon which his practice has been based.