262 . PETTERSSON, ON WATER AND I C E. 



denote the average coefficients of expansion of mercury and 

 glass. 



In the case, that q should be experimentally deterinined 

 with exactness for temperatures below zero, all the elements 

 are at hand for a recalculation of the following results. 



B. Measurements of volume. 



These are made by means of the dilatometer represented 

 by fig. 2 on plate 20 in the following way, The reservoir B is 

 filled to ^'3 with mercury and cautiously heated, until the air 

 adhering to the surface of the glass, in contact with the mer- 

 cury, has escaped. When the last traces of air are removed, 

 and a real ebullition of mercury seems to begin, the stopcock 

 d is closed and the flame allowed to sweep the upper part of 

 B. The pressure of the heated air and the mercury vapour 

 makes the hot mercury rise in the capillary stem of the dila- 

 tometer and flll the side tubes b and c and the upper reser- 

 voir A. The volume of the mercury, which remains in B, 

 must not exceed Vi o of the space of the reservoir. Then the 

 stopcocks a, b and c are shut and the flame removed from 

 B to C, the water of which is heated to boiling. The stopcock 

 d is opened, and a jet of hot water passes from C to B, on 

 account of the rarefied air in the reservoir. When B is half 

 filled with water, a separate flame is applied to it, and the 

 water in B and C is assiduously boiled, until every trace of air 

 is removed. If B is allowed to cool, it is rapidly filled with 

 air-free water from C, which is still kept boiling. When the 

 temperature of the water in B has sunk to about 30° or 35'' C, 

 the capillary tube of the dilatometer is made to dip in the 

 mercury, which covers the bottom of C, and then, on farther 

 cooling, a tiny column of mercur}^ enters the capillary tube 

 in the manner shown in fig. 2. The stopcock d is shut, a is 

 opened, and the operation of filling the instrument with water 

 is completed. 



If the dilatometer in this conditiou is combined with a 

 calibrated scale-tube, the whole dilatation of water, from its 

 boiling to far below its freezing point, can be easil}" deter- 

 mined, but if also the change of volume ht/ freezing and the 

 dilatation of the frozen water must be ascertained, which was 

 the case in my experiments, another preparatory operation, 

 which requires the greatest caution, is necessar3^ 



The freezing of the water in the dilatometer must be con- 



