Freezing-Points and the Freezing-Point Methods. 473 



of a '01° or a platinum instead of a Hg-thermometer, &c. is 

 used as long as we do not know what they measure. More- 

 over the study of the registering instrument itself cannot 

 properly begin till a good equilibrium of the heterogeneous 

 system itself has been secured, e.g. many of the so-called 

 " properties " of the Hg-thermometer do not exist at all and 

 come from the incorrect interpretation of the observed varia- 

 tions of temperature. For the determination of molecular 

 depressions in very dilute solutions, a very fine and accurate 

 registration is required. But at very high or very low 

 temperatures, where a "01° or *001° error or more does not 

 come much into consideration, the arrangements for accurate 

 registration can be easily made, and the correct determi- 

 nation of temperature depends almost entirely upon how 

 the equilibrium of the heterogeneous system has been 

 established *. The convergence-temperature being here 

 very far removed from the measured temperature, the 

 observed results can easily be affected by very great errors. 

 If the principles given above be carried as far as possible, 

 very important advance will be made in these regions also. 



Experimental Proof. 

 I. Determination of C and t g in the method I use. 



dt t - *' 



From ±p=C(tg-t) it follows that Ofo-c^log^- 



The arrangements of the experiment, the quantity of liquid 

 used (1250 c.c), the stirring arrangements, &c, &c. have 

 here been the same as in a freezing-point determination. 



I investigated distilled water (the value of C can be put 

 equal for water and dilute solutions). I used for the measure- 

 ments a xoo° thermometer. 



"While measuring the velocity of overcooling, the cork 

 must always be continuously tapped, the stirring must be 

 strictly regulated and the experiment so conducted that the 

 time required for the reading of the thermometer does not 

 influence the obtained results considerably (I made one reading 

 in each five minutes). The reading of the j-^° thermometer 

 has been made to "001°. Since the liquid is not only warmed 

 by stirring but also by the warm air which is injected by it, 

 the convergence-temperature depends not only on the number 



* A more detailed theoretical discussion of the conditions necessary 

 for accurate measurements at high temperatures will be soon given in 

 one of the subsequent articles on the freezing-point method. 



