326 Messrs. Haldane and Penibrey on an Improved Method 



gravimetric method the bottle method and also the tube 

 method of Pettenkofer. 



The bottle method is in general use in this country and in 

 Germany, and is in many ways exceedingly convenient. 

 There are reasons, however, for suspecting that even in expe- 

 rienced hands it may give seriously inaccurate results for free 

 air. Thus although the French experiments referred to above, 

 and a series at present in progress by our own method, show 

 that the real proportion of carbonic acid is, as a rule, under 

 3 vols, per 10,000, and fluctuates within pretty narrow limits ; 

 yet with the Pettenkofer method a very different average 

 proportion is often obtained, and the apparent limits of fluc- 

 tuation are, as a rule, much greater. 



We may quote a few recent instances. Marcet and Lan- 

 driset* found at two country stations near Geneva 2*7 to 4*9 

 vols, per 10,000, mean 3'76 vols. Carnelley, Haldane, and 

 Andersonf, who titrated with the barium carbonate in sus- 

 pension, found in the outskirts of Dundee and Perth 1*7 to 

 3'5 vols., mean 2'9. FeldtiJ, who also titrated with the car- 

 bonate in suspension, found in Dorpat 1*85 to 3*65, mean 

 2*66 vols. The results of pairs of simultaneous analyses made 

 by him sometimes disagreed by as much as 30 per cent. 

 UfTelmann§, in Rostock, found a mean of 3"5 vols., whereas 

 Schultze at the same place had previously obtained an average 

 of 2*9 vols. Yet these three observers seem to have used very 

 nearly the same modification of Pettenkofer's method. All 

 titrated straight into the absorption-bottle, the carbonate being 

 still in suspension. 



In the following comparative experiments, which were 

 nearly all made within doors, every care was taken to obtain 

 the air from the same place in the room, and at the same 

 time in each pair of determinations. The bottles were 

 pumped full of air when the aspiration through the tubes was 

 half over. 



For the Pettenkofer analyses we used two large bottles of 

 10,680 and 10,800 cub. cent, capacity, with ground-glass 

 stoppers || . Before each analysis these bottles were washed, 

 first with tap-water, and then with distilled water, and dried. 

 They were not washed with acid. The air was pumped into 

 the bottle with a bellows, 100 cub. cent of baryta- water 



* R. Meteorol. Soc. Journ. xiii. p. 167 (1887). 

 t Phil. Trans., B, 1887, pp. 66, 68. 

 X Der Kohlensaih-egehalt der Luft zu Dorpat, 1887. 

 § Archiv fur Hygiene, viii. p. 252 (1888). 



|| India-rubber caps are objected to by Blochniann, Liebig's Annalen, 

 ccxxxvii. p. 39 (1887). 



