The fuel values for the diet are at times extraordinarily high. Thus, in the experiment of February 16, 1906, the subject consumed a breakfast of mixed diet having an actual fuel value of approximately 4,400 calories. In several other instances the fuel value was over 3,300 calories. In general, however, it was not far from 500 to 1,200 calories.

The total increments, also given in table 249, varied from minus values (which are obviously due to faulty technique, defects in the determination of the basal value, or undue activity in the basal period) to the increment of 334 calories noted in a heavy supper experiment. Of special significance is the relationship between the total increments and the fuel values of the food intake, i. e, the cost of digestion. The highest noted is that of 23 per cent for an experiment with J. J. C. on May 9, 1910, with glidine. Values above 10 per cent appear chiefly in the protein experiments, thus emphasizing strongly the fact that the ingestion of protein causes not only an actual maximum increase in metabolism higher than values obtained with the other nutrients, but a greater proportional increase when compared to the fuel value of the intake. Striking irregularities may be noted, and even with protein we find values under 10 per cent as frequently as above 10 per cent.

With bananas and sugar the cost of digestion is relatively high in practically all instances, averaging about 7 per cent. In the one experiment in which it is low, namely, 3 per cent with F. M. M. on April 8, 1909, there was an ingestion of but 9 grams of sugar. It is possible that the low increment of 19 calories noted on that date in a 3-hour observation may have been due to the fact that the superimposed effect of the cane sugar included in the diet on the other days was here absent. On the other hand, on the days when bananas alone were eaten, the two experiments with Dr. H. showed a cost of digestion of 5 and 7 per cent. Still another experiment made with F. M. M. showed no change.

While averaging results as diversified as those recorded in this table may appear to be a questionable procedure, yet we may tentatively state that the cost of digestion, or the relationship between the fuel value of the intake and the increase in the heat production due to the ingestion of food is, with carbohydrates, not far from 6 per cent on the average. With fat, aside from the two high values found with cream in the experiments with the subjects A. L. L. and A. H. M. in the spring of 1906, a small cost of digestion is noted, the average being not far from 2 per cent. With protein, although wide differences are found, the values ranging from 2 to 23 per cent, the average value is approximately 10 per cent.

The three experiments with milk on March 21, 22, and 23 were planned to be comparable as the same amount of milk was given the subjects; the fuel value ranged from 382 to 444 calories. The increments obtained are somewhat irregular. A minimum of 11 calories was found with H. R. D. on the first day and fairly comparable values of 59 and 51 calories, respectively, were obtained on the two succeeding days. The fuel value of the milk taken in the last two experiments was practically the same. We see no reason for omitting the experiment on the first day, and hence the three experiments represent an average cost of digestion of milk of approximately 10 per cent.

In the previous comparisons, the fuel value did not exceed approximately 1,900 calories, except in one experiment with butter and potato chips, in which the intake of energy was 3,202 calories. In a group of experiments with an excess amount of food, characterized as "heavy-breakfast" experiments, the fuel value ranged from 2,142 to 4,378 calories. The cost of digestion in these experiments was fairly uniform, ranging from 4 to 6 per cent, with an average of 5 per cent. There were also two experiments of much longer duration than any of the other experiments included in this table, viz, those with H. R. D., April 10-11, and A. L. L., April 6-7, in which "heavy suppers" were taken with high fuel values. The cost of digestion was 10 and 12 per cent respectively. These higher figures may be due to faulty basal values or unusual activity in the food experiment, or the experiments may have been long enough to obtain all the increment which actually took place. They do not lend themselves, however, to very critical analysis.

Under ordinary conditions the normal individual rarely eats a meal containing a pure nutrient or a meal in which there is an excessive proportion of any single nutrient, but usually a fairly balanced combination of nutrients. It is accordingly of considerable practical significance that the six experiments with a heavy breakfast show such uniform percentages. While the use of average figures for the several groups of carbohydrates, fats, and proteins may be somewhat questionable, with mixed diets we may fairly state that the excess heat production as a result of ingesting such a diet is 5 per cent of the fuel value of the intake. In all of the heavy-breakfast experiments, the basal metabolism was not reached during the experimental period; the value of 5 per cent is therefore probably somewhat low and a value of 6 per cent would be more nearly in accord with the actual facts. We suggest, therefore, that as a general factor a heat production equivalent to 6 per cent of the fuel value may be expected as the result of the ingestion of a mixed diet.

The respiration experiments, summarized in table 250, were primarily designed to study the maximum effect rather than the total increment, and were therefore shorter than the calorimeter experiments. With the exception of one experiment with a mixed diet and one with beefsteak and potato chips, the fuel value of the diet did not exceed 600 calories, this value being much smaller than that of the diets used in the calorimeter experiments. The duration of the increment was also shorter, although it is evident that in many instances the basal value had not been reached at the end of the experimental period. The values given in such cases may be partial rather than maximum.