Freezing- Points and the Freezing-Point Methods. 481 



ascertain the minimum value of K. The velocity with which 

 my thermometer in all these measurements approached the 

 freezing-temperature was in no case smaller than 6, conse- 

 quently K is > 6. 



There is an indirect way of determining K : 



We have : , __ C t g — t' , T > 



C is known ; we arrange the convergence-temperature over 

 the freezing-temperature ; t g is known. We overcool the 

 liquid sufficiently and register t 00 . After the separation of 

 the ice we get the apparent freezing-temperature t' ; K and 

 t remain unknown. After the freezing-temperature observed 

 on the •001°-thermometer has remained for two or three 

 minutes constant, we change the temperature of the ice-bath 

 and with it the convergence-temperature. We read the new 

 apparent freezing-point, when it remains constant for two 

 or three minutes. We have 



C t ' — t" 



<"-«..-&&£. < IL) 



where t , C, K, t ov are the same as in (I.) , t g , t'' } C, t ov are 

 known, t , K unknown. 



From (I.) and (II.) we get : 



or, since t' —t" is very small, 



~ K(t -t ov )' 

 Thus we obtain 



These experiments can be carried out only in narrow limits 

 of temperature, as we soon get to conditions when the freezing- 

 point can no longer be obtained constant. In a good freezing- 

 point method the value of t'—l" is very small ; so that the 

 reading-error, even when it is, as in our case, '0001° or # 0002° 

 for each point, makes a very great part of the total value 

 observed. In Table II. (p. 478) are given such observations. 

 From them we get in a very rough way the result that the 

 value of k is under my experimental conditions about 5 or 6. 

 It may be remarked that we cannot use conditions different 

 from those of the freezing-point method in order to get greater 



