164 HEMOGLOBIN 



However doubtful may be the exact nature of the process which 

 accounts for the partition of haemoglobin between oxygen and carbon 

 monoxide, certain properties of the reaction are amongst the best 

 estabUshed facts with which we have to deal. 



Of these we may commence with the shape of the curve which 

 forms the graphic representation of the reaction. This curve is a 

 rectangular hyperbola, as was proved by Haldane and Lorrain 

 Smith (4). But the interesting point with regard to it is that while 

 the evidence in favour of the curve being a rectangular hyper- 

 bola seemed to be very good, the precise position of the curve was 

 a matter of some doubt. 



In Fig. 52 is shown Haldane and Lorrain Smith's cm^e (1897). 

 The ordinate of this curve requires no special discussion. If the whole 

 quantity of hsemoglobin is called 100, the figure at any point indicates 

 the percentage of that haemoglobin which is saturated with carbon 

 monoxide, it being always assumed that the remainder is saturated 

 with oxygen and therefore that no appreciable quantity of reduced 

 haemoglobin is present. The abscissa of the curve requires careful 

 consideration. The figures along it are as given in Haldane and 

 Lorrain Smith's paper and consist of "percentages of carbon mon- 

 oxide in air" — this fact may be grasped without a full understanding 

 of its significance, for it is imphcit in the figures that 21 per cent, 

 of oxygen is present the whole time. The figures therefore are not 

 absolute quantities of CO but, hke those standing against the ordinate, 

 they are really ratios and mean if fully written out : 



CO -05 ^ 



02~'2l' 21'®^°- 



In theory there is an error involved in spacing out equally the 

 figures -1, -2, -3, etc. per cent, of CO in air because if you dilute 

 the air with CO you reduce the quantity of oxygen present, and there- 

 fore if the curve commences at 



C0_^ 



02~21' 



it would end as shown on the paper with 



CO -5 



the air being diluted to the extent of -5 parts in 100. In practice 

 the error is so small as not to matter. But it is possible to draw the 



