53 



1)ecaii.se of uncertainty regarding the nitrogen factor for protein/' 

 The use of " coetScients of digestibilit\' " moreover is attended with 

 uncertainty' for, while they yield reasonably accurate results in the 

 average of a considerable number of studies, the actual error in some 

 individual results has been found to be as high as 5 per cent in one 

 or more nutrients.* The method of average weights which it was 

 necessar}' to employ appears, therefore, to give results that are sonie- 

 wiiat less accurate than the usual methods, but it is believed that the 

 results are nevertheless sufficient!}' exact to be trustworthv and of 

 ])ractical value. 



Another source of error in dietary work lies in the failure to ol)tain 

 representative samples for anal3'sis. In this investigation the unusu- 

 ally large variety of foods eaten by the various men made it practi- 

 cally impossible to sample and analyze every article; on the other 

 hand, it should be pointed out that many of the materials were con- 

 sumed to such a limited extent that, as far as the effect on the results 

 is concerned, the}' might be estimated with ample accurac}' from 

 available data, or, indeed, neglected altogether without materially 

 changing the final results. With a knowledge of the number of orders 

 of each food eaten during the periods of study, it was possible to judge 

 whether any given article was of sufficient importance to demand 

 analysis. Table 1, page 6, gives the percentage of protein and energ\' 

 per gram of the foods used during the dietary studies, including those 

 analyzed in connection with these studies and those for which such 

 figures were estimated from previous anahses. There is no wa}'^ of 

 learning precisely how nearly the figures obtained by our method of 

 sampling, analysis, and estimation represent the true values of protein 

 and energy for the amount of food involved. The writer is of the 

 opinion that, although the sampling, etc., in this case may not have 

 been as accurate as in the ordinary dietary study, nevertheless the 

 error is no greater than that involved in the use of the average weights 

 before discussed. 



From the above consideration it seems reasonable to conclude that 

 the dietaries express the daily rations of the subjects to within 3 to 4 

 per cent of both the protein and the energ}^ involved. 



RANDALL HALL DIETARIES COMPARED WITH THOSE OF OTHER 



STUDENTS. 



In the discussion of the results of the individual dietaries, they Have 

 been compared with each other and with the commonly accepted stand- 

 ard for a man of sedentary occupation, a basis of comparison selected 

 because it was believed that none of the subjects were occupied m such 



«8ee At water and Bryant, Connecticut Storrs Station Rpt. 1899, p. 76. 

 ^ Ibid., p. 86. 



