( 669 ) 



model, which yields the connode and the tangent chords, and the 

 simplified construction in the plane given in Suppl. N°. 8, § 7, to 

 which a small correction was applied. After viz., a provisional 

 connode, that of the mixtures taken as homogeneous, has been found 

 by tracing curves of double contact to the t/V-r-curves, anc j nv 

 determining conjugate points b on the gas branch at some points a 



on the liquid branch of that connode, so that every time — is the 



d.v 

 same for the two conjugate points, the lines which join every two 

 of these points a and b, are produced outside the provisional connode, 

 till they cut the isobars which pass through a, in points c, 

 which together represent the required gasbranch of the connode; c 

 and a are then considered as conjugate points. In the if'-surface at 

 5° C. the two constructions yielded fairly well corresponding results, 

 both with regard to the chords and to the connode itself, as appears 



from pi. II fig. 4, where denotes the connode and the 



connodal tangent-chords found by rolling a glass-plate on the model, 



those found by means of the just mentioned construction. 



That the simplified construction, which was more particularly plotted 

 for equilibria far below the critical temperature (see Suppl. N°. 8, § 7) 

 still leads to our end, is probably due to the fact, that we have 

 here to deal with a mixture of the II type. 



With the if.'-surface for 20° the slight depth of the plait rendered 

 it necessary, to considerably diminish the longitudinal scale for the 

 v-coordinate of the model. This compression (cf. pi. I, tig. 2) rendered 

 the plait sufficiently clear to determine the connode and the place 

 of the connodal tangent chords by rolling a glass plate. By means 

 of the simplified construction the connode was still to be obtained, 

 but the determination of the tangent-chords became uncertain. 



With the if>-surface of 26° the depth of the plait (here split into 

 two) becomes so exceedingly slight, that it does not appear but with 

 a computation with 7 decimals, and even then it manifests itself 

 almost quite in the two last decimals. Hence it is not possible to 

 model a if'-surface (we mean a surface derived from the i/'-surface, 

 on which the coexisting phases are still to be found by rolling a 

 plane), on which this plait is visible, nor is it of any avail to confine 

 ourselves to a small part of the surface, because the curvature of the in- 

 curves is very strong exactly there where something important might be 

 shown. The determination of the connode and the connodal tangent 

 chords by construction according to § 8 of Comm. N°. 59«, which 

 can always be carried out provided enough decimals are worked 

 with, remained still uncertain up to 7 decimals, so that we have 



