130 REPORT—1874, 
of absolute measurement been applied separately to each specimen of the 
list instead of only to a few rocks which furnish data for calculating the 
absolute conductivities of the remainder. Circular disks of linen well wetted 
with plaster of Paris (mixed with a little glue or white of egg) were laid over 
the surfaces of two or three of the rocks, enclosing under them and against 
the rock the two points of the thermopile-pincette, which were also first 
dipped into plaster. When these had set quite hard under pressure, and were 
thoroughly dried by a gentle heat, they were placed in the apparatus, and a 
measurement of the absolute temperature-difference and accompanying heat- 
flow was thus obtained, affording the real conductivity and a means of com- 
paring it with the apparent one found by similar observations of the same 
rock when no plaster was used, and when the points of the thermopile 
merely pressed against its surface. Thus the thermoelectric difference ob- 
tained with the wire couples merely touching the surfaces of white statuary- 
marble between velvet faces was 16°; while for the same heat-flow, when 
the arms of the thermopile were firmly plastered to the marble plate, the 
temperature-difference observed was only 6°2*, being more than twice . 
and a half as large a difference in the former as in the latter case. With 
whinstone the corresponding temperature-differences were 26° and 8°65, in 
the proportion of very nearly 3:1. A similar experiment was made with 
cannel-coal, of which the conductivity is much less than those of the last 
mentioned rocks, the temperature-differences obtained being for the same 
heat-flow in the plain and plastered plate 53°-4 and 39°-7 ; in the proportion 
of only 1:37:1, a far smaller reduction than was observed in the two fore- 
going cases. Care is, however, necessary to introduce wet plaster under as 
well as over the points of the thermopile in cementing them to the rock, that 
air may be excluded and the junction may be solid—a precaution which was 
omitted in this case, and plaster without size was used, which in drying 
sometimes flakes off from the rock-surface, either entirely or in places, which 
may render an experiment, as that on cannel-coal may not impossibly have 
been from this cause, entirely valueless ; yet this result presents itself, with 
many others met with in the investigation, as very well worth repetition, 
with fresh precautions and with new arrangements, to guard against the 
possibility of false conclusions. 
Adopting for the present, as probably not far from the truth, a common 
reduction-factor of 22 as the proportion in which the recorded temperature- 
differences of the plain rock-surfaces between velvet faces exceeded the true 
temperature-differences of the surfaces of the rocks examined, and intro- 
ducing some very small corrections for the thicknesses of the plates, the 
thermal capacity of the metal cooler, &¢., which are all probably (as well as 
the allowance for heat-absorption in raising the temperature of the rock 
plates very slowly during the observations) really negligible in comparison 
with the uncertainty that attaches (except in one or two well-ohserved cases 
of absolutely measured temperature-differences of the rock-faces) to the 
great majority of the determinations from unknown peculiarities of surface- 
contact and heat-transfer where air surrounds the thermopile, the following 
Table gives the absolute thermal conductivities (in centimetre-gramme-second 
* The heat-flow through the plate was actually greater in this latter than in the former 
case in the proportion of about 5:4, showing that the rough plaster-washed linen surface 
received and delivered heat to the velvet coyers much more readily than the smoothly 
dressed surface of the stone; and the whole resistance was less in the former than in the 
latter case, although the rock plate itself had been made thicker. The same diminution 
of the total resistance oceurred also in the experiment with plasteged whinstone, 
