Lerts & Buaxe—The Carbonic Anhydride of the Atmosphere. 139 
The preceding results—and especially those made with arti- 
ficial mixtures—show, we venture to think, that Pettenkofer’s 
method, if carried out with the precautions we have described, ig 
one of greater accuracy and delicacy than is generally supposed. 
Although smaller receiving vessels were employed than were 
used by Pettenkofer himself, and by most of the observers who 
have since used the method—the results obtained by us vie in 
point of accuracy with those obtained by the elaborate and tedious 
methods of Reiset, and Mintz and Aubin, in which hundreds of 
litres of air were operated on. 
As regards the degree of absolute accuracy very few observers 
appear to have tested the point in the various methods which they 
have used, so that it is difficult to compare our results with 
analogous figures obtained by others. 
Mintz and Aubin, as we have already stated, are an exception 
to this rule, as they describe one experiment on similar, but not 
exactly the same, lines as our own, in which they found 98-5 per 
cent. of the added carbonic anhydride, or in parts per 10,000, 
2°98 instead of 3:03 (see p. 118). 
A word ought, perhaps, to be said regarding the correction for 
aqueous vapour present in air. Probably from the minuteness of 
such a correction (relative to the magnitude of the incidental errors 
attending the determinations) very few observers appear to have 
taken it into account. In all determinations involving extreme 
accuracy it is, however, a factor which, no doubt, should not be 
neglected. 
In the eighth series we have calculated for each determination 
the extent of the correction, on the assumption that the air-sample 
was saturated with moisture at the given temperature—an assump- 
tion which is probably not strictly accurate, owing to the greasy 
surface of the paraffin wax which localized the moisture, and also 
to the short interval of time which elapsed between adding the 
absorbent and the collection of the air-sample. But at any rate it 
is the maximum correction. In the succeeding tables we have 
omitted the details of the correction, but give the mean. 
We shall conclude this part of our Paper by giving the results 
of duplicate determinations made on 23 days, with air collected in 
the grounds of Queen’s College, Belfast, in 1897, during the 
months of March, April, May, June, and July. 
We may draw attention to the fact that, with two exceptions, 
the differences in the duplicate samples are in parts per million. 
SCIENT. PROC. R.D.S., VOL. IX., PART II. M 
