130 Dr. M. Wildermann on an Experimental 



In any case, what has still to be said is, that there are 



present subsidiary causes which might in part affect the 



equation and the constant 1*87 : such as dilution-heat put 



forward by Van't Hoff himself, by Nernst, and recently 



by Dieterici and by Evans ; or as the formation of higher 



molecules, which I believe to be present in greater or lesser 



number in all solutions. But the influence of such causes 



could only be slight, and the only relations here quite 



certainly expressed are those given by Van't HofT in the 



0*02 . T 2 

 equation t = — . Besides this I regard it for various 



reasons as very probable that the absolute value of the scale 

 of my thermometer is, between 0°'4900 and 0°'3900 or 0°3600, 

 1-2 per cent, too small. My further investigations will 

 decide this question. In every case Van'tHofFs thermo- 

 dynamical equation has found excellent confirmation in dilute 

 solutions, much better even than most generalizations estab- 

 lished on the thermodynamical basis. 



The results of Loomis (Wied. Ann. li.) obtained with the 

 yJjq thermometer, which show a continuous decrease of the 

 values of /3, may now be calculated, starting from the most 

 dilute solutions (yoo mo ^ normal) (Table IV.). 



The observations of Loomis (made with a yoq° thermo- 

 meter) extend to y^ molecule-normal. The lowering of the 

 freezing-point for a yoq molecule-normal solution of a 

 "non-conductor" is about o, 0187 ; with such a fall of the 

 freezing-point no formation of an ice-cap round the bulb of 

 the thermometer is possible. Here we have taken the j Jq 

 molecule-normal solution as starting-point, and the result is 

 that Loomis's investigations point indubitably to Van't Hoff's 

 constant. In the case of cane-sugar the figure obtained with 

 almost all the more dilute solutions is 187; in the case of urea 

 and alcohol we obtain constants which are too small by 2 or 

 3 per cent. But the fact that these constants are almost the 

 same at all concentrations shows most clearly that Mr. Loomis 

 has not correctly determined the strength of his original 

 solutions — a mistake which does not depend on the method 

 itself. The low values of the molecular lowering of the 

 freezing-point obtained by Loomis, values which for a y-J-Q 

 normal solution are already too small by 10 and more per 

 cent., may be therefore here attributed in the first place to 

 incorrect determination of the freezing-point of water. But 

 the not unimportant fluctuations in the value of b in the cases 

 of alcohol and urea (after the elimination of the error which 

 may depend on the freezing-point of water), which will be 

 noticed in the above tables, shows this, that results obtained 

 with a y<yo° thermometer, in spite of the exactitude which 



