6 Chapter I 



described, he made from the same solution, run from the same burette, 

 sixteen oxygen estimations. These were divided into four groups : 

 the average of each group was taken, and the mean of these four 

 averages was taken. The greatest error which entered into Peters' 

 blood-gas analysis was doubtless in the measurement of 3 c.c. of 

 fluid from an ordinary 50 c.c. burette, the meniscus in the case of 

 haemoglobin solution being none too well defined. This error, serious 

 though it appears, is discounted by the fact that the 16 samples for 

 analysis were run consecutively out of the burette. There may be 

 a certain ruggedness in the individual figures, but the averages of the 

 groups of four are very close to one another, since a positive error 

 in one sample entails an equal negative error in the next : in the 

 aggregate 48 c.c. used there is no appreciable error as compared 

 with the 50 c.c. used for the iron analysis. 



Peters' figures for the oxygen of a single experiment are as 

 follows : 



Oxygen in 3 c.c. of solution A in c.c. 



Group (1) Group (2) Group (3) Group (4) 



3667 -3667 -3576 -3527 



3449 -3374 -3640 -3510 



3482. -3439 -3638 -3614 



3455 -3517 '3455 -3537 



Total 1-4053 1-3997 1-4309 1-4188 

 Mean -3513 -3499 -3577 -3547 



Taking the average of these four means we get 



3518 

 3499 

 3577 

 3547 



Total 1-4136 

 Mean -3534 c.c. 



When it is remembered that the whole of these operations, both iron 

 analyses and gas analyses, could be carried through successfully in a 

 day, it will be clear what an advance Peters has made by the use of 

 the modern technique both as regards the concordance of his figures 

 and the certainty with which he has been able to put them forward. 

 His figures for the volume of oxygen per gram of iron are as follows : 



Ox 394, 401, 399 



Sheep 387, 384 



Pig 388 



Dog 384 



Average 391 



