( MH'y ) 



we do not slir vigorously, an unexpected cliango takes place in the 

 disti'ilmtion of the temperature in the bath and hence in the tempe- 

 rature of any measuring apparatus placed in it. In measurements 

 of the kind considered here, we cannot allow such irregularities and 

 fluctuations in the temperature of the hath, either as to time or place. 



Of the various methods of preventing this sudden ebullition, the 

 simplest is the generation of small bubbles of gas by means of the 

 heat of a short resistance (boiling thread). If, however, there are 

 ignitible gases among those successi\ely iulnnbicc^d into tiie a|»paralus 

 and if cojisequently an e.\|)h)sive mixture with aii- might be formed, 

 this method is not without danger. 



To l»ring about cbulHtion a currcnl of gas is often led through 

 the liquid, which. howev(M-, has tlie disadvantage of contaminating 

 the evaporated gas. To avoid this ditHiculty I have led through the 

 bath a current of the gas itself, 'i'his means was applied f(»r instaju*e 

 to avoid the retardation in boiling ij» the vacuum vessel mentioned 

 at the end of ^ 2, and alsc» in order to cause a strong stirring in 

 the bath by means of the current of gasbubbles. Ihit this means 

 also presents many ditïiculties, mostly ai-ising from condensation 

 phenomena in the delivery tube, or higher temperature of the gas- 

 bubbles; I therefore, preferred, tiie arrangement as described in § 4. 



If the cryostat is used as it was intemled to be in Comm. 51, 

 the requirements for very accurate measurements would not be 

 fullilled, even though a uiiitorm temperature throughout the bath was 

 attained. There still remains a systematic regular rise of the tempe- 

 rature, because the gas used is never perfectly pure and the more 

 j)ermanent |)art evapoi-ales first. In cases where measuring ap|)aratus 

 require longer to adopt the temperature of the bath than the time 

 in which the tenq)erature changes the amount permitted by the 

 accuracy of the observation, we cannot reach more accurate residts 

 without additional means, 



§ 4. We iu)w |»ass on to the description of tiie arrangements 

 which form the subject of this communication. The unifoi-m tempe- 

 rature in the bath is obtained by stirring. The stirring apj)aratus 

 is placed concentrically to the bath, thus leaving room in the most 

 profitable way for the measuring apparatus. From this space the 

 stirring ap[)aratus (as in Comm. 27 May and .lunc '1)6 PI. Ill) is 

 separated by a protecting cyHnder |„ (comp. the figure to the left of 

 plate I). The upj)er ring /„j is provided with small valves Xo4 

 covering openings of the same form. If the stirring apparatus moves 

 in the cylindrical space between ^o and B02 the valves shut up 



