It has been stated that in order to avoid loss of heat from a flame through the burner, that burner should present only a small mass of material (_i.e._, be as light in weight as possible), and should be constructed of a bad heat-conductor. But if a small mass of a material very deficient in heat-conducting properties comes in contact with a flame, its temperature rises seriously and may approach that of the base of the flame itself. In the case of coal-gas this phenomenon is not objectionable, is even advantageous, and it explains why a burner made of steatite, which conducts heat badly, in always more economical (of heat and therefore of light) than an iron one. In the case of acetylene the same rule should, and undoubtedly does, apply also; but it is complicated, and its effect sometimes neutralised, by a peculiarity of the gas itself. It has been shown in Chapters II. and VI. that acetylene polymerises under the influence of heat, being converted into other bodies of lower illuminating power, together with some elemental carbon. If, now, acetylene is fed into a burner which, being composed of some material like steatite possessed of low heat-conducting and radiating powers, is very hot, and if the burner comprises a tube of sensible length, the gas that actually arrives at the orifice may no longer be pure acetylene, but acetylene diluted with inferior illuminating agents, and accompanied by a certain proportion of carbon. Neglecting the effect of this carbon, which will be considered in the following paragraph, it is manifest that the acetylene issuing from a hot burner--assuming its temperature to exceed the minimum capable of determining polymerisation-- may emit less light per unit of volume than the acetylene escaping from a cold burner. Proof of this statement is to be found in some experiments described by Bullier, who observed that when a small "Manchester" or fish-tail burner was allowed to become naturally hot, the quantity of gas needed to give the light of one candle (uncorrected) was 1.32 litres, but when the burner was kept cool by providing it with a jacket in which water was constantly circulating, only 1.13 litres of acetylene were necessary to obtain the same illuminating value, this being an economy of 16 per cent.