( 816 ) 
The holder and. the torsion spring are, on the other hand, completely 
altered. On account of the smallness of the couple to be measured 
all foreign magnetic actions had to be eliminated as carefully as possible. 
Preliminary experiments showed us that a metallic holder could not 
be used, not only on account of the traces of para-or ferro-magnetic 
impurities that are never absent from workable metals but also on 
account of the difficulty of keeping the surface sufficiently clean; 
this difficulty was encountered repeatedly in Mhe silver ellipsoid 
that we used in our experiments, and it is probable that the 
constant contact of the hands with iron tools plays a part in causing 
it. Glass seemed to be by far the most suitable material both on 
account of the absence of inherent magnetization and of the fact, that 
the surface on account of its smoothness can be kept quite clean. 
The holder which we finally adopted was made completely of 
glass: it consists of a tube b' 5 mm. in diameter that at b\ is drawn 
out to a narrow but thickwalled stem, 0.7 mm. in diameter. To this 
stem the silver ellipsoid was attached; for this purpose a hole of 
sufficient width to fit • was bored along one of its greater diameters 
and the ellipsoid was then fixed at the desired height by means of a 
little wax that completely filled the narrow space between the glass 
and the metal. The tube was then pumped free from air and sealed 
off, so that the liquefaction of the air that it would otherwise contain 
would be prevented. The flat mirror for measuring the angle of 
torsion and the oil-damper were also attached to the holder. 
The torsion springs. On account of the smallness of the couples 
to be measured (the constants of the springs were of the order of 
1200 e.g.s. while those used for the investigation of the ferro¬ 
magnetic substances were some tens of thousands) it. was found more 
suitable to use a straight instead of a helical spring. We took a 
strip of phosphor bronze about 5.5 cm. long \V) and 0.2 ><[0.01 sq. cm. 
in cross-section. The upper end was soldered to a spiral spring of 
three turns made from a much thicker strip than the other; the 
greatest dimension of this strip was horizontal so that in this way it 
fulfilled its purpose of being elastic to tension while taking no part 
in torsion; its presence is essential to prevent the breaking of the 
thin glass stem or of the platinum-iridium stretching wire that is 
soldered to the stem. This stretching wire is made from a platinum- 
iridium wire of 0.1 mm. diameter, which was rolled very thin so 
as to make its torsion constant extremely small without diminishing 
to any great degree its resistance to breakage. The stretching wire 
is fused at b\ to the lower end of the glass stem, and at its other 
extremity it carries a knob c' which is held fast in a ring f \. 
