562 Mr. 8. C. Laws on the Thomson Lffect 
required length and diameter, sealed at one end; these were 
heated to about 240° C. ina bath of oil, and the molten metal, 
which had been previously kept liquid and well stirred for 
some hours, poured in and allowed to cool slowly. By cutting 
with a diamond and gently tapping, the glass was removed 
from the rods; these were then annealed by placing them 
in a.wide glass tube and heating in an oil-bath to a tempera- 
ture as near the melting-point as possible and very slowly 
cooling. 
Owing to the greater expansion of bismuth on solidifying, 
glass tubes could not here be used as moulds, for the metal, 
at the point where it first began to solidify, expanded to such 
an extent as to crack the tube, thus allowing any metal that 
still remained liquid to escape. A satisfactory mould which 
did not introduce impurities into the metal was made from 
slate. The mould was obtained in two pieces by clamping 
together two long slabs of slate, rectangular in cross-section, 
and drilling a hole of the required diameter along the length 
of the block with its axis coinciding with the line down the 
centre of the plane of junction. The walls of this cylindrical 
space were carefully polished with emery-powder, and the 
mould placed inside a cylinder around which a coil of german- 
silver wire was wound. 
By sending a current of about 4 amperes through this coil, 
the mould could be heated to about 250° C. before the liquid 
metal was poured in. When the meial had solidified the two 
halves of the mould were separated and the rod withdrawn 
and annealed as before. 
Ill. Description of the Apparatus. 
To increase the magnitude of the effect to be measured and 
also to eliminate as far as possible errors due to the want of 
homogeneity of the material, two rods were used. 
These were placed parallel to one another at a distance 
apart of 12 cms., with one end of each passing through a 
rubber stopper into a bath in which water could be kept 
boiling, and the other end in a bath which might contain 
melting ice, or through which a current of cold water could 
be circulated. 
The ends in the hot baths were joined by a copper rod so 
that a current could be sent through the two rods in series. 
To the other ends of the rods leads were soldered, and the 
circuit completed through a battery of accumulators B, 
adjustable resistance R, ammeter A, and reversing key K, 
figs aL), 
: ito measure the changes of temperature, thermocouples 
