reduced to a small aperture in tlie hlow-tlame. C is a capil- 

 lary joined to half a metre of rubbei' tnhiiig, which leads to A'. /), j^, i'"' 

 form (ooether a second glass [)arl ; the lower end of F is joined 

 through another half metre of rubber tubing to G, which is bent 

 three times up and down. At P a platinum wire has been sealed 

 into the glass tube. 



Half of bulb .4 and all the rest of the glass apparatus was filled 

 with mercury distilled in a vacuum. A stout rubber pressure tube 

 was slid over the narrowed top of A and joined at the other end 

 to a reservoir in which air had been brought to a pressure of 2 to 3 

 atmospheres. The arrangement described served the double purpose 

 of preventing heat being carried by conduction from the heated tube 

 H to the platinum wire I\ and also procuring an easy way for 

 i-etilling bulb A with mercury. 



For each merciwy jet a similar apparatus had been prepared. 

 The two parts (t were tied together and placed in a vessel with 

 water. (48 X 24 X 20 cm.). To the platinum wires /-* were soldered 

 the ends of flexible wires covered with rubber and silk, leading to 

 the galvanometer, the junctions and the bare platinum and copjjer 

 ends being thickly coated with shellac. Two metres of the leads 

 were immersed in the water, thus prevenling any conduction of heat 

 either through the mercury or through the copper, and securing 

 perfect equality of temperatuie of the two mercury-platinum-copper 

 junctions. The resistance of the connecting wires, the tubes with 

 mercury and the mercury jets was 1.5 ohms. The galvanometer 

 used was one of the TnoMSON-type, made by Carpentier, in which 

 the original astatic system had been replaced by a system according 

 to DU Bois-RuBENs, made by Siemens and Halsk!*:, suspended by a 

 quartz fibre of 7 jr The coils had a resistance of 2.7 ohms; the 

 scale-distance was 2.8 metres; the magnifying power of the reading 

 telescope was 33 times. 



By adjusting the directing magnet the sensitiveness was raised to 

 a deflection of i mm for 5.8 ,- 10^^ ampère ; the total resistance 

 being 4.2 ohms, 1 microvolt produced a deflection of 4J mm. With 

 this sensitiveness, however, we could not make any measurements 

 till after the cessation of the tramway traffic, but then we could 

 trust the reading to within 0.1 mm, unless unusually large fluctua- 

 tions of the magnetic declination occurred. 



One of the apparatus ABC was attached to a solid stand permit- 

 ting slow displacements in three perpendicular directions. 



In order to collect the mercury the nozzles were placed in a short 

 vertical glass tube of 4 cm. diameter, two vertical slots allowing 



