A Study of the Diet and Metabolism of Eskimos. 21 



From the quantities of Og and CO^ consumed and produced 

 during a period the corresponding quantities per hour are calculated 

 and also the respiratory quotient. 



3. Accuracy and sources of error in the respiration experiments. 



As the respiration chamber was scarcely completed when we 

 had to start for Greenland we were unfortunately prevented from 

 carrying out any experimental tests at home. We attempted to test 

 it in Greenland but the resources there at our disposal did not allow 

 any very effective tests being made. 



We tried to determine the CO^ and the O^ deficit arising from 

 the combustion of a definite quantity of alcohol, but failed because 

 the combustion was not quite complete and the traces of alcohol vapour 

 present in the air made accurate gas analyses absolutely impossible. 



We produced a certain amount of CO^, inside the chamber from 

 acid and bicarbonate of soda and found it again in the outgoing air 

 with only a slight absolute error, but the quantity evolved was al- 

 together too small for the test to be conclusiA^e. 



We have therefore since our return made some test experiments 

 on a smaller apparatus constructed and worked on the same prin- 

 ciples as that used in Greenland. We have used a box of about 200 

 liters capacity out of which we have sucked a current of air of about 

 17 liter per minute by means of a gasmeter acting as a pump, thereby 

 causing a negatiл^e pressure amounting to 1.2 — 1.4 mm. of water. 

 CO2 was admitted into the box and measured by means of a small 

 meter. The current of CO^ was varied extensively during the ex- 

 periments to imitate as far as possible the actual experimental con- 

 ditions. 



The sampling of the air in the box was performed as in the Green- 

 land experiments with this difference only that the analyses of the 

 air in the box at the beginning and end of each period were done in 

 duplicate because the box was much larger in proportion to the ven- 

 tilation during each period than was the case in Greenland. The 

 mixing of the air produced in actual experiments by the respirations 

 and movements of the subjects was in the test experiments performed 

 by a small electric fan. 



The table shows the presense of a systematic error amounting 

 to about 2.Ö "/0 and accidental errors of 1 — 1.5 "/q. The systematic 

 error must be due to diffusion of CO2. out from the box, while the 

 accidental correspond closely to the unavoidable inaccuracies of the 

 gas analyses. 



The systematic error caused by diffusion in the Greenland ex- 

 periments is unknown but it may safely be assumed to be less than 

 2.5 "/0 and is probably something like 1 ^Jq. 



