52 



of the ration slight. It is clear, then, that the exactitude with which 

 we should know the weights of the various servings or portions of the 

 different foods depends both upon the chemical composition of the food 

 and upon the number of orders eaten. vSince we know the amounts of 

 protein and energy furnished by the various classes of food, viz, soups, 

 dairy products, cereals, breads, vegetables, meats, etc., and the per- 

 centage variation apph'ing to each class, it is quite possible to com- 

 pute, approximately at least, the effect produced b}' these variations 

 in the final figures expressing the energy and protein in the average 

 ration. The following table shows the amounts of nutrients furnished 

 by the different classes of food during the whole study, the percentage 

 variations, and the effects on the totals: 



Table 13. — Proportion of protein and energy in average daily ration per man {for the 

 entire inrestigation) famished by various classes of food and probable maximum error 

 introduced by variation in the size of consecutive orders of the various dishes. 



Classes of food. 



Soup 



Hash, fish, meats 



Dairy products 



Breakfast cereals 



Breads and crackers. 



iJes.serts 



Vegetable.s 



Fruits 



Beverages 



Combination meals. . 



Sugars 



Eggs 



Total 



Multipl3nng the total probable maximum error in protein, 6.70 

 grams, by 100, and dividing by the total digestible protein furnished, 

 TS. 1 grams, gives 8.6 per cent as the average variation for digestible 

 protein; a similar computation gives 7.8 per cent as the average vari- 

 ation for available energ}"; and the average of both is 8.2 per cent. 

 This tigure is believed to be the largest error that abnormal variations 

 in the size of orders is likel}' to introduce into the dietaries. In the 

 opinion of the writer, the actual error lies considerably below this, 

 ])eing perhaps 3 or i per cent. A carefully conducted dietary stud}^ 

 where the method of weighing, sampling, and analyzing all food on 

 hand at the beginning, received during, and remaining at the end of 

 the period of examination is employed would probably yield results 

 within about 1 per cent of the actual amount of nitrogen and energ}^ 

 in the food. The error inherent in the analytical determination of 

 nitrogen and heat of combustion is so small that it ma}' be neglected; 

 ])ut when one computes "protein" from the total nitrogen by using 

 factors, another and .sometimes considerable error may be introduced. 



