MEMOIRS OF THE NATIONAL ACADEMY OE SCIENCES. 281 



and carbon. These error.- arc hardly avoidable, but on the whole they appear to counterbalance one 

 another so that their effect is eliminated in the averages of a considerable number of experiments. 



2. Differences in the activity of the subjects in the twoclassesof experiments. These differ- 

 ences arc not easy to avoid. The man in the chamber may make more muscular effort on one 

 day than on another in taking down his bed in the morning and in setting it up at night, or he 

 may move about more in caring- for the food and excretory products and weighing himself and 

 t e absorbers. In the work experiments there may be differences in the external muscular work 

 despite the best efforts to make the amounts constant from day today. These differences in 

 muscular activity, though small, may affect the metabolism of matter and energy. 



3. The energy furnished by the alcohol may not be as efficient, calorie for calorie, in meeting 

 the demands of the body as the energy from the materials which it replaces. It is hardly to be 

 supposed that the experimental errors in categories (1) and (2) will be considerable. It is still less 

 probable that they will be so concentrated in either the alcohol or nonalcohol experiments as to 

 materially affect the average results. If. therefore, the differences between the figures for the 

 experiments of the two classes are large and reasonably constant, it would seem fair to attribute 

 them to differences in the actual value of the alcohol as compared with isodynamic amounts of 

 fats and carbohydrates. 



The figures of Table 17 show differences to the disadvantage of the alcohol. The differ- 

 ences are, however, mainly within the range of experimental error.' 1 



In the more directly comparable experiments (Group I) the conditions with and without 

 alcohol were closely similar. In Group II there were not inconsiderable differences between the 

 amounts of protein and energy in the diet, in the number of subjects, in the number of experi- 

 ments, and in the amounts of muscular exercise. These differences do not, in our judgment, 

 destroy the value of the comparisons in Group II, though thev T do make the differences in result 

 less decisive. The results of Group II are, therefore, valuable as confirming those of Group I. 



Gains and losses of oody mah rial as indicative of tin relative effectiveness of alcohol. — The 

 differences in the gains or losses of protein and fat in the experiments with alcohol as compared 

 with the others are slightly to the disadvantage of the alcohol. They thus imply that, calorie 

 for calorie, the energy furnished to the body by the alcohol was less effective than that furnished 

 by the carbohydrates and fats. These differences may be due to experimental errors, but even 

 if they are wholly charged to the alcohol they make it only slightly inferior to the nutrients 

 which it replaces. The inferiority is found only in the work experiments; in the rest experi- 

 ments there is practically no difference between the alcohol and the ordinary nutrients in 

 effectiveness. 



Amounts of energy metabolized as indicative of the relative effectiveness of alcohol. — The 

 results here are similar to those found in the comparison of gains or losses of material. This is 

 to be expected, since the two measures are really different expressions of the same fundamental 

 fact. In the rest experiments the results with and without alcohol are practically identical. The 

 inferiority of the alcohol is limited to the work experiments. 



Energy of material metabolized in work experiments with and without alco/iol. — In the work 

 experiments more material was oxidized than the food supplied, and the deficiency was made up 

 by drafts upon the previously accumulated store of body protein and fat. Under these circum- 

 stances the body may be supposed to use the energy economically so as to make the drafts upon 



a The differences between the results with ami without alcohol are in all cases small. Considering them from 

 the ordinary mathematical standpoint, they are, of course, noticeable; but in such physiological experimenting as 

 this the unavoidable errors of individual experiments are considerable, and it is only when a large number of 

 such experiments are averaged that differences of one or two parts in one hundred could properly be regarded as 

 significant. Indeed, in this whole discussion there is danger of being misled by the figures in the tables unless one 

 constantly recalls the fact that the range of unavoidable variation is wide. When, however, the averages of large 

 numbers of experiments show a constant difference on one side or the other, it may be permissible to use such 

 differences for conclusions and generalizations. On the whole, it might seem that in these experiments the results 

 were sufficiently numerous to imply a slight inferiority of the alcohol in respect to the economy of the use of energy; 

 but this inference rests upon the rather questionable assumption of the absolute equality of all conditions other than 

 the presence or absence of alcohol in the diet. 



