598 
DR. J. T. BOTTOMLEY ON THERMAL 
referred to in this paper is thus temperature on Eegnault’s ‘‘normal constant 
volume” air thermometer. All the mercuiy-in-giass thermometers employed for 
subsidiary purposes were also compared directly with the air thermometer, and the 
suitable corrections applied wherever they were sensible. 
One part of the arrangement of the thermo-electric circuit still remains to be men¬ 
tioned. By far the most satisfactory way of experimenting for purposes like those 
under consideration is to take readings (in the way to be described immediately) of 
the galvanometer deflection on each side of the middle, or zero, position, at equal 
intervals of time, instead of reading’ only the deflection on one side, as is often done. 
Accordingly, I inserted in the circuit a mercury reverser, constructed with care to 
avoid the introduction of small uncertainties as to resistance. The use of this piece 
of apparatus also enabled me from time to time to apply to the galvanometer and its 
leading wires a very small known current from a standard cell, in order to assure 
myself of the constancy of the magnetic field. 
I may now explain the mode of experimenting. The value of the thermo-junction 
having been determined, and the copper globe screwed on to the plug, and the 
apparatus, which is supported by clips on a retort stand, having been adjusted so 
that the globe shall hang in the centre of the enclosure, the lower half of the enclo¬ 
sure is brought up and held pressed against the upper half by means of temporary 
clips. Owing to the edges of the flange having been well prepared by grinding, and 
by having a very thin layer of solder or tinning applied to them, the parts fit together 
very exactly; and in this position they are now soldered together, wdth the help of a 
soldering bolt and a Bunsen burner. Besin is used for the soldering, except just at 
the end of the operation, when a little soldering fluid is apjflied round the outer edges 
to make sure of perfect filling up of the minutest holes. The soldering up of these 
edges in such a wmy as to be tight for the Sprengel vacuum is a very troublesome 
operation, and one which requires the greatest care. 
When the soldering is finished, the joint between the brass tube and the glass tube 
is made complete by means of siegelwachs; and finally the apparatus is connected to 
the Sprengel pump by means of a portable blow^pipe. The intention, generally 
speaking, in these experiments, was to determine the loss of heat from surfaces 
prepared in various ways, both with air jDresent (and later with other gases), and also 
in more or less complete vacuum. Sometimes the cooling in vacuum was tried first, 
and the cooling in air afterwards ; sometimes the order was reversed. Sometimes the 
cooling in air was determined, then in vacuum, and then in air again. 
This being understood, I may say that everything being prepared as described above, 
water was now poured into the rvater bath ; a flame wms applied beneath, and the 
boiling water communicated heat to the enclosure, and gradually to the suspended ball 
within. The galvanometer was watched and the time roughly noted to show the progress. 
I have not, horvever, made use hitherto of these observations on speed of heating. 
With full air present, the process of heating was not inconveniently slown With 
