420 RESPIRATION 



more water being also added to the other tube if necessary. The solution I 

 under examination is then saturated with coal gas and the addition to 

 the normal blood solution of carmine is continued until the tints are 

 again equal. To illustrate the method of calculating the result we may 

 suppose that in the first result equality of tint was observed with 1.2 and 

 1.3 cc. of carmine, mean 1.25, and that in the second 6.4 and 6.8 cc. gave 

 equality, mean 6.6; the percentage saturation X is then given by the 

 result of the following proportion sum : 



6.6 . 1.25 



; 7T * ; ' : 100 : -X" 



5 + 6.6 5 + 1.25 



or, more simply, 



6.25 6.6 



100 X X = 35.1 per cent. 



1.25 11.6 



It is clear that the more carmine has already been added to the 

 normal blood solution the less effect on its tint will any further addition 

 have. Hence in approaching the point of equality only o.i cc. is added at 

 a time if not more than 2 cc. have already been added, whereas after 

 already adding 6 cc. it is useless to add less than about 0.4 cc. at a time. 



The titration is repeated with the other half of the blood solution 

 for further safety, and it will be found that apart from accidents the 

 two results will nearly always agree within i per cent of the total satu- 

 ration. This accuracy is very surprising at first sight, since colorimetric 

 determinations have in general a rather bad reputation among chemists. 

 The carmine titration is also no ordinary colorimetric titration, but one 

 in which the quality, and not the density, of tint is estimated. We believe 

 that the bad results commonly obtained with "colorimeters" are due to 

 the two solutions being in some fixed position determined by the apparatus 

 used. An error of 10 per cent or more may easily occur from this cause. 

 Far more accurate results can be obtained with two ordinary test tubes 

 repeatedly transposed, as above described, than with complicated and 

 expensive colorimeters. 



It will be found that the amount of carmine giving equality varies dis- 

 tinctly for different individuals. The proportional difference is, however, 

 the same at the two stages of the titration, so that the percentage result 

 obtained is the same. For the same individual the amount of carmine 

 needed varies, also, with different qualities of daylight, and is usually 

 less towards evening. This does not affect the percentage result, however, 

 provided that the two stages of the titration are completed by the same 

 light. 



All these differences are due to the facts that the two solutions are 

 not spectrally identical; nor is the daylight at different times of day; 

 nor are the retinae of different persons equally sensitive to differences 



