484 Dr. M. Wildennann on Real and Apparent 



It is to be seen from the above table that, if the convergence- 

 temperature is by '1°, '2°. below the freezing-temperature, 

 the difference between the real and apparent freezinq-temper- 



atures is not greater than "00002°— '00004° as far as ,-^-,-— - — —4 



v> (t t ov ) 



is concerned, and so also when the variations of the con- 

 vergence-temperature are in a series of experiments only 

 "1°, # 2°, '3°, the obtained freezing-temperature for one and the 

 same liquid, as well as the apparent and real freezing-point 

 depressions, cannot differ from one another more than by '00002°, 

 "00004°, or '00006°, as far as the equilibrium of the hetero- 

 geneous system is concerned, as C" = about 15. 



The method with the convergence-temperature below the 

 freezing-temperature is to be preferred to that with it above, 

 for the following reasons : — 



(1) The ice-cap need not be avoided. 



(2) Since C" is essentially greater than K, we are able 

 more easily to reduce the experimental error and the 

 variations of temperature, as far as the equilibrium of 

 the system is concerned. 



I will now show how by the use of the results given above 

 to calculate the experimental error in other methods. 1st 

 example : in a given method 70 c.c. liquid and a liquid-bath 

 at — "3° are used, and the quantity of ice present in the 

 liquid at equilibrium is equal to '3° overcooling. Then we 

 have : — The value of C is found for 1250 c.c, about 20 times 

 greater if a liquid instead of an air-bath is used. For 

 70 c.c. C is 2*7 times greater than for 1250 c.c, i.e., C is 

 54 times greater than mine. Again t — t ov is 3 times smaller 



., . T1 , C {t g -t<) . 0-003. 54.0-1 



than mine, lnerelore r^ni. — r~r 1S , . ,-. „ = about 



0°'004 for t 9 -t< = 0-1, 0°-008for tg-l'=0°% &c.,i.e., already 

 about 20 x 2'7x3, or about 160 times greater than mine for 



the same value of tq—tJ. In the same way ttt,; 3 — 3 : = 



J V" (to -tov) 



0°004 for each 0°-l of tg—tj, i.e., the obtained apparent 

 freezing-point of one and the same liquid by repeating the 

 experiment as well as the apparent molecular depressions are 

 already affected by errors of a 004 for each t g — 1 g ' = 0°'l, i.e., 

 160 times greater than in my method for the same value of 

 tg — tj. We see how very easily the error in a freezing- 

 point method may become great. 



2nd example : in a given method the quantity of the liquid 

 used or some other data, which enable us to calculate the values 

 of C, tg — tg', tg — t', are not given. Assume, the only data we 



