ON THE THERMAL CONDUCTIVITIES OF CERTAIN ROCKS. 21 



ohms when the wires were coupled for observing a difference of temperature ; 

 and it was assumed that, with this resistance and with the probable tendency 

 of twelve wires similarly circumstanced to neutralize each other's false efl'ects, 

 no sensible errors from local disturbances would arise. The instrument was 

 submitted to some careful tests, with a result that, at the highest temperatures 

 of the experiments, errors in the temperature- difference amounting to about 

 1° F. may have been committed. At the ordinary temperatures of the wires 

 between 100° and 120° F. it was found, by substituting a heated iron disk 

 (coated on the faces with thin paper) in the place of a rock-plate, so as to 

 heat both sets of wires equally, that the only permanent deviations produced 

 as the plate sunk verj- slowly in temperature also sunk gradually with it from 

 an equivalent value of about |° to about g° upon the scale. As the coiTect- 

 ness of the small temperature-differences (of 6° and upwards) lying usually 

 between the above two temperatures was thus fairly checked, and for exception- 

 all}' higher differences and temperatxires the conditions could not easily be more 

 exactly assimilated to those of the actual experiments so as to control and 

 estimate them, the effects of these small errors have not been further regarded 

 in the calculations ; but in order to avoid changes of value in the divisions of 

 the scale, and to enable the actual temperature of each set of wire-junctions 

 to be directly observed, an arrangement of the thermopile was made by which 

 each set of junctions could be separately combined with a similar set in con- 

 tinuous circuit with the galvanometer placed in a small rectangular water- 

 bath. The latter is made of tin, and, as well as its lid, is well jacketed with 

 cork, and provided with an agitator ; so that by adding hot or cold water, 

 which can be withdrawn below, any temperature of the water in the bath 

 can be obtained. A simple commutator enables the circuit with the galvano- 

 meter to be closed, either through the two principal sets of junctions or 

 through one of them and through a set corresponding to it in the bath ; so 

 that by changing the temperature of the latter until no current passes through 

 the circuit the actual temperature of each rock-face could be observed. This 

 mode of observation is fi-ee from all objections, excepting those of false 

 currents arising in long wires and plates of the same metal maintained at 

 very various temperatures ; but with the exception of the twelve loops of 

 German-silver wire projecting on one side from the rock-plate, the corre- 

 sponding loops on the other side, and all the rest of the circuits made to the 

 galvanometer, were formed from the same piece of iron wire freshly annealed. 

 The comb-like teeth of the commutator are pieces of narrow hoop-iron about 

 3 inches long, closely set together in wood, and also thoroughly annealed, to 

 which the proper terminals of iron wire are soldered at their feet, while the 

 upper ends are filed to chisel-edges ; and a small hand-rack of iron wedges set 

 on wood at proper distances apart, thrust between them in different positions, 

 completes the connexion in the three different orders that are required. The 

 additional branch wires used in the arrangement are few, and, as will be seen 

 from the following description, add very little to the total lengths of iron 

 wire which conduct the currents. The twelve-turn coil of wire in which the 

 rock is pressed consists of twelve half-turns or loops of German silver and 

 the same number of iron loops. The twelfth looj) of German silver (see 

 figure, p. 22) completes the circuit or connexion from the beginning to the 

 end of the coil through the medium of the galvanometer. There are 

 thus twelve junctions of dissimilar metals above, and twelve below the rock- 

 plate in a closed circuit with the galvanometer. To produce a new set of 

 twelve junctions corresponding to each of these, the loops of German-silver 

 wire are all cut through in the middle, and the free ends soldered to twenty- 

 four short pieces of iron wire, the junctions being laid side by side across a 



