GENERATOR IMPURITIES.--The generator impurities present in the crudest acetylene consist of oxygen and nitrogen, _i.e._, the main constituents of air, the various gaseous, liquid, and semi-solid bodies described in Chapter II., which are produced by the polymerising and decomposing action of heat upon the carbide, water, and acetylene in the apparatus, and, whenever the carbide is in excess in the generator, some lime in the form of a very fine dust. In all types of water-to-carbide plant, and in some automatic carbide-feed apparatus, the carbide chamber must be disconnected and opened each time a fresh charge has to be inserted; and since only about one-third of the space in the container can be filled with carbide, the remaining two-thirds are left full of air. It is easy to imagine that the carbide container of a small generator might be so large, or loaded with so small a quantity of carbide, or that the apparatus might in other respects be so badly designed, that the gas evolved might contain a sufficient proportion of air to render it liable to explode in presence of a naked light, or of a temperature superior to its inflaming-point. Were a cock, however, which should have been shut, to be carelessly left open, an escape of gas from, rather than an introduction of air into, the apparatus would follow, because the pressure in the generator is above that of the atmosphere. As is well known, roughly four-fifths by volume of the air consist of nitrogen, which is non-inflammable and accordingly devoid of danger- conferring properties; but in all flames the presence of nitrogen is harmful by absorbing much of the heat liberated, thus lowering the temperature of that flame, and reducing its illuminating power far more seriously. On the other hand, a certain quantity of air in acetylene helps to prevent burner troubles by acting as a mere diluent (albeit an inferior one to methane or marsh-gas), and therefore it has been proposed intentionally to add air to the gas before consumption, such a process being in regular use on the large scale in some places abroad. As Eitner has shown (Chapter VI.) that in a 3/4-inch pipe acetylene ceases to be explosive when mixed with less than 47.7 per cent. of air, an amount of, say, 40 per cent. or less may in theory be safely added to acetylene; but in practice the amount of air added, if any, would have to be much smaller, because the upper limit of explosibility of acetylene-air mixtures is not rigidly fixed, varying from about 50 per cent. of air when the mixture is in a small vessel, and fired electrically to about 25 per cent. of air in a large vessel approached with a flame. Moreover, safely to prepare such mixtures, after the proportion of air had been decided upon, would require the employment of some additional perfectly trustworthy automatic mechanism to the plant to draw into the apparatus a quantity of air strictly in accordance with the volume of acetylene made