568 Thoday. — On the Capillary Eudiometric Apparatus of 
Notwithstanding the thickness of the walls of the capillary, the volume 
of air in the apparatus was observed many times to change in a direction 
corresponding to fluctuations in the temperature of the laboratory; and 
when the temperature was low an increase of volume was noticed, which 
might be attributable to the proximity of the observer. The graduated 
tube was therefore enclosed in a case made of long strips of glass to protect 
it from draughts, and a cardboard screen interposed between it and the 
observer. More recently I have surrounded the tube with a trough of 
water, in order to gain as complete control as possible of any change 
of temperature which may occur during analysis. Even then changes of 
0*1-02° C. were often observed, for which corrections were necessary of + 0-03 
to 0-07 in the percentage of oxygen, or C 0 2 . Over the trough a long strip 
of glass is placed, which serves to screen the trough from the breath of the 
observer, and also as a rest for the lens or other device used to avoid errors 
of parallax. 
Meanwhile, the effect of varying the composition of the pyrogallate 
solution was tried. The solution first employed was that recommended by 
Hempel in earlier editions of his c Gas Analysis and the percentage of 
oxygen obtained varied between 20-5 and 20-7. This solution had the dis- 
advantage that a single film is not powerful enough to absorb all the oxygen 
even during prolonged contact. 1 It was necessary to renew the film several 
times before absorption was complete. It is probable that the conditions 
were especially favourable to the production of CO, for at the first contact 
practically the whole length of the film was exhausted ; and after it was 
renewed the portion which the air first met again was exhausted. 2 Thus it 
is probable that the production of CO accounts in part for the large deficit 
of oxygen obtained with this solution. 
In more recent editions Hempel recommends a solution made up with 
much stronger potash (120 grm. in 80 c.c. water instead of 60 %), and with this 
a series of analyses of atmospheric air gave an average of 20*89 as the com- 
bined percentage of C 0 2 and oxygen. This close approximation' to the 
correct 20*96 % (C 0 2 + 0 2 ) I failed, however, to repeat, getting usually 
about 20-8 %. 
Similar results were obtained with Haldane’s solution, 3 which is much 
more powerful and is stated to produce no CO. A single film of this solution 
nearly suffices for complete absorption, and only requires to be once renewed. 
It seemed improbable, therefore, that the error wasdue to the production of CO. 
1 Bonnier and Mangin remark on the necessity for repeated contact. Ann. d. sc. nat., Bot., vi, 
18, 1884, p. 291. 
2 Cf. Muller, on the necessity of slowing down absorption when much oxygen is present, in 
Abderhalden’s Handbuch der biochem. Arbeitsmethoden, iii, p. 624. 
3 Methods of Air Analysis, 1912, p. 13. Difficulty was found in making a solution of potash of 
the strength mentioned by Haldane, and the solution used was made up with potash of specific 
gravity 1.5 instead of 1.55. Cf. below, p. 570, and see also F. G. Benedict, The Composition of 
the Atmosphere, Carnegie Institute of Washington, 1912, pp. 80 and 111-13. 
