Iwo or Ihree analyses of air per day. With varying ([uantities of 

 CO2 in Ihe air, an air-tesl must be made simultaneously Avilh every 

 estimation of assimilation. 



During the summor of 1920 Ihe normal carbon dioxide ])ercentage 

 j)er lilre of the air around Ihe Kcological Station was on the average 

 as follows (18" and 760 mm.), 



22 . VI— ö . VII : O.r.T mg 



8 . VII— 15 . VII : 0.5H mg.. Lovvest (X)o percentage : 0.4i mg.tS . VII) 

 16 . VII— .'U . VII : 0.008 mg.^Highest y> » : O.72 ^ (:\ . \X> 



1 . VIII— ;iO. VIII: O.oismg 



A higher CO 2 percentage Ihan that of the air was ohtained by 

 introducing cerlain (juantities of pure carbon dioxide into the closed 

 apparatuses. The carbon dioxide \vas taken from bottles of arti- 

 ficial Vichy water. The measuring and ])umping in of the gas was 

 performed över (juicksilver in a sort of gas-burette with a capillary 

 starting-tube, which was connected with the assimilation-chamber 

 by means of a fine-calibred indiarubber tube d. 



Numerous test-experiments convinced me that the carbon dioxide 

 used was perfectly pure and only saturated with aqueous vapour. 

 In the calculation of the amount of CO„ introduced into the aj»- 

 paratus the tension of the acjueous vapour must therefore be de- 

 ducted. 



The repeated raising and lowering of the quicksilver vessel of the 

 burette, after the introduction of the carbon dioxide, ensured not 

 only that all the CO., was washed out of the burette, hut also that 

 the air in the assimilation-chamber was properly mixed. 



The degree of accuracy given by the apparatus is determined 

 almost solely by the error in reading on pi])etting and titrating, for 

 all loss of CO.^ by dilTusion or by incomplete absorption isexcluded. 

 In order to prevent as far as possible the adsorption of CO^ to the 

 walls of the api)aratus^ these. were coated with paraffin. The rubber 

 tubes and the stopj)ers Avere carefully smeared with vaseline. The 

 small constant error arising from the rinsing and from the carbon 

 dioxide of the air in Ihe titration-flask was eliminated by deter- 

 mining the titer of the baryta solution after it had been |)ut into the 

 apparatus and sucked out again. 



The error in titration varies with the strength of Ihe solutions. 



' As BoRNEMANN too has found (1920, p. 55), CO^ is strou^ly adsorbed to me<al> 

 (rp. also Wii.i.stXttfr and Stoi.i. 1918, p. 280). 



