ON THE ELECTRICAL PROPERTIES OF PURE SULPHUR. 
143 
Auxiliary Condenser .—This is necessary to swamp the effect of the changes in 
capacity produced by putting in or taking out the sulphur. YVe used ten “ gallon ” 
Leyden jars in parallel with the guard-ring condenser. 
Transformer .— It appeared from the experiments of one of us and Mr. Allen, that 
of all the methods of excitation tried, nothing was so steady as a large coil used as a 
transformer, the alternator being excited by a small current only, and driven by rather 
loose belting from a well loaded gas engine which has a short period of irregularity. 
The alternator has a very massive field-magnet system which rotates, and this acts 
sufficiently well as a fly-wheel to smooth most of the irregularities of the gas engine, 
when these are short in period. Some irregularities, however, remain and impose the 
limit to the accuracy attainable. It was found best to have a weak field on the 
alternator, and use all the armature coils, instead of strongly exciting it and using a 
few armature coils in series with resistances. 
This was contrary to expectation. The induction coil could also be used with its 
break, or with a clock-work break. Mr. Allen found, however, that these were not 
so steady, and the same remark applies to a number of attempts made to use a 
Wimshurst machine driven at a constant speed, and having one terminal permanently 
to earth, and the other earthed by a high resistance vacuum tube, with a bit of caustic 
potash in it. An irregular excitation will very quickly spod the knife edges of the 
balance. The frequency of the alternator was about 60, and we got better results with 
this than when the frequency w r as less. 
In order to keep a check on the voltage, the latter was measured continuously by a 
Kelvin static voltmeter reading to 12,000 volts. About 5000 to 6000 was the voltage 
generally employed, 
Adjustments .—The apparatus having been provisionally set up, and all rough 
adjustments made, the following procedure was adopted. The glass plate was levelled 
as accurately as possible by an ordinary laboratory spirit-level, reading perhaps to 
something like SO" of arc. The guard-ring was then similarly levelled at a suitable 
distance, from 1 *5 to 2 centims. in our experiments, from the lower plate. This was 
then checked by measuring the distance between the top of the tin-foil and the 
bottom of the guard-ring, in several places by a wedge reading to A millira. directly. 
The adjustment being found correct, the hanging plate was adjusted to the guard-ring 
by eye, the under surface of the ring being flat and level, and the suspended plate 
also being flat. This can be done with great accuracy, just as in sighting a barometer. 
Of course the preliminary adjustment was good enough to let the plate hang concen¬ 
trically in the guard-ring. The adjustment must be made with the plate counterpoised, 
and is rather tedious, but presents no difficulties that may not be overcome by patience. 
The lower plate was then excited in order to find if the suspension stretched 
appreciably, but this was found not to be the case. The screens were then placed in 
position, and the experiment was readv for observation. 
