ON THE THERMAL CONDUCTIVITIES OF CERTAIN ROCKS. 23 



tho upper junctions (a, a') and carried through the, bath to a separate tooth of 

 the commutator ; and by moving the wedges of the rack-piece together one 

 tooth-space to tho right or left (as shown in new positions by a x in tho 

 figure), combinations of junctions in the bath with junctions (a a') above or 

 {bb') below the rock-plate are put into connexion with the galvanometer. 



By the same mode of trial as before, a heated iron plate coated with thin 

 paper being substituted in the place of an experimental plate, the tempe- 

 ratures of its two faces, as exhibited by the thermometer in the bath when 

 the commutator was shifted from one of its two supplementary positions to 

 the other, were sensibly the same as the heated plate slowly cooled, and no 

 false difference of temperature arising from false currents differently excited 

 in the two circuits thus joined up were found to be indicated as a result of 

 several such determinations of the really equal temperatures of the two faces 

 of the plate. This mode of observing the actual temperatures and the 

 temperature-differences of the rock-faces in the present series of experiments 

 was therefore constantly employed, and the values of the scale-divisions in 

 degrees for the other more usual method of employing the thermopile were 

 not determined with special care, although this adjustment of the commutator 

 was also used to check and follow the gradual variations of temperature- 

 difference that were less speedily, although more certainly, measured by the 

 absolute method of determination. The only case of failure to observe a 

 sensible difference of temperature between the two sides of an experimental 

 plate occurred with iron-pyrites, which (as weU as galena), being a good 

 conductor of electricity, it was foimd necessary to coat with two thicknesses 

 of the thinnest tissue-paper on each face; and the apparent difference of 

 temperature recorded (which was decidedly less than 1°) may have arisen 

 from the resistance offered by the slight obstructions of these thin paper 

 sheets (soaked with water) to the passage of the heat : although certainly very 

 great, no definite value of the thermal conductivity of ordinary iron-pyrites 

 can therefore be assigned. It was also necessary to use oil junctions instead 

 of wet bladders, from the galvanic effects produced by the saturated salt 

 solution, when rock-salt was tested ; and it appears probable from some 

 measurements of quartz with the same kind of luting that the conductivity 

 of rock-salt thus found is somewhat less than, rather than likely to be in excess 

 of, the real thermal conductivity of that substance. As a good assurance that 

 when membranes wetted with water were used to press the thermopile 

 against the rocks the true temperatures of their faces were very nearly 

 marked, the experiment with iron-pyrites may be instanced, as the small 

 temperature-difference of less than 1° could not have been observed if the 

 wires were not very nearly indeed at the same temperature as the two paper- 

 covered faces of the pyrites against which they were pressed ; and as the 

 circumstances of their adjustment in other cases were exactly the same as in 

 this instance, it may be assumed that the method of pressing the thermopile 

 against the rocks with wet bladders adopted in the present series of experi- 

 ments exhibited the true temperature -difference of the faces, and afforded 

 correct values of the thermal conductivities. The pressure was applied by 

 means of strong spiral springs (instead of the weights described in the last 

 Report), whose extensions in a graduated tube indicated the pressures which 

 they were made to exert. The pressure thus applied was usually SO lbs. 

 upon a surface of nearly 20 square inches of the rock-plates, or about 4 lbs, 

 per square inch. The general agreement of the results with those formerly 

 obtained also serves to verily the correctness both of the thermal conduc- 

 tivities now air^signed and of those previously observed. The principal 

 differences in the two m.elhods of determination consist in the use of an im- 



