( 544 ) 



hence the error of the correction owinp^ to the coefficient not bein^ 

 preci.sely known, nuiy he neglected. 



The vohune in Eh al)0ve Eh^,^, wliicii could not he delerniined 

 \)\ means of mercnry, niav he determined witii suflicient accuracy 

 vohimenomelrically, as we have to <lo with |)roportions of much 

 hirp:er vohimes. 



Measurement of tiie ])rcssure: 



In most cases it will be possible to follow the rulo of makinj; 

 the adjustment of the mercuiy in the \ oliimenometei' at a mark so 

 that the pressure of the gas abo\e it is not less than 0.5 at in. For 

 a measurement of the j)ressure 4 cathetometer rea(hngs are re(iuir('d; 

 lot us put for the probable error of the result: 2X^^-^^2=^^-^''l '"•"- 



1 



this is on 40 cm. : -. Onlv at the volumein)metrical determina- 



10000 



tion of the relation of FJi,_ and what is above it to E}>.^^ the pressure 



was smaller, but this relation, being, as it has been said, oidy used 



as an auxiliary (piantity, need not be so accurately known. The 



same may be said of the measurement of the remainder of the first 



gas in E, aftei- as much gas as possible has been transferred to 



the mixing vessel F. 



Although E had been accurately vertically mounted, so that the 



mercui'v menisci of both Eh^.^ and I'^h^ could be seen sharply without 



altering the focusing of the cathetometei- telescope, it yet appeared 



that the windows PJi„ were not exactly vertical, but that they ail 



deviated from this position in the same direction at an angle estimated 



at 0.2° by means of a level fastened at right angles to a Hat piece 



of steek Therefore as the windows are not placed perpendicidarly 



to the line of visioji of the cathetometer telescope, refraction occurs, 



which causes an a[)parent displacement of the mercury meniscus of 



■ ^i — 1 



<L t. cm., (I being the distance from the meniscus to the window, 



n 



i the angle of the normal on the window and the horizon, ?i the 



refractive index of water. In our case (</ = 8.6 cm., // = 1.38) this 



displacement amonnts to : 0.0075 cm. This gives on atmosphere an 



u 



1 



error of . x\s we have to do with relations ot (luantities to each 



5000 ' 



of which a correction in the same sense would have to be applied, 



the error in the result is smaller. If a higher degree of accuracy 



is wanted, the angle formed by the windows and the vertical will 



have to be determined more accurately, so that the correction may 



be applied. Koi' our pui-pose this was not necessary. 



