}fitfi(al S)luhili(ij of Li<pti(t;, 639 



alcohols and t\\c p-f tlia«>niin illastratino- these cases*, it will 

 bo soon that there the branch of the critical curve which has 

 its or;ii;in at the critical point of the alcohol ends in the lower 

 critical point o£ the licjuids which these mixtures possess. 

 This same branch is bound to have a different character for 

 the mixtures with methyl-alcohol, owing to the non-existence 

 of a similar lower critical point. 



The experiments were conducted in the following manner : 

 the compression-tube was charged with some methyl-alcohol 

 and vapour-pressures and critical point were determined. 

 Increasing quantities of ethane were then introduced into the 

 tube, and each time similar measurements made ; or the tube 

 was first filled with ethane, a certain quantity of methyl- 

 alcohol was added, readings were taken, and small quantities 

 of alcohol let out at the bottom, each time a set of determina- 

 tions being made. The number of experiments required for 

 an exhaustive survey oF the phenomena is naturally very 

 large, but fortunately at the lower temperatures it appeared 

 to be possible to use one and the same mixture for the deter- 

 mination of a whole set of critical points, a fact which was 

 already mentioned in our previous paper f ; propftrly speaking, 

 one mixture can have one critical point only, but the compo- 

 sition of the critical phase in the liquid state varies so slowly 

 with change of temperature, that a given mixture shows the 

 critical phenomena at a whole range of temperatures, not 

 always at the middle of the column, but at the lower tempera- 

 tures iu the upper half, at the higher temperatures in the 

 lower half of the tube ; for further explanation we refer to 

 former publications on the critical phenomena of mixtures. 



We confine ourselves to the publication of the critical data, 

 the results relating to the saturation-curves having no parti- 

 cular importance. Our general experience with regard to 

 the critical mixing-points of liquids is that a great consistency 

 in the numerical results is not easily obtained ; slight admix- 

 tures of air or other impurities which are not wholly to be 

 excluded and naturally vary in amount appear to have a 

 considerable influence on the critical pressure at a given 

 temperature, or vice versa, much more so than in measure- 

 ments of the critical point between vapour and liquid in 

 ordinary mixtures. With other fillings than ihase to which 

 the following results refer equally smooth curves of the same 

 general character were obtained, but the pressures at the 

 lower temperatures were sometimes as much as five atmo- 

 spheres higher than those in the Table I. For publication we 

 * Kuenen & Robson, /. c. p. 100, fip-. 8. 

 t Kuoueii X: llob.=*oi), /. c. p. U)I». 



'21: '2 



