It is not meant to suggest that the impurities generally to be found in bulk supplies are necessarily harmful to pyrotechnic results. This is not so; salts give far better results in their natural or mineral form than do those prepared synthetically. As an example of this saltpetre may be cited. For pyrotechnic purposes the best obtainable is that from Bengal, yet an analysis of this would probably be found to be less pure than that synthetically prepared in Germany. But experiments have shown that samples of the latter, taken from the same cask, but in different parts, produce very distinctly varying results pyrotechnically. Pyrotechny is an art, chemistry is a science, and although it is impossible to deny that the former is greatly indebted to the latter for the supply and production on a commercial scale of chemical ingredients, yet it is possible to overestimate the position of chemistry in the art, or possibly it might be more correct to say that pyrotechny has its own chemistry. Chemistry without pyrotechnic experience is apt to lead to erroneous conclusions. To take a concrete instance: in an article in a famous encyclopædia, obviously written by a chemist of standing, a portion deals with the use of metal salts in the production of colour; the writer gives copper as producing green, which no doubt it does in the laboratory; in practice, however, copper is used solely for the production of blue. The question of purity in chemicals used in pyrotechny is a secondary consideration, that is, of course, as long as the adulterants have no adverse effect either as regards the pyrotechnic result or the safety of the worker in manipulation. What is of first importance is its pyrotechnic suitability, that is, it must produce the required result and must be consistent. Unequal results are the bugbear of the firework makers. As we have seen, constant experiments are necessary to keep an even standard, but with irregularly functioning chemicals these would be multiplied to an impossible degree. The first group of compositions for consideration is that nearest related to gunpowder, in fact, for the purposes of a work on pyrotechny, gunpowder may be considered a particular case of this class. The governing principle of this group, and one may say of all firework compositions, is the same. For combustion to take place oxygen must be present. When an inflammable article such as a piece of paper is set on fire it takes up oxygen from the air. A pyrotechnic composition, however, is so arranged that one of the ingredients has a supply of oxygen which it is ready to give up; another, or others, are of a kind ready to receive and combine with this oxygen. The oxygen-supplying ingredient which is by far the most frequently used is saltpetre, or, as it was formerly called, nitre, known chemically as nitrate of potash.