Joty—On the Volume Change of Rocks and Minerals attending Fusion. 291 
heated air. On the other hand, it loses little, comparatively, by radiation. It is 
not impossible also that a selective absorption of rays from heated platinum by 
platinum may be involved. This would be of interest if it turned out to be 
involved, but no experiments were made to investigate the matter. In any event 
the temperatures, as determined by the melting of substances such as potassium 
carbonate, must afford closely reliable data, such bodies sufficiently resembling 
the substances subsequently under experiment in their physical properties, and all 
conditions attending the calibration and the subsequent experiments being alike. 
Such bodies will alike absorb and radiate freely and gain little by conductivity. 
It may indeed be inferred that when an error no greater than 7°5 per cent. attends 
the substitution of substances so different in physical properties as are platinum 
and potassium carbonate, but very little difference will attend the behaviour of 
nonconductors among themselves. The temperature assumed by the body hanging 
in the oven, supposing the ovens closely alike in dimensions, must be a question 
of the conductivity for heat and the absorptivity of the substance for the radia- 
tions to which it is exposed. The behaviour of a metal in both respects will, it 
may be safely assumed, differ more markedly from the behaviour of such a 
substance as potassium carbonate than will this last substance differ from the 
silicates subsequently dealt with. Except, however, investigation showed that 
there was some special absorption, as suggested, of the rays from heated platinum 
by the platinum bead, the thermometric curve determined by the platinum 
thermometer will be that proper to use when metallic substances are under inves- 
tigation. It will be presently seen that the form of the thermometric curve, 
whether for metals or non-metals, is in its general features the same. The lower 
part of the curve is concave to the axis of temperature, the upper part convex. A 
general explanation of this is to be found in the fact that the rate of loss of 
radiation of the heated ribbon forming the oven inereases rapidly with the tem- 
perature, and therefore with the current traversing it. Thus the heating effect 
of the current is proportional at all points to the square of the current (neglecting 
change of resistance of the platinum), whereas the rate of loss of heat by radiation, 
at first simply proportional to the temperature, will rapidly mount up. For a 
certain range it will also be proportional to the square of the temperature. Over 
this interval the thermometric curve will be approximately a right line. Ultimately 
the radiation loss will become more nearly proportional to the cube of the current. 
At this stage the curve will become convex to the axis of temperature. The course 
of the curve will be complicated by the electrical behaviour of platinum at the 
various temperatures. 
A thermometric curve was now constructed by heating to their melting points 
beads of chloride of silver, sulphate of silver, potassium carbonate, sodium fluoride, 
and copper oxide. As this curve applies to some of the results recorded in this 
TRANS. ROY. DUB, SOC., N.S. VOL; VI., PART XII, NE 
