42 



August Krogh and Marie Krogh. 



An inspection of the figures in table 6 shows that the influence 

 cannot be estimated from individual periods because the production 

 of heat has been too varied under the influence of other factors (mus- 

 cular movements etc.). In order to get a result it is necessary to treat 

 the material statistically. We have done this by first reducing the 

 heat produced during each period to a common standard of rest. For 

 each experiment and each group of corresponding periods we have 

 calculated the mean production of heat per kg. and hour as shown 

 in the adjoined table: 



II III IV 



10- (; 1.23 1.302 1.143 



6-2 1.453 1.62 1.63 



2-lOJ 1.72 1.723 1.663 



We have then divided the actual production of heat in each period 

 with the mean of the group and multiplied with the lowest mean avail- 

 able viz. 1.143. The resulting figures, w^hich are strictly comparable, 

 have been tabulated together with the corresponding excretions of 

 N in two groups (Л' in urine below 22.5 gr. and N in urine above 

 22.5 gr.). The means for each group and the corresponding mean 

 errors of the heat production have been computed. 



We find: 



Mean deviation of one 

 determination 



±51 + 73.8 



The result indicates a slight and not very distinct specific dyna- 

 mic action of the protein since the difference between the two means 

 is less than the mean error on the larger of them. In each group how- 

 ever one figure deviates more than double the mean error from the 



