651 
Fig. 1 and 2. 
85 
80 
(5 
2500 _ 5000 
=< op 
The measurements by Dewar and FLEMING give the largest results for 
w! 3 WE30 K 
- and at the same time the smallest results for — — They 
to 2730 K 
were probably obtained with extremely pure material. Braku worked 
with a large number of different bismuth wires. One of these, labelled 
! 
a 
5b, gave a larger value for — than the others for which he gives 
w 0 
a mean value. The wire with which I worked gives exactly this 
