21

               machine (which was not available in the different jails), no great
              amount of attention was given to this part of the work on
              prisoners; a certain number of observations were, however, carried
              out in the Physiological Laboratory on the menial staff and on
              animals.

Methods of work.

     6. A short account of our methods is necessary in order to give some
indication of the extent of the investigations; and at the same time it will
obviate any repetitions in the account furnished of our results.

           I. Analyses of Food-stuffs and Excreta.—The food materials in use
                in seven different jails in Bengal were analysed as regards their
                protein, carbohydrate, fat, moisture and ash. As will be seen from
                the tables of analyses, we were not content with a single analysis,
                but made repeated analyses of the same sample to obtain an
                average that would be, as far as possible, absolutely accurate. To
                ensure this we estimated the protein in at least six samples—often
                more—of the food-stuffs from the different jails, and never
                accepted any result that departed more than 0.5 per cent. from
                the average of the analyses closest to each other in amount. This
                we considered necessary as the food-materials are often con-
                taminated and it is difficult to obtain samples absolutely pure—
                this remark applies particularly to the dals which are often mixed
                with one another. As will be seen, we were able to obtain results
                that are exceedingly close to each other, and may justly claim that
                the average percentages presented are, for the samples analysed,
                absolutely accurate.

                       The protein estimations were all carried out by the Kjeldahl
                process as modified by Bernard Dyer. A modification which we
                found of the greatest service was to add a few drops of a weak solu-
                tion of phenolphthalein to the cross-tube leading from the distilla-
                tion flask; as soon as the ammonia begins to come over this indica-
                tor becomes red and remains so until all the ammonia is distilled
                over, when the colour disappears. By using this simple method a
                great source of trouble in Kjeldahl's process is eliminated, viz., a
                want of knowledge of the proper time to stop the distillation. It
                will be evident that to lay down any definite time, such as half -an-
                hour, as usually stated, is manifestly absurd: the time required will
                depend largely on the source of heat used in the distillation, and
                furthermore, we found that different samples of the same substance