Freezing -Points and the Freezing-Point Methods. 465 



K. / (t —if) is unavoidably negative, G(t g —f) must unavoidably 

 be positive ; that is to say, in the ice-melting process, the pro- 

 cess of cooling the liquid must be compensated by the counter- 

 reaction of warming by the experimental arrangements in 



order that ^- can = 0. 

 dz 



When the convergence-temperature is above the freezing- 

 temperature, the temperature of rest of the mercury -thread is 

 not at the real freezing-temperature t but at a temperature t', 

 the apparent freezing-temperature, which differs more or less 

 from t , and the temperature t will be obtained between the real 

 freezing-temperature t and the convergence-temperature t g . 



I shall now, using the equation (A ; ), explain a series 

 of rules for the freezing-point method which have been 

 previously found empirically and communicated in the paper 

 of my late friend P. B. Lewis and in my paper, " Zur 

 Bestimmung des Gefrierpunkts des Wassers,'"' Zeitsch. phys. 

 Chem. xv. p. 538. 



In order that the apparent freezing-temperature t' may 

 coincide with the real freezing-temperature as far as possible 



it is necessary that -^ be kept as small as possible. 



How are we to obtain this by the freezing-point method ? 



(a) is (p. 461) inversely proportional to the quantity as 

 well as to the heat-capacity of the liquid used and directly 

 proportional to the conducting surface of the liquid. 



The first condition is therefore that the quantity of liquid 

 used for the experiment be as large as possible in order that 

 C may be kept as small as possible. Lewis and I used 1250 

 c.c., Jones 1000 c.c., Loomis* 70 c.c., and Abeggf 100 c.c. &c. 

 The second condition is to arrange the overcooling so that 

 the same quantity of the liquid, at the same value of t g —t', 

 be cooled by the arrangements of the experiments as little as 

 possible. For this an air-bath (Lewis, Wildermann, Jones, 

 Abegg) is much preferable to a liquid-bath (Loomis*, and 

 others). The value of C is in a liquid-bath about 20 times 

 greater than in an air-bath. 



(b) K must be kept as large as possible : in the equation 



j~= K(t — i) (the velocity of ice-melting) K is directly 



proportional to the total surface of the separated ice. It 

 follows from this that the overcooling t — t op of the liquid before 

 the separation of the ice must be sufficiently great (Lewis and I, 



* "Wiedem. Ann. vol. li. 



t Zeitschrift fur physik. Chemie, vol. xv. p. 682. 



