( 598 ) 
itself at this particular point amounted to 25 kilogausses; it was 
measured from point to point by means of a small standardised 
spherical test-coil. The sensitiveness of the torsion-balance could be 
varied; it was determined in the usual way by means of applied 
additional moments of inertia. 
The furnace consisted of a porcelain tube wound with platinum 
wire and insulated with kaolin powder and asbestos. With a con¬ 
sumption of 1.2 kilowatts a temperature of 1250° was attained, 
which was measured by means of a thermo-element, previously checked 
by observations on the melting-points of tin, tellurium, antimony, 
and gold. 
Test-samples. The great difficulty with all experiments in this 
sphere of work is and always will be the prevalence of iron, 
with its overwhelming ferromagnetic properties, though it hardly 
ever seems to act quite freely. In the case of fifteen elements, their 
binary alloys with iron were examined in Tammann’s laboratory, not 
in the very diluted state, however, which generally corresponds to 
ferrugineous impurities. Of 81 elements, 43 were tested; many of 
them were supplied as pure as possible by Kahlbaum; Prof. Cohen 
and Dr. Hoitsbma of Utrecht kindly placed several elements at our 
disposal; as yet the 10 gaseous elements have not been tested; Li, 
Rb, Cs, Ca, Sr, Ba could not be obtained sufficiently free of iron; 
while Be, Sc, Ga, Ge, Y, Rd, and the rare metals could not be 
procured. Fe, Co, Ni, of course, form a class by themsejves. Dr. M. 
Handa kindly determined the percentage of iron eolorimetrically by 
the Berlin blue-reaction. 
The experimental results, moreover, furnish certain physical criteria of 
their own reliability, for in so far as the susceptibility proves indepen¬ 
dent of the field there can hardly be question of a ferromagnetic ingre¬ 
dient. With about one third of tne samples this was not the case, for the 
susceptibility diminished (in the algebraic sense) with an increasing field 
according to a hyperbolic law. From this Mr. Morris Owen calculated 
the value which would hold asymptotically for an infinite field; 
and, in addition, the influence of the ferromagnetic ingredient, which 
at most amounted to only one sixth — and generally much less even — 
of what could be imputed to the iron in the free state. The thermo- 
magnetic properties also afford a test of purity up to a certain point, 
a few strongly ferrugineous substances show a great diminution 
of susceptibility between 5(XF and 600 \ whilst above 70(P the 
influence of iron hardly need be feared. In no case is there reason 
to doubt that the value of the susceptibility of absolutely non-ferru- 
gineous elements would remain constant, at least within the usual 
