Work on the felt filters started about June, 1918. Great difficulties were met in the beginning, as a felt satisfactory for this purpose must be made under carefully controlled conditions and production conditions during the war did not readily lend themselves to such control. However, the opportunities afforded in felt making for uniform packing and arranging of the fibers (the whole process of making a felt, is one of gradual packing of fibers into a relatively small volume) are such as to assure a greater degree of success than is the case in paper making. Very successful filters have been obtained with the use of felt. There are two serious objections to its use, however. The first is the great cost of the filter (this was above one dollar per filter at the close of the War); the second is that felt is a valuable industrial commodity. It is thus very desirable that a cheaper and a less important industrial material be found. THE 1919 CANISTER Just before the Armistice, the Gas Defense Long Island Laboratory brought out the so-called “1919 Canister,” which consisted of an oval section, perforated metal, war gas material container with a central, flat, perforated breathing tube connected to a nozzle at one end. (See also page 228.) After this inner container is packed with the war gas chemicals, a filter jacket is slipped over it and the top edge sealed to the inner container. [Illustration: FIG. 105.—1919 Felt Filter Canister.] Attempts were made to put paper filters on this canister by wrapping it with layers of paper. In some cases, layers of coarse burlap or mosquito netting were applied between the layers of paper to give mechanical strength and air space. The fact that many filters gave good protection showed that a filter of this type and material is possible, but the operations of wrapping and sealing require careful work in production and inspection and even with the greatest skill and care, imperfections are almost impossible to avoid. This chance of defects, together with the labor involved, makes the process undesirable. A THEORY OF SMOKE FILTERS Tolman, Wells and Gerke, during the course of their work on toxic smokes, developed the following theory of smoke filters. The phenomena occurring in the filtration of smoke are exceedingly complicated, but the general nature of the process may be simply described in terms of the kinetic properties of the small particles comprising the smoke. A filter may be regarded as a series of minute capillaries through which the smoke slowly flows. In order that filtration may take place, it is not necessary to assume that the capillaries of the filter are smaller than the particle, for the particles may diffuse to the walls of the capillaries and it is believed that with typical filters this is the actual method of smoke removal for particles less than 10⁻⁴ cm. in diameter.