A Study of the Diet and Metabolism of Eskimos. 19 



come more or less vitiated by atmospheric air. The sample of air 

 from the chamber was taken from the outgoing pipe, a special trial 

 having shown that an average sample was best obtained in this way. 



The gas samples were analysed by means of the Haldane apparatus 

 which one of us has employed for a number of years. It is read to 

 0.005 "/o and the accuracy corresponds to about 0.01 "/o both for 

 oxygen and carbon dioxide. The analyses of the continuous samples 

 were always made in duplicate and repeated anew if the agreement 

 was not satisfactory. From the samples taken in medicine bottles 

 only single analyses were made. 



During a series of analyses some dirt will accumulate in the burette 

 while the amount of moisture will become diminished. This causes 

 small errors and the first and last analysis of a series of 20 or .30 are 

 not strictly comparable. To avoid serious errors on this account the 

 analyses of the outside air were in most cases taken alternately with 

 the corresponding analyses of air from the sampling apparatus, while 

 the samples from the chamber were analysed in a single series after 

 the close of each complete experiment. 



2. The calculation of the respiratory exchange. 



The data from which to calculate the respiratory exchange of 

 the subjects during a period are: 



1. The quantity of air shown by the meter to have passed through 

 the chamber during the period. This quantity reduced to 0°, 760 mm. 

 and dryness is denoted B. 



2. The volumes of air enclosed in the chamber at the beginning 

 and end of each period are likewise reduced to 0°, 760 mm. and dry- 

 ness, but allowance must be made in this case for the fact that the 

 air is not saturated with moisture. For this purpose the psychro- 

 metric table in Landolt Börnstein: Physikalisch Chem. Tabellen was 

 employed. The reduced quantities are denoted С and D respectively. 



3. The percentage of 0^,, CO^ and N2, in the outgoing air (B). 

 The results of the duphcate analyses are calculated to three decimal 

 places and averaged. The percentages are denoted Oß, Cß and riß. 



4. The percentages of O^, CO^ and N^ in the outside air 0^, 

 c^, Hj^. To find these, curves are constructed from all the analyses 

 and smoothed graphically. The percentages to be used in each period 

 are taken from the curves ^. 



5. The percentages of the gases in the samples from the chamber 

 are denoted o^, c^, tic and o^, c^, and Uj). 



^ The variations observed in the composition of the atmospheric air are small 

 (not exceeding 0.02 during one experiment) but even as they are they may very 

 possibly be unreal. 



2* 



