G. D. Barnett 269 
matical effort. Andsoon. It would no doubt be possible to 
evolve many other forms equally applicable. The following table 
summarizes the comparative merits of the Ambard type of formula 
and the three modifications suggested. 
No. of errors 
greater than 10 
Formula. . Average deviation. volumes per cent. 
eet vol. per cent 
9 Lek hee 
COz = 80 — va VC 6.99 15 
D 
COs, = 80 — 5 y2 5.45 11 
CO; 8050.2~74D 5.85 , 12 
CO, = 80 — 0.9+/NH; 6.35 Hii 
While it is thus apparent that many formulas can be derived 
which show a certain degree of applicability to the facts of acid 
excretion, it does not at all follow that they are to be regarded 
as true quantitative relationships. Certainly before we accept 
any formula as such we must justify it further than by the mere 
statement that ‘“‘the margin of error to be accepted in this in- 
stance appears to be about 10 volumes per cent of plasma COs.” 
For certainly if the relationship be a true one this margin of error 
must bear some fairly close relation to the deviation produced 
by possible errors in the original observations. Suppose, then, 
in the original formula above, we allow an error of 5 per cent in 
determining urinary acidity, a similar error in determining am- 
monia, 2.5 per cent error each in measuring urinary volume and 
the weight of the individual, and 6 per cent error in determining 
plasma bicarbonate, all of which are much greater than the errors 
of the methods involved. If such maximal errors occurred in 
every urinary, weight, volume, and CO, determination in the en- 
tire series and always occurred in such a manner as to make 
their effects additive, the error deviation of the calculated from 
