For field concentration work two complete and separate lines of trenches were used and also several impact grounds. The trenches were built to simulate the trenches actually used in warfare. Each line of trench contained several concrete shell-proof dugouts and was also equipped with shelves into which boxes could be placed for holding the sample bottles. At intervals of one yard throughout the trenches there were electrical connections available for electrical sampling purposes. The various impact grounds were used for cloud gas attacks, and experiments with mustard gas or in many cases for static trials. The samples were collected by means of an automatic sampling apparatus. The work of the Division consisted in the first instance of determining the proper bursting charge. While a great deal of this work had been carried out in Europe, American gas shell were enough different to require that tests be carried out on them. The importance of this work is obvious, since phosgene, a substance with a low boiling point, would require a smaller bursting charge to open the shell and allow the substance to vaporize than would mustard gas, where the bursting charge must be not only sufficient to fragment the shell but also to scatter the liquid so that it would be atomized over the largest possible area. In the case of low boiling liquids it was necessary that the charge be worked out very carefully as a difference of one or two grams would seriously affect the concentration. Too small a charge would allow a cup to be formed by the base of the shell which would carry some of the liquid into the ground, while too great an amount of explosive tended to throw the gas too high into the air. After the bursting charge had been determined a large number of shell were repeatedly fired into the trenches, wooded areas, rolling and level ground, and the concentration of gas produced and the effect upon animals placed within the area ascertained. From the results of these experiments the Proving Division was able to furnish the artillery with data regarding how many shell of given caliber should be used, with corrections for ranges, wind velocities, temperatures, ground conditions, etc. Trials were also held to determine how many high explosive (H.E.) shell could be fired with gas shell on the same area without unduly affecting the concentration. This was important, because H.E. shell were useful in disguising gas bombardments. Gas shell can usually be distinguished by the small detonation on bursting. Experiments were performed to determine the decomposition of various gases on detonation. The shell were fired at a large wooden screen and burst on impact. Samples of gas were taken immediately and analyzed.