1002 
But one can still quite well imagine, however, that at higher 
temperatures negative values can be obtained in weaker fields in 
the course of the change which /// as a function of H undergoes 
with the temperature. The part played by admixture would then be 
restricted to a displacement of the temperature at which a negative 
value could still just appear, and this temperature would be higher 
for bismuth of greater purity than for impure bismuth. This would 
be analogous to the diminution of the negative effect at lower 
temperatures in the case discussed in $ 14 of Comm. No. 129c in 
which the axis stands perpendicular to the field. 
At lower temperatures we found the Harreffect positive in all 
fields, which is not what BrcqurreL found to be still the case at 
liquid air temperatures. It is further worth noting that RH shows 
no further change with temperature below the temperature of liquid 
air. This makes it important to amplify the measurements given in 
Table XIII for the axis perpendicular to the field by others at the 
temperature of liquid air. 
It is seen from Fig. 6 that for fields greater than 2000 gauss at 
+30x10° CGA. Tee 
| ae 
| u ale 
Pe oe: Wh rade es ek 
le | | Be 
i 7 6 
| ekal3HK / | 
20 = : EE | + 
r | HAN 
Mn NE | 
15 en bias ~- — +— 
ay ee ee 
10 F ==> = ae on I 
5 fo Me EEE aes 
Eon kel pe Wey Ses Ree 
AK x deo } 
t 0 Vo fe = heet 
NRE EEE 
a 2000 4909 Lave 8000 10003 Gare 
Fig. 6. 
low temperature, and in fields greater than 6000 gauss at ordinary 
temperature, RH is clearly a strictly linear function of the field. If 
we write 
RH =aH tb, 
in this region, we obtain 
T= 290K T= BOK T= 20.3°K 
nee oe a = + 2.56 a = + 2.56 
h' = — 5600 h' = + 1300 . b' = + 1100 
