124 Scientific Proceedings, Royal Dublin Society. 
part of the glass siphon tube. Figure 2 (p. 123) explains the 
arrangement attached to it, for drawing off a charge of 50 c.c. in 
such a way that the air taking its place was first freed from car- 
bonic anhydride. The mode of using the apparatus will be 
apparent ata glance. The stock of solution having been intro- 
duced into A and its rubber-cork H, and all attachments fastened 
in, the corks J and J (also of indiarubber) are removed, and the 
siphon tube Z filled with the solution by suction. ‘The pinch-cock 
G is now attached, and the cork J fastened in the position shown. 
Before filling the pipette C with a charge of the solution, we usually 
washed it out with a little of the liquid. ‘The use of the soda 
lime tube Bis obvious during the charging operation. It is of 
course corked when the apparatus is not in use, and also closed at 
the other end by the pincheock G. The whole apparatus is con- 
veniently kept on a somewhat high shelf. 
The standard solutions.—These were made as dilute as was com- 
patible with the delicacy of the colour indicator employed in the 
titrations. In all our determinations the indicator used was 
phenol-phthalein, as in addition to its extreme delicacy, it pos- 
sesses another important advantage, viz. that the colour effect to 
be observed is no mere change of tint (as is the case with turmeric, 
litmus, methyl orange, &c.), but the production of colour in a 
colourless solution, or vice versd. Hence this indicator is charac- 
terized by its sharpness. By employing acid and alkaline solu- 
tions of equivalent strength, and made so that 1 c.c. of either cor- 
responded with 0:1 ¢.c. carbonic anhydride at N.T.P. and by only 
using a small quantity of phenol-phthalein, we found that the 
“end” reaction required an addition of less than {5 ¢.c. This 
end reaction was, in our determinations, the final disappearance 
of the faint pink colour which remained when the baryta had 
been nearly neutralized by acid. As the total quantity of 
carbonic anhydride to be determined was usually about 1 e.c., 
it is obvious from the above that the limit of accuracy in 
titrations with solutions of the above strength was about +45 ¢.c. 
of carbonic anhydride or 1 per cent. of the total amount to be 
determined. 
Experience has shown us that these solutions are very suitable 
for the determinations as we conduct them, and they were em- 
ployed in most of our experiments; but in a few cases where the 
