A preliminary study of the situation indicated that any filter for fine smoke particles must have a high resistance per unit of area, but that the total resistance must be comparatively low. In order to secure the large area necessary to bring the total resistance within reason, the experimental work was developed along three lines: The formation of a filter into a bag, cup, or jacket to surround the outside of the canister; the use of an arrangement sufficiently compact to go inside the canister; and the use of a filter as a separate unit, to be attached to the canister by an air connection. A survey of the possible filtering materials indicated that only two offered promise, namely, paper and felt. PAPER FILTERS Reports that the British had developed thin, creped, sulfite-cellulose wood pulp paper for filters led to an intensive study of this material by the Chemical Warfare Service. [Illustration: FIG. 104.—Crepe Paper Doughnut Filter Canister.] In general we may say that the development of paper filters (in sheet form) met with little success. Papers affording the required protection did not live up to the resistance specifications. The reason for this probably is in the method of making paper. The pulp is fed onto the screen of a Fourdrinier machine under conditions that do not permit of uniformity in the distribution of the fibers and consequently there is no uniformity in the size of pores. In order to eliminate the large holes, which allow the smoke to pass readily, the paper must be pressed to reduce these pores to the proper magnitude. This naturally results in an approximately equal decrease in the size of the small pores, with a consequent increase in the final resistance out of all proportion to the protection gained. A very satisfactory paper was finally produced, but the resistance was too high and it was necessary to increase the total available filtering area, which resulted in the accordion type of filter. This filter was incapable of development on a large scale because of the large amount of hand work required in assembling. The lack of uniformity in a single sheet has been overcome with some success by making up a filter from 40 to 80 layers of tissue or crepe paper, trusting that the law of chance would bring the large pores in some successive layer. Such a filter was adopted by the British, but since it did not give protection comparable with that afforded by felt filters, it was rejected in the United States. In the so-called “doughnut” filter use was made of tissue paper. Instead of seeking for uniformity in a vertical direction through a block of tissues, it was sought along the axis horizontal with the sheet. The effectiveness of such a filter was less than that of felt. In addition, serious difficulty was met in cutting the pile of tissue paper into the proper shape so that eventually it was abandoned as a production possibility. FELT FILTERS