It has been found that, if the optical readings from the Tyndall meter are plotted as ordinates against the time _t_ (the time elapsed after detonation) as abcissas, and that portion of the curve between _t_ = 0 and _t_ = 30 considered, the curve generally descends sharply at first, from a high point representing the density immediately after the production of the smoke, to a point in the neighborhood of _t_ = 8, where it flattens out and descends much more slowly with a slope that changes little. The area under the significant portion of the curve, that is, the area circumscribed by the curve from the point _t_₃₀ to _t_₀, the vertical axis from this point to the origin, the horizontal axis from the origin to _t_₃₀ and the line perpendicular to this axis, cutting the curve at _t_₃₀, is a rough measure of the relative values of different smokes. This area is calculated as the sum of two rectangles, from _t_₀ to _t_₈ and from _t_₈ to _t_₃₀. Some results are as follows: Area 30 Phenyldichloroarsine 181 Triphenyldichloroarsine 178 Diphenylcyanoarsine 137 Diphenylchloroarsine 101 Cyanogen bromide 94 Methyl dichloroarsine 70 Phenylimidophosgene 69 Mustard gas 38 The curves in Fig. 101 show the way in which the readings fall off with time. Each substance of course has its characteristic curves. [Illustration: FIG. 101.—Typical Curves Showing the Decrease in Concentration of Smoke Cloud with Time.] TOXIC MATERIALS The selection of materials for the production of toxic smokes can only be carried out experimentally. A number of very toxic substances have been shown to be valueless as toxic smokes because of low penetration, decomposition during the process of smoke production, or for other reasons. Arsenic compounds produce smokes distinctly better than the average. Inorganic compounds which have high melting and boiling points are very poor smoke producers. The only exception to this is magnesium arsenide, which may suffer decomposition. Compounds like mercuric chloride and arsenic tribromide, which boil or sublime at comparatively low temperatures, produce good smokes. Most materials which boil below 130° C. produce no smoke as they evaporate on dispersion. It is difficult to set any upper limit for the boiling point beyond which materials do not produce good smokes, but in all probability 500° C. is not far from the maximum. Liquids and solids are, on the whole, almost equally good as smoke producers. The physical condition of the material has no great effect upon the amount of smoke which it will produce. This seems to depend only upon the physical and chemical properties of the material. TOXIC SMOKE APPARATUS