ON THE THERMAL CONDUCTIVITIES OF CERTAIN ROCKS. 55 



was had, in consequence, to soft cements, to make the junctions between the 

 surfaces air-tight. An experiment was also made with slate, by plastering 

 the thermopile solidly to it with thin sheets of india-rubber moistened with a 

 mixture of red-lead and oU, and testing the conductivity of the plate when, 

 after remaining for two weeks under pressure, the cement appeared to have 

 Bolidiiied to ijerfect hardness. The result in this case (391) * scarcely dif- 

 fered either from the estimated number for the same rock-plate (392) given 

 in our former list, or from two other determinations (414 and 425), when, 

 instead of the solid junction, a thin paste of boiled starch and another of red- 

 lead and oil were used in succession to effect the junction. Two new 

 specimens of slate, cut from one piece, and tested by the same process, with 

 moist lutings of starch and linseed- meal to secure the junctions, gave as 

 values of their conductivities in different experiments the rather lower num- 

 bers, apparently belonging to this different sample of the stone, 340, 346, 

 349. The i)late of white Sicilian marble described in the last Eeport as pre- 

 senting when tested, by attaching the thermopile to it solidly with plaster, a 

 resulting conductivity 559, now afforded, with moist linseed luting, the 

 number 497. Whinstone, which formerly exhibited in the same manner an 

 absolute conductivity 3] 2, now afforded, with linseed luting, the conduc- 

 tivity 333. The results obtained with moist lutings are sometimes in excess 

 and sometimes in defect of those observed with solid junctions, and nothing 

 nccessaiy to be preferred in soUd over liquid attachments of the surfaces to 

 each other was found to be indicated by these preliminary trials. 



The inconstancy of some determinations now attempted of new rock- 

 specimens with the iron and palladium' thermopile led to the discovery that 

 its roUed branches produce thermoelectric currents by alteration of the con- 

 dition of the pressed metal ; and no anneahng by heat was able to remove 

 this serious objection. While unequally pressed parts of the wire along its 

 branches were subjected to unknown temperatures, it was obvious that small 

 differences of temperature could not be measured satisfactorily with the 

 thermopile, and no reliance in respect of ultimate accuracy could be placed on 

 the values found up to this time with the instriiment in its first constructed 

 form. German-silver wire was, however, substituted successfully for palla- 

 dium in the thermopile, with which the present series of experiments were 

 made. It consists of three rolled wires of German silver and two of iron in 

 series, between two rolled iron -wire terminals of a Thomson's reflecting 

 galvanometer, three junctions of the flattened helix into which they are 

 wound upon two bracing-bars of wood being above, and three below the 

 rock-section, which slips easily between them. A stout india-rubber collar 

 (half an inch in thickness) surrounds the rock-section before it is placed be- 

 tween the wire grating of the thermopile ; and similar collars round the con- 

 fronting ends of the boiler and cooler make a continuous non-conducting 

 casing of the rock-plate and of adjoining parts of the apparatus, protecting 

 them entirely, when the pressure acts upon them and upon the lutings that 

 connect them, from heat-communication to the outer air. Although German- 

 silver wire (even if unflattened) produced, when slightly heated at certain 



effects of undetected external influences might become sensible were the rate of heat- 

 transmission to be measured less than 0°'2 F. per minute. The rate actually observed 

 with the apparatus above described was between 0°'194 with cauuel-coal and 0°"400 with 

 quartz ; and in the following list of absolute conductivities the proper correction for the 

 external air-temperature round the cooler has in every instance been applied. 



* The significant figures only of the decimals representing the absolute conductivities- 

 in the following Table are hero used, for brevity, to denote them. 



