If from tlii.^ wc iiilci-polaU' in tlic same n\ a v as in § o wc liiid that 



1 = l.()l() lor tlic (lcii--il\ 225.50, to w liicli heloims tlic xoliime: 

 dTj, • ' 



().(K)443. 



This \ahic docs not a,u-i'ee with tiuit derived al>ove from the densities. 

 The diJfereiicc is hiruer than can he aHcribed to the errors in tlie 

 ohsci-vation. The deviation is in the same sense as foUows from 

 Ama(tAt's experiments. The fohowing- might serve as an exphmation 

 of til is (hlfercnee : 



Aceordinji' to a remai'k of I'rof. I\A.MKRLiNr.n ( )nnks from whom a 

 new paper on the cause of the (h^xiations near the ci-itical point may 

 l)e expected (eom[i. ("omni. X". 74, Arcli. Xcerl. série II, t. A I, [>. <S<S7) 

 the appearance of tlie mist in the nci,ulil)onrhood of fjie critical 

 point seems to indii-ate that a [)ai-t of the snbslancc condenses round 

 numci-ons centra e([uallv distriltuted o\ cr the \N"iiolc s[)ace. That oidv 

 in this area this appears so distinctlv, niijiht lie ascribed to the cir- 

 cnmstance that here small forces are snlKicient to cause great variations 

 of dcnsitv. These con(h'Usation-> niiiiht ha\e a perceptible influence 

 for instance on the satni-alion \-olume although the \ariation of the 

 j)rcssure would not become perceptible in conse(pience of it. 



In a comparison of mixtures with pure sul)stances accordiiig to 

 the \a\y of correspoiKbug stales we must disregard those conden- 

 sations, as ihe\ do no! occur at corrcs[»onding points (in mixtures 

 near the piaitpoiut . Hence our jmrpose will ])robabl\- be served best 

 if \\(' adopt foi- the critical xobiuie; ().()044o, as tliis \alue according 

 to a thermodviiamical rehiliou roull.-- from deleriuinations not so iiear 

 to the critical point, and where therefore parlicuhir plieuomeiia which 

 occur in its immeibale neighbourhood havi' pla\ ed no pai't. 



