520 
ME. W. CEOOKES ON EEPULSION EESTJLTING- EEOM EADIATION. 
enable me to work with more convenience and accuracy. Instead of trusting to the 
comparison between the barometric gauge and the barometer to give the internal 
rarefaction of my apparatus, I have joined a mercurial siphon-gauge to one arm of the 
pump. This is useful for measuring very high rarefactions in experiments where a 
difference of pressure equal to a tenth of a millimetre of mercury is important. By 
its side is an indicator for still higher rarefactions ; it is simply a small tube having 
platinum wires sealed in, and intended to be attached to an induction-coil. This is 
more convenient than the plan formerly adopted (51), of having a separate vacuum- 
tube forming an integral part of each apparatus. At exhaustions beyond the indications 
of the siphon-gauge I can still get valuable indications of the nearness to a perfect vacuum 
by the electrical resistance of this tube. I have frequently carried exhaustions to such a 
point that an induction-spark will prefer to strike its full distance in air rather than 
pass across the inch separating the points of the wires in the vacuum-tube. A pump 
having these pieces of apparatus attached to it was exhibited in action by the writer 
before the Physical Society, June 20th, 1874. 
83. The cement which I have found best for keeping a vacuum is made by fusing 
together 8 parts by weight of resin and 3 parts of bees-wax. For a few hours this 
seems perfect, but at the highest exhaustions it leaks in the course of a day or two. 
Ordinary or vulcanized india-rubber joints are of no use in these experiments, as when 
the vacuum is high they allow oxygenized air to pass through as quickly as the pump 
will take it out. Whenever possible the glass tubes should be united by fusion, and 
where this is impracticable mercury joints should be used. The best way to make these 
is to have a well-made conical stopper, cut from plain india-rubber, fitting into the wide 
funnel-tube of the joint and perforated to carry the narrow tube. Before fitting the tubes 
in the india-rubber, the latter is to be heated in a spirit-flame until its surface is decom- 
posed and very sticky ; it is then fitted into its place, mercury is poured into the upper 
part of the wide tube so as to completely cover the india-rubber, and oil of vitriol is 
poured on the surface of the mercury. When well made this joint seems perfect ; the 
only attention which it subsequently requires is to renew the oil of vitriol when it gets 
weakened by absorption of aqueous vapour. Cement has to be used when flat glass 
or crystal windows are to be cemented on to pieces of apparatus, as subsequently 
described (99, 102). 
It would be of great service could I find a cement which is easily applied and removed, 
and will allow the joint to be subjected to the heat of boiling water for some hours 
without leaking under the highest rarefactions. Hitherto I have failed to find one 
which answers these requirements. I mention this in the hope that some one who 
happens to read this may be in possession of the recipe for such a cement, and will 
communicate it to me. 
84. Before my first paper on this subject was read before the Boyal Society I 
had discarded the balance form of apparatus there described, and commenced experi- 
menting with bulbs and tubes in which quantitative results could be obtained. On 
