The third method developed was the sulfuric acid smoke. This smoke was produced by passing dry air through a tower of solid pieces of sulfur trioxide and then mixing the vapor with a large volume of air at 50 per cent relative humidity. It is not a clogging smoke and the filtering efficiency does not change materially in the time of exposure required for a test. The smoke lends itself easily to chemical analysis and offers data as to exact particulate cloud concentrations which will penetrate canisters; photometric measurements are also applicable. [Illustration: FIG. 106.—Tobacco Smoke Apparatus for Testing Canisters.] The fourth method consists in the use of tobacco smoke. This is generated by passing air over ignited sticks of a mixture of tobacco (63 per cent), rosin (30 per cent) and potassium nitrate (7 per cent). This smoke is composed of particles of extreme uniformity in size; chemically it is relatively inert. It is not a clogging smoke and is not sensitive to moisture and dilution. The density of the effluent smoke is compared with that of the entering smoke in a Tyndall beam, and the filtering capacity of the material determined in terms of the amount of air necessary to dilute the entering air to the same concentration of the effluent air. The method is simple in manipulation and the test is a rapid one (50 canisters per day). Because of the apparent superiority of tobacco smoke as a testing smoke, the accompanying disadvantages are possibly outweighed. From the standpoint of inherent chemical properties, the general desirability of a suitable testing smoke would decrease in the following order: tobacco, sulfuric acid, ammonium chloride. CHAPTER XIX SIGNAL SMOKES The success of pyrotechnics in night signalling led, during the World War, to considerable attention being paid to the development of pyrotechnic signals for day use. This was mainly directed to the production of distinctive smokes, which should have the same long range visibility under varying light conditions. Since a gray or white smoke might be confused with the smoke produced accidentally by the explosion of shell, it was necessary to use smoke of definite and unmistakable colors, and red, blue, yellow, green and purple smokes were developed. During the early part of the war, only a yellow smoke was in use, though others were added later. PRODUCTION OF COLORED SMOKES There are three possible ways of obtaining signal smokes. I. Mechanically dispersing solids. II. Chemical Reaction. III. Volatilization of colored materials.