( 350 ) 
from flowing immediately to the bottom and to diminish the eon- 
duction of heat from below, another little vessel « of compressed 
paper with a wooden bottom has been placed under and around 
sj. The space left beneath the pole pieces has been filled up with 
wool. In the rim t six holes have been made, connected with the 
copper tubes v, four of which lead off the gaseous oxygen, whilst 
two have been shut with a stopper and serve as a safety arrangement. 
In order to obtain the room wanted for these jackets between 
the coils of the electromagnet, it was necessary to considerably 
lengthen the pole-pieces. For weak currents this did not much 
diminish the strength of the field; for strong currents the loss was 
considerable. 
During the experiment all parts of the apparatus except the 
observing-glasses were wrapped in wool. The receptacle a was com- 
pletely filled with liquid oxygen, and as an additional precaution 
liquid was even allowed to flow out until a considerable quantity 
had passed through the holes r and had collected in «. The apparatus 
stood this well; especially the compressed paper, which appears to 
be a very suitable material for this work. 
7. The Hawu-effect at the boiling point of liquid oxygen. 
For the Harr coefficient 2 in a magnetic field of 4400 C.G.5. 
units the value 41,4 was found. Hence the product RM is 182000. 
Before, at a temperature of 10°C., was found to be 11,0. This 
does not wholly agree with the value 10,15, which may be obtained 
by interpolation from the table given in § 3 of this paper for the 
field 4400 and the temperature 10° C. Recently Prrror !) has 
noticed, that the thermo-electric constants of crystalline bismuth 
showed irregular variations with time, which he at first was inclined 
to ascribe to the influence of repeated heating and cooling; this 
however appeared later?) not to be the case. In order to see whether 
perhaps something of that kind really happened in consequence 
of the strong cooling in my experiments, I repeated the deter- 
mination of the HArr-coefficient R at 10°C. shortly after the experi- 
ment in liquid oxygen, and found 11,1. The difference from the 
value of R 11,0, determined immediately before the experiment is 
too small to be worth attention. As formerly I have also not noticed 
continuous variations with time in electrolytic bismuth, I think we 
1) Arch. d. Sc. phys. et nat. (4) 6 p. 105 and 229, 1898. 
| Oe AN OK no mom (4) 7 p. 149, 1899. 
