The need for an efficient, practical laboratory method for determining the potency of tvphoid vaccines long has been recognized by immunologists concerned with the development, production, and assay of the product. The need in relation to production and control in routine manufacture of typhoid vaccine is mainly for a testing method which will permit the ready demonstration that an individual lot, or successive lots, of vaccine equal or exceed an established level of potency. Researchers, on the other hand, require an assay method which, in addition to the foregoing, permits quantitative estimation of potency, i. e., which permits numerical expression of potency in absolute or relative terms with an estimable degree of uncertainty. Only by the use of such a method is the investigator enabled to detect or evaluate differences or changes in potency as effected by experimental treatments or conditions of interest. Methods early used in the determination of the potency of typhoid vaccines were based upon parenteral administration of the products to rabbits and subsequent in viro determination of the agglutinin response (15), or the actual protection of such immunized rabbits against subacute infection following injection of living, virulent Salmonella typhosa (18). It is beyond the scope of this report to discuss or criticize these tests in detail and it is considered sufficient to state that such methods have proved to be neither adequately definitive nor reproducible to warrant their continued use. Investigation and establishment by Siler and his associates at the Army Medical School (22) of the mouse-protection test for assessing the antibody response of humaDs to typhoid immunization stimulated

The determination of values, even repeated measurements of a certain dimension of a single inanimate article, is subject to errors or lack of precision. Likewise, in the mass production of articles designed to be alike, determination of a certain dimension or attribute of each article produced yields a series of values differing from unit to unit. If these values are plotted as a frequency distribution, they describe a pattern characteristic of the process under consideration. Every process, no matter now rigorous the attempt to control it, is subject to inherent variability, and the pattern of variability for each process is unique unto itself. When any change from the basic pattern occurs, some new source of variability must have entered the process. The methods of quality control are based on this principle. Prior to World War II, Shewhart (1931, 1939) and Pearson (1935) studied the application of statistical methods to problems pertaining to the control of variability of manufactured articles and introduced' the basic techniques of modern quality control. The urgent demand for testing procedures combining accuracy and economy, occasioned by the accelerated production schedules of