v EG A -EXPEDITION EKS VETENSKAPLIGA ARBETEN. 



263 



•ducted very slowhj from the upper part downwards, until the 

 whole mäss is transformed into a cylinder of ice. The ten- 

 dency of the ice to form at the surface of the mercury, which 

 on account of its superior conductivity is cooler than the 

 glass, must be prevented by keeping the nether part of the 

 instrument isolated with cotton. * The operation is best per- 

 formed on a cold winter day, by placing the instrument in a- 

 snowdrift or exposing it to the free air. The freezing process 

 is introduced from above, by surrounding the capillary tube 

 with snow and a few crystals of CaClo + Caq. Ordinarily the 

 formation of the whole cylinder lasted 6 to 8 hours. This 

 cylinder was without a single bubble of air and entirel}^ trans- 

 parent, if the instrument was filled with pure water - and all 

 operations correctl}^ performed. The ice from salt water is 

 somewhat opaque and shows a radial structure. 



Afterwards the dilatometer is either dipped for a moment 

 in water of + 10^ or + 15° C or warmed by the hand, until 

 the outer layer of ice, nearest to the glass, is melted, then the 

 stopcock d is opened. The pressure of the mercury makes a 

 thin film of metal ascend between the glass and the compact 

 ice cylinder in the reservoir, which will thus be surrounded 

 hj an envelope of mercur}'. The melted ice leaves the instru- 

 ment by the capillary tube. If this operation is conducted 

 with address, the ice will be almost completel}' enveloped with 

 a metallic rind and only be visible from without as a narrow 

 stripe, wliere the ice cylinder touches the glass, (see plate 20). 



Finally the stopcock d is shut and the dilatometer placed 

 in an iron vessel, filled with mercury. An iron hook is fixed 

 in the bottom, which keeps the reservoir and a part of the 

 stem and the capillary tube of the dilatometer immersed in 

 the fluid, which is constantly stirred by an iron paddle. A 

 thermometer is placed beside the reservoir, and a calibrated 

 scale-tube is joined to the dilatometer at c. Outside the vessel 

 a cooling mixture of snow and NaCl is applied, wdiich makes 

 the temperature of the mercury sink very slowly. 



When an observation is to be made, the regular sinking 

 of the temperature in the mercury vessel is compensated by 



* This part of the experiment requires the constant watchfuhiess of 

 the experimentalist. 



■" In this case the danger of bursting is infinitely greater than other- 

 wise, on account of the extreme hardness of absolutely pure ice. Ice from 

 salt water is much weaker and seldom causes the bursting of the glass 

 reservoir. 



