RESPIRATION 407 



or an opal shade covering a gas flame or electric lamp. At every observa- 

 tion the tubes are transposed. This is essential since it will be found 

 that in all probability the tint of one tube will appear deeper when it is 

 held on one side than when on the opposite side. If, for instance, the 

 tubes are nearly equal in depth of color they will appear equal when one 

 tube is on the right or left side, but not vice versa. A slight inequality 

 of this kind is rather a help to accuracy, as probably only one reading 

 will give equality on both sides. With careful work any error in a de- 

 termination should not exceed 0.5 per cent. The method is thus one of 

 great accuracy. 



It is often loosely assumed that colorimetric estimations are uncertain. 

 This is certainly not the case if they are properly carried out, with ap- 

 preciation of the precautions needed to avoid the errors referred to 

 above, of physiological origin. Another common misconception is that a 

 uniform colored surface is necessary, and that, as a tube does not give 

 this, a method such as that just described must be inaccurate. The 

 surfaces need not be uniform, provided they are similar to one another, 

 as in the case of two similar tubes. 



The correctness of a Gowers-Haldane haemoglobinometer can be 

 checked at any time by the ferricyanide method described under A or 

 C. Another check on the correctness of the standard solution is that it 

 must have practically the same pink tint as fresh blood saturated with 

 CO. If there has been any defect in filling, the standard tube will appear 

 yellower. With a proper standard tube one can tell at once by the 

 absence or presence of yellow color whether a patient's blood is free from 

 methaemoglobin or other abnormal blood pigments. 



For ordinary clinical work it is convenient to work ordinarily with a 

 picrocarmine standard tube, and only occasionally ascertain the correc- 

 tion necessary with this standard. The correction can easily be made by 

 comparing the results for the same person and time with the two tubes. 



C. Determination of Oxygen and Carbon Dioxide in Blood by 

 Ferricyanide and Acid 



As mentioned in Chapter IV, a method, based on the use of ferri- 

 cyanide, was described in a paper by Mr. Barcroft and myself in 1902.'^ 

 The principle of this method is that, without permitting any previous 

 contact of the blood with air, the oxygen of a small measured volume of 

 blood is liberated by ferricyanide in a closed vessel, and the pressure 

 produced by the liberation measured without any alteration being allowed 

 in the volume of gas in the vessel. The CO2 is then similarly liberated by 

 ^ Barcroft and Haldane, Journ. of Physiol., XXVIII, p. 232, 1902. 



