kept li'ood (I'lriiiii' a \ov\ Ioiiü: time and no or \vv\ liltlc soil \v;is 

 deposited ill liic lid)os. 



Tlie pi-essuro cvliiidei-s and \\\v coniKM'liiifï tubes had been well 

 cleaned beforehand with bciiy/Mie, which was removed by heating 

 while air was stickod (hioiiiiii. The connections were tightened by 

 leather washers soakc^l in paraflin. The |)acking which had to prevent 

 leakage of nuM-ciiry along iIk^ pixot of tiie high pressure cocks through 

 which the mercury passed which was to be forced into the observation 

 tubes, consisted of rings cut from selected S|)anish corks. During the 

 observations the two observation cylinders \\ (m-c disconnected from 

 that where tiie pressure of the glycerine was transferred to the 

 mercury in ordei- to be inde]»endent of leakage that might occur in 

 the i)uin|) or in Ihe glycerine lead'). A perfectly tight tit of this 

 enclosed |»ortioii (Men at the highest j)ressures was secured. 



§ 3. T/ic in('as/ir<'iiirn/ of the ro/inui's. The determination of the 

 normal volume was made in the same way as has been described 

 in Ounm. N". 70 \. These Proc. IV June 21), 1901, p. 107 and in 

 C'onim. N". 7<S, These Proc. Y\ April U), 1902, p. 7(11 , especially the 

 same precautions for the constant tein|)eralui-c and |)ressure were taken. 



The normal volume was determined at least twice before and 

 twice after the measurements. It must \)v recoi'ded, howevei-, that 

 this was not done with the tirsi (luaulity of carbon dioxide of which 

 the isolhermals tVom 25". 55 V. to o7\09 (^ were im'estigated, because 

 the experimenlal tube had broken, ^vllil(Mn the case of the manometer 

 wc may protilably substitute a dii-ccl compai'ison with a standard 

 manometer, lo which c(unparison 1 shall refer later. 



I fonud lor ihc normal \(iluuic of ihe liydroiicii luaiiometer: 



22 Sept. "02: 23.217 cc. 



23.194 „ 

 23.192 „ 

 12 Nov. „ 23.220 ,. 



As the lirst 3 measurements are not made in Ihe halli of c(Histanl 

 temperature 1 haxc in Ihe calculation of Ihe mean assigned the 

 weight 3 to the last delermiualion and ha\"e adopted Ta' ^ 23.210' cc. 

 From the following the advantage of a hydrogen manometer may 

 ap|)ear-). Most of the determinations willi Ihe lirst niixlure (0. 1 

 oxygen in carbon dioxide) were made with an air manometer. During 

 the expei-iments phenomena (tccui'red which pointed lo \'ariations of 



') Small ViU-ialioiis of picssnrc could llicti he aitplicd by scrcwinL;,- sli^^litly in 

 or out Ihc |iivol ol' one ol' the line hiuh |ii'essni'e cocks in Ihe enclosed portion. 

 2) Conip. (Jomm. x\". 'o(J, Tiiese Proc. II June U, 1899, p 29. 



