SPECIFIC OXYGEN CAPACITY 55 



The theoretical accuracy of the ferricyanide method was regarded 

 at that time as proved by the researches of Haldane, confirmed by 

 those of Franz Miiller in Berlin (13). Certainly these authors proved 

 it not to be grossly inaccurate. Recent research has shown a certain 

 source of error in its use for the determination of the oxygen in blood (14 ) . 

 The fats in the plasma react with the ferricyanide and use up some 

 oxygen in the process. This effect in normal blood is very slow, hence 

 in a method as rapid as the differential method of oxygen estimation 

 which Peters used, this error is scarcely perceptible, though in the 

 case of anaemic and consequently lipaemic animals it becomes very 

 appreciable. By using the differential method Peters had no difficulty 

 in doing half a dozen determinations with as many cubic centimetres 

 of blood in two hours. Even had Peters used blood — not a suspension 

 of corpuscles — he would have been at a great advantage as compared 

 with his predecessors, whose individual analyses, if theoretically 

 somewhat more accurate, extended over two or three days, and 

 entailed the use of large quantities of haemoglobin. It was impossible 

 in their case to obtain large numbers of analyses from which the 

 errors could be eliminated to some extent by statistical treatment. 



In estimating the iron by titration with titanium advantage was 

 taken of new methods of analysis which were much simpler and more 

 accurate than the older permanganate titrations. 



The theory of the titanium method of estimating iron is represented 

 by the following equation : 



Tiag + FeClg = TiCl4 + FeCLj. 



The method has this great advantage over that of titration with potas- 

 sium permanganate, that it is not vitiated by the presence of chlorides. 

 In practice defibrinated blood was centrifugalised, the corpuscles 

 washed twice in isotonic salt solution and as much as possible of 

 the fluid removed. The cream of corpuscles was laked by twice its 

 volume of dilute ammonia (4 c.c. of strong ammonia per litre). This 

 solution, which we shall call solution A , was centrifugalised again to 

 rid it of any corpuscles which did not lake and of other debris. 

 Portions of it were then measured out of the same burette both for 

 the iron estimations and for the oxygen analyses. For the former 

 50 c.c. of the solution were evaporated in a platinum crucible and 

 carefully "ashed." It was at this point that the nicety of the deter- 

 minations really entered, if accurate results were to be obtained. 

 The "ashing" must take place at a temperature which is neither 



