DEDUCTION FROM Tlfl'; GASEOUS THEORY OF SOLUTION. 297 



water, so as to leave the same (iiiantity of aniline distributed tlironj^li- 

 out a smaller spai^e. And what will be the lesnlt of (l()in.ij;- this while 

 temperature is kept ecnistant .' Evidently, as in the other case, the 

 point a (Fig. -) will travel to the left, aeross lines of increasing 

 osmotii! pressure, until it reaches h — that is, until the solution is a 

 saturated one; and alter that, if more water be abstracted, some of 

 the aniline will l)e thrown out or condeused, not as [)ure aniline but 

 as a saturated solution of water in aniline, so that two layers will now 

 coexist — the aniline in one- having the specitie solution volume repre- 

 sented at />, and the aniline in the other having that r(;preseuted at c. 

 This transference from h to c, will continue, as water is abstracted, until 

 the ratio of residual water to aniline is Just enough to give the whole 

 of the latter the specitie solution volume shown at c. At this stage 

 the water layer will disap])ear, and only a saturated solution of water 

 in aniline will be left; and after that only a very small volume change 

 can possibly result from further abstraction of water, as the specific 

 solution volume is already not far from the si)ecific volume of i)ure 

 aniline itself at the same temi>erature. 



To complete the C:>mj)ai"isou of the two curves, let me jtoint out that, 

 Just as we can from Fig, 3 calculate the distribution of alcohol between 

 its li(|uid and its vapor layers under givesi conditions, so can we <'alcu- 

 late from 1^'ig. 2 the distribution of the aniline between the aniline 

 layer and tiie water layer under given conditions. In the former case, 

 if the total volume of a tube containing ii grams of alcohol, at, say. 

 230^, be )t X ./', and if .r be mai'ked olf (Fig. o) between h and c on the 

 line of that temperature, then (.r, />, and c standing for the volumes 



wliich can be read off on the horizontal base line) )i . , is the 



h — c 



weight of the alcohol in tlie vapoi- lav<'r, and ii . , is its weiiiht in 



the li(|Uid la.N'er, and the \-olnnies of the two lavcrs in cubic cent'- 

 metersarey/ .!>.', ~ and ii . <■ ' resi>ecti\('l\', which arc together 



e(pnd to // , .V. .lust also witli the anilim^ and watei' mixture (Fig. 2), 

 If II X •'■ i>e the total volume of tin; mixture (both layers together) 

 containing // grains of aniline, at, say, 140^, and if .r be nKii'ked off 



as it was in the (»ther case, then ;/ . , is the weight of aniliiu' in 



II — c 



h - X 

 the water layer, and " • / _ .is its weight in the aniline iaxcr. an<l 



the total volumes (»f the two layers are n . h . , and n . c . , 



— c h — c 



respectively, together equal ii . .v. 



If the actual Aveights of aniline and water in the mixture be given, 

 the value of x can be calculated with a very fair approach to accuracy 

 by the iiu'thod adojtted in ])lotting the curve; and thus all the facts 

 with regard to the <listribution at any tem[)eratur(^ can l)c obtained. 



