PREDICTION OF THE BASAL METABOLISM OF YOUTH. 203 



basis alone this point seems to be a justifiable place for separation. 

 From a comparison of the figures for the prediction of metabolism from 

 weight for boys in table 33, it will be seen that in every instance the 

 prediction for boys is materially better than that for girls. Thus, 

 with body-weights of less than 10 kg., the error for the boys in the 

 prediction of the metabolism was 8.7 per cent, and for the girls 

 11.8 per cent. Above 10 kg. the metabolism of the boys was pre- 

 dicted with an error of 6.3 per cent and the girls 7.5 per cent. 

 For 15 kg. or above, it was 5.8 per cent for the boys and 6.1 per 

 cent for the girls. It should be emphasized that these differences are 

 not due to an error in the curve (which is a smoothed curve repre- 

 senting all our measurements), but to the more pronounced physio- 

 logical differentiation with children of the lower weights. This is 

 particularly the case with children in the first months of life, a con- 

 siderable number of whom appear on our charts. 



PREDICTED HEAT FROM TOTAL CALORIES REFERRED TO 



SURFACE (GIRLS). 



While our analysis of the prediction of the metabolism from the 

 surface for boys gives little reason to expect a better method of pre- 

 diction for girls with this criterion, it is necessary to analyze carefully 

 the data in table 34 for the heat predicted on this basis. 



The actual differences range from to 148 calories, an extremely 

 wide range; but again reference must be made primarily to the per- 

 centage differences, indicated in column k, for purposes of comparison. 

 With 11 subjects the prediction has an error of 20 per cent or more, 

 but 10 of these are girls with a body-weight of 10.6 kg. or below, 

 corresponding to approximately 0.50 square meter of body-surface. 

 Thus, on the basis of surface as well as weight, the larger children 

 apparently exhibit less errors of prediction by this method of analysis. 



When the actual averages are compared (see table 35), we find that 

 the average deviation of the predicted from the actual measurement 

 is 9.8 per cent, this being the largest general deviation by our 

 methods of prediction, considering the boys' as a whole and the girls 

 as a whole. This error in prediction is also slightly higher than that 

 found on the basis of weight and confirms the deduction made from 

 the values for boys, i. e., that the prediction made from surface is on 

 the whole slightly inferior to that from weight. 



Realizing that the smaller children, i. e., those with the smaller body- 

 surface, show the greater proportion of large deviations, it remains to 

 be seen whether or not any particular area will give the better predic- 

 tion. As with boys, a body-surface area of 0.45 square meter, which 

 corresponds to a body-weight of approximately 10 kg., has been 

 selected as the dividing-line. According to table 35, the percentage 

 deviation for the girls with a body-surface area below 0.45 square 



