CRITICAL DISCUSSION OF RESPIRATION APPARATUS. 259 



we have had a wide experience in the use of various forms of gas- 

 analysis apparatus may have been a factor in acquiring and teaching 

 the technique of this apparatus. 



To be able to place absolute reliance upon the results of the analyses, 

 they must be controlled in some way. The best control of analyses of 

 expired air is the analysis of samples of atmospheric air. Haldane 1 

 points out that such analyses are sometimes used for calibrating his 

 gas-analysis apparatus, as he assumes that the composition of outdoor 

 air is constant, i. e., 20.93 per cent for oxygen and 0.03 per cent for the 

 carbon-dioxide content. Benedict, 2 in studying the oxygen content of 

 the atmospheric air, has found that both the carbon dioxide and the 

 oxygen content are very constant at all seasons of the year and in all 

 parts of the world where such investigations have been made. Wolff 

 and Heele 3 have recently based the accuracy of the gas-analysis appa- 

 ratus used by them upon the constancy of the composition of outdoor 

 air as reported by Benedict. Results of analyses of expired air can be 

 properly taken as reliable when a series of analyses of outdoor air, 

 made under the same conditions, show constancy. 



In this laboratory it has been the practice to control our gas-analysis 

 apparatus with frequent analyses of outdoor air, and when constant 

 results could not be obtained with samples of outdoor air, the apparatus 

 has been examined to find the cause of the discrepancies. In some 

 cases it has been found that the burette was dirty; in other cases there 

 has been a slight leak or the sample has been contaminated with outside 

 air in transit. Unfortunately, we have no method of controlling the 

 analyses of expired air; that is, we have no air that can be analyzed 

 which is both similar in composition to expired air and constant in 

 composition. While analyses of outdoor air may be made and accu- 

 rate results obtained, it is barely possible that the sampling of the 

 expired air may be imperfect and duplicate results still be obtained. 

 Outdoor air has so nearly the composition of any air which may sur- 

 round the apparatus that even if other air were admitted there would 

 be no possible way of detecting it. Notwithstanding these facts, it is 

 strongly recommended that all gas-analysis apparatus be controlled 

 by analyses of outdoor air and that results be obtained in general 

 within 0.02 per cent for either oxygen or carbon dioxide. The values 

 for atmospheric air obtained by Benedict with the Haldane solution in 

 a Sonden gas-analysis apparatus were for carbon dioxide 0.031 per 

 cent, and for oxygen 20.952 per cent in carbon-dioxide-free air. 2 

 Investigators do not, as a rule, publish the results of their analyses of 

 atmospheric air, and when published, they frequently show large varia- 

 tions; these variations must certainly be taken as an indication that 



'Haldane, Methods of Air Analysis, London, 1912, 44-45. 

 2 Benedict, Carnegie Inst. Wash. Pub. 166, 1912, p. 114. 

 3 Wolff and Heele, Journ. Physiol., 1914, 48, p. 430. 



