570 Tkoday.—On the Capillary Eudiometric Apparatus of 
air with a new and apparently still more accurate form of apparatus. 
Results obtained with the same absorbing solution were remarkably con- 
cordant, but he found constant differences in the percentage of oxygen 
according to the particular solution used. With Hempel’s solution he 
obtained 20-85 %, with a solution made with rather weaker potash than in 
Haldane’s solution 20-938 %, while with Haldane’s solution he obtained 
20-956 % oxygen. He infers that a small trace of CO is produced with 
even slightly weaker solutions as compared with Haldane’s solution. 
Thus it appears probable that the small deficit which my analyses still 
show is due, at any rate in part, to the evolution of traces of CO, for the 
pyrogallate solution was not made up quite according to Haldane’s instruc- 
tions. 1 The error is small, however, and relatively constant with the solution 
used in these analyses. Being constant it is for all practical purposes 
negligible, for it enters into all analyses to appreciably the same extent, 
and is eliminated when changes of composition are determined by difference. 
At the same time it appears advisable to use a potassium pyrogallate 
solution made up accurately according to Haldane’s instructions. 
The following series of seven analyses of a sample of expired air 
(which had been retained in the lungs for some seconds) is a further illustra- 
tion of the degree of concordance obtainable with Bonnier and Mangin’s 
apparatus when used in the way described. 
Analyses of Sample of Expired Air. 
Analysis 
(I) 
(2) 
( 3 ) 
( 4 ) 
( 5 ) 
(6) 
( 7 ) 
Average. 
%co a 
1-44 
1*41 
i -43 
i- 3 B 
1-38 
1-42 
1-44 
i* 4 i 
% 0 2 
19*55 
19-61 
19*55 
19-61 
19-61 
19-58 
i 9'54 
19-58 
% (CO a + O a ) 
20-99 
21-02 
20-98 
20*99 
20-99 
21-00 
20-98 
20-99 
The maximum difference from the mean percentages is + 0-04 %, so 
that a single analysis may be taken to be correct at least to the nearest 
o-i %. 
In the following risumd of the procedure finally adopted a number of 
details are given which have not yet been referred to, but which are never- 
theless important, as they greatly facilitate the use of the apparatus. 
Accessory Apparatus. 
The reagent tubes are fitted with short lengths of glass-rod as handles 
(Fig. 2). Three are necessary for the reagents, and another (empty) for 
drying the capillary. Tubes of the same form are convenient for holding 
the samples of air to be analysed. They can be supported conveniently 
over small dishes of mercury in small stands of the form shown in Fig. 2, 
1 See footnote 3, p. 568, and references there given. 
