RESPIRATION 433 



The following scalfe of corrections was used for human blood. 



For mouse blood the corrections used were 50 per cent higher, since 

 the partial pressure of CO required to produce a given saturation of the 

 blood with CO is about 50 per cent higher for mice than for men. 



As already mentioned, the results of duplicate or triplicate titrations 

 of the same sample of blood agree very closely, the variation in the per- 

 centage saturation found hardly exceeding i per cent or 0.5 per cent 

 from the mean. When, as in determinations of arterial oxygen pressure, 

 two samples not differing much in percentage saturation are compared 

 successively with the same standard blood solution, the difference in their 

 percentage saturations with CO can be determined with corresponding 

 accuracy; for any errors due to imperfect preparation of the standard 

 solutions, or to the allowance for dissociation, will affect both results 

 equally. To determine the absolute range of the latter errors we made a 

 number of analyses of definite mixtures of normal blood with the same 

 blood saturated with coal gas. The coal gas contained about 7 per cent 

 of CO, and allowance was made for the small amount of CO present in 

 simple solution in the saturated blood. 



The ox blood used for these mixtures was measured out from a pipette, 

 the blood being kept constantly stirred to prevent sedimentation of the 

 corpuscles. This method, though fairly accurate, is liable to slight 

 errors on account of variations in the quantity of blood which is left 

 adhering to the pipette. The following percentage saturations were ob- 

 tained on different occasions. The same carmine solution was used by 

 both observers. 



In series (2) and (3) the mixtures were made with blood laked by 

 dilution to half, and were unknown to the observer. In (i) and (4) the 

 mixtures were made with whole blood, and were known to one observer. 

 In (4) each observer made up his own carmine solution. 



