' ( 409 ) 



some time that if the plate were moauted between the poles as liad 

 been done until then, the condition was never steady enough to make 

 the measurement in this way; the irregular variations of temperature 

 as compared with the fall of temperature to be measured were too 

 large. This difficulty was almost entirely removed by using a U- 

 shapcd water-jacket placed between the poles of the electroinagnet, 

 through which water was streaming under a constant difference of 

 pressure, so that the temperature remained constant for several hours 

 within a few tenths of a degree. 



The new mounting of the plate may be seen from fig. 2. 



m 



P. = Plate of bismuth. 

 l^l^ If M. ilk T/;. = Thermo-element. 



W. — Water-jacket. 

 K. = Copper tube through which 

 the steam is driven. 



A. B. C. = Grooves for intro- 

 ducing the thermo-ele- 

 ments. 



B. = Piece of wood for clamping 

 the plate. 



iSPi, 



Fig. 2. 



The plate having been put into this jacket, the remaining space 

 was filled with cotton-wool. A test-experiment in which the jacket 

 was closed on all sides with paper did not yield any different results, 

 so that we may assume that no errors arise from currents of air. 



In inserting the thermo-elements in their places we had to secure 

 proper contact on the one hand, and on the other, we had to make 



28 



Proccediugs Royal Acad. Aiuoterdam. Vol. I. 



