1268 



which tlie narrow tops had been oul off. A small difference of level 

 on the two sides then sufficed to draw a slow current of mercury 

 through the thin tube. Without this current we could not detect 

 anv deflection when the small lube was heated on eitlier side of 

 the constriction. Now the section of the .mercury in the tube was 

 0.50 mm', of the glass walls 0.88 mm', and in the constriction 0.08 

 ram' and O.JO mm' lespectively. The thermal conductivity of mercury 

 being ten times that of glass, the conduction through the latter could not 

 be of any moment here, as it w^as in Benf.üioks' experiment. Besides 

 we could greatly increase the sensitiveness of our test by having 

 the mercury streaming against the flow of heat. Thus the cold mercury 

 streamed through the constriction and was heated immediately 

 afterwards by a Bunsen flame, which surrounded the bare glass 

 tube. In this w^ay the walls of the constriction were certainly kept 

 cool by the flow of mercui-y, the velocity of wiiich was six times that 

 ill the heated tube. A temperature difference of 250° over a distance 

 of a few millimetres was thus obtained. The galvanometer did not 

 show the slightest deflection while 0.1 mm could have been detected 

 with certainty, the zero being extremely steady in this case. Therefore 

 any effect caused by the temjieiature gradient mentioned should be 

 less than I.IO-^ Volt. 



The above mentioned difference of 1.5 mm for ditferent kinds 

 of contact of the jets may very well be explained without the 

 assumption of a thermo-electric force depending on the temperature 

 gradient. From the quantity of mercury delivered per minute and 

 the diameter of the jet we found its velocity to be 5 metres per 

 second. Therefore the time of contact of the jets was of the order 

 of 10"^ sec, and a small part of the mercury is cooled in this short 

 time from 300° to half this value. It may well be that in this 

 mercury the internal equilibrium between the electrons and the 

 metal is upset, which would cause it to behave thermo-electrically 

 like a different substance. As this different substance would have 

 unequal temperatures at the two ends, a thermo-electric current 

 would be produced. 



lu our opinion this investigation therefore disproves the existence 

 of an effect as described by Benedicks, and there is accordingly no 

 ground for modifying the existing theory of thermo-electricity. 



Physical Laboratory University Groningen. 



