generator-house may eventually become so offensive to the nostrils that the liquid has to be renewed; but when this happens it is due to the accumulation in the water of the water-soluble impurities of the crude acetylene. If, as should be done, the gas is passed through a washer or condenser containing much water before it enters the holder the sulphuretted hydrogen and ammonia will be extracted, and the seal will not acquire an obnoxious odour for a very long time. Four principal substances have been proposed for lowering the freezing- point of the water in an acetylene-holder seal; common salt (sodium chloride), calcium chloride (not chloride of lime), alcohol (methylated spirit), and glycerin. A 10 per cent. solution of common salt has a specific gravity of 1.0734, and does not solidify above -6° C. or 21.2° F.; a 15 per cent. solution has a density of 1.111, and freezes at -10° C. or 14° F. Common salt, however, is not to be recommended, as its solutions always corrode iron and steel vessels more or less quickly. Alcohol, in its English denatured form of methylated spirit, is still somewhat expensive to use, but it has the advantage of not increasing the viscosity of the water; so that a frost-proof mixture of alcohol and water will flow as readily through minute tubes choked with needle- valves, or through felt and the like, or along wicks, as will plain water. For this reason, and for the practically identical one that it is quite free from dirt or insoluble matter, diluted spirit is specially suitable for the protection of the water in cyclists' acetylene lamps, [Footnote: As will appear in Chapter XIII., there is usually no holder in a vehicular acetylene lamp, all the water being employed eventually for the purpose of decomposing the carbide. This does not affect the present question. Dilute alcohol does not attack calcium carbide so energetically as pure water, because it stands midway between pure water and pure alcohol, which is inert. The attack, however, of the carbide is as complete as that of pure water, and the slower speed thereof is a manifest advantage in any holderless apparatus.] where strict economy is less important than smooth working. For domestic and larger installations it is not indicated. As between calcium chloride and glycerin there is little to choose; the former will be somewhat cheaper, but the latter will not be prohibitively expensive if the high-grade pure glycerins of the pharmacist are avoided. The following tables show the amount of each substance which must be dissolved in water to obtain a liquid of definite solidifying point. The data relating to alcohol were obtained by Pictet, and those for calcium chloride by Pickering. The latter are materially different from figures given by other investigators, and perhaps it would be safer to make due allowance for this difference. In Germany the Acetylene Association advocates a 17 per cent. solution of calcium