( 497 ) 
48 m.m. a screw thread was etched as deep as possible *). On the 
upper end, at a distance of 1 e.m., two copper blocks a, and ag are 
fastened to the glass by means of screws passing through holes. In 
order to avoid uneven tension some leaves of tin-foil were laid 
between the blocks and the glass. To prevent the blocks from turning 
round the screws they rest on the rim by means of two wings. 
Finally to prevent the wire from rubbing against the wall of the 
vessel (see fig. 3) protruding points of glass b were made at the lower 
and the upper end of the cylinder. The wire was wound double, for 
which purpose another projecting point of glass was sealed on at c, 
round which the wire was turned, the ends of this wire were soldered 
on to the blocks, where also the leads were fastened. Hence the wire was 
free from induction which was a great advantage in the measurements 
with the WHEATSTONE’s bridge although on the contrary the conduction 
through the liquid has now a much greater influence and hence the 
bath had to be kept very pure. If the wire is wound on two con- 
centric cylinders, the thermometer occupies more space, but then 
the insulating power of the liquid need not be so great as was 
required for this thermometer. We tried to make the resistance so 
high that the measurements could be sufficiently accurate even at the 
lowest temperature observed; this was attained both by taking a long 
wire and by choosing it as thin (0.1 m.m.) as could be done with 
advantange without making the surface too large in proportion to the 
resistance of the wire. As the numerous (50) turns are distributed 
over the whole height of the cylinder, their resistance expresses as 
well as possible the mean temperature of the bath over a large range, 
if this should not be the same everywhere. 
Four copper leads are used, 2 at each end. This number is 
required in the measurements with the WHEATSTONE’s bridge for 
the entire elimination of the resistance of the leads. Moreover this 
gives an opportunity for measurements by means of the differential 
galvanometer and similar apparatus. 
For the first experiments the leads were 0.5 m.m. in diameter. 
But the variation of resistance of these relatively thin wires some- 
times proved too great in the measurements, so that later copper 
wires of 2 m.m. diameter were taken. The ends of these in the 
bath had been flattened over a length of 11 e.m. to 5 m.m.’s width 
and 0.15 m.m.’s thickness, which prevented the increased conduction 
of heat along them from being of any influence. 
1) For the resistances used later, similar screw threads were ground in the glass. 
