113 



This may be also shown in the following manner. The direction 

 of the solntion ' path of /'' is determined by 5 (IV), that of the 

 boundary curve by 16 (IV) and that of the four-phase curve by 8. 



If now between the phases occurs a reaction of group c we get: 



u—^ _yi—y_yx—? 



X — a .t', — X A' J — « 

 From this relation follows : 



M .V — a 



N a; J — tV 

 so that 5 (IV) passes into 16 (IV). 



With the aid of the above relation we readily find from (2) and 

 (3) the formula 16 (VI), with which the above mentioned [)roperty 

 is indicated. 



{To be continued). 



Physics. — "Further experiments with liquid heliiun. H. On the 

 galvanic resistance of pure metals etc. VII. The potential 

 dijference necessary for an electrical current through mercury 

 beloio 4°. 19 K." By Prof. H. Kamerlingh Onnes. (Continued.) 



§ 11. Local nature o f the loss of heat by a mercury thread enclosed 

 in a glass capillary carrying a current, lohen the temperatur^e sinks 

 beloiv 4°. 19 K. While the supposition that the thread should acci- 

 dentally consist of some other substance than mercury for a small 

 part of its length, is in contradiction to the regularity of the poten- 

 tial phenomena, yet on the other hand the supposition that the mer- 

 cury thread has a microresidual resistance similar to the ordinary 

 resistance in Ohm's law (therefore independent of the strength of 

 current, see § 4j, gives rise to no less dilficulties ^). Such a micro- 

 resistance proper to the mercury will be evenly distributed over the 

 whole thread. If we calculate from the potential differences observed 

 during the warming up at low temperatures and the strength of 

 current to which they belong, the resistance of the thread under 

 the conditions of the experiment, then we find that the thread, when 

 the threshold value of the strength of current is only very slightly 

 exceeded, must for a part of its length be partly heated distinctly 

 above the vanishing point. Let us take for example the experiments 



1) Besides those mentioned in § 9, the difficulties here treated also present 

 themselves it we try to explain the potential phenomena by an even distribution 

 of additive mixlureresislance. 



8 



Proceedings Royal Acad. Amsterdam. Vol. XVI. 



