56 KEPOET — 1875. 



points of its length, sensible thermoelectrical effects of local disturbing cur- 

 rents, •wbich (as vras also observed in the palladium wire and in some other 

 metals which -were tried) did not appear to be removed, like those of strained 

 iron wire, by annealing at a considerable heat, yet the high electromotive 

 force of the German-silver iron couple and the neutralizing eifect of the 

 several loops of German-silver wire, all exposed to the same temperature 

 variations, were found, in some suitable experiments made in the process of 

 determining the scale of the instrument's indications, to have almost entirely 

 eliminated the influence of these small disturbing actions ; and the smallest 

 differences of temperature, of only two or three degrees, occurring in the 

 rock-conductivity experiments could be accurately measured. Eesistances of 

 two, five, and ten ohms were included in the thermoelectric circuit succes- 

 sively in each experiment ; and the proportionality of the galvanometer read- 

 ings in these several conditions being constant, showed the constancy of the 

 resistance, and accordingly the unvarying scale-value of the indications of 

 the instrument for every observation ; while the zero of the scale-readings 

 could also be conveniently determined at any moment by unplugging, in the 

 rheostat connected in the circuit, a very large resistance of 1000 ohms. 

 Twelve, seven, and four divisions of the scale represented respectively l°r., 

 when the several resistances above named were used * ; and three readings 

 being taken for every temperature- difference observation, the results reduced 

 in this proportion were accordant to one or two tenths of a degree, as long 

 as the total resistance of the circuit had undergone no variations from acci- 

 dental injuries of the connexions, and as long as the thermometric value of the 

 scale-divisions had accordingly been preserved. The electromotive force of 

 a German-silver iron thermopile is shown, by Prof. Tait's representation of 

 the specific curves of these metals in his ' First approximation to a Thermo- 

 electric Diagram,' to be at ordinary temperatures very exactly proportional 

 to the difference of temperature between the junctions ; and the temperature- 

 differences noted in these experiments may, it is presumed, be accordingly 

 regarded as affected by only very small errors of uncertainty. The thickness 

 of the flattened wire (about 0-5 millim.) of the thermopile occasioned, 

 however, even with the considerable pressure used, a certain thickness of the 

 lutings ; and it is probable that, from this cause as well as from the partial 

 fluidity, instead of perfectly solid nature, of the cement, the temperature-difter- 

 cnccs observed somewhat exceeded those which actually existed between the 

 faces of the tested plates. Besides the small corrections for escape of heat 

 from the cooler to the outer air, an addition of one tenth to all the observed 

 conductivities is made in the present list for the cooler's thermal capacity and 

 for that of the brass foot, wire gauze, and agitator, which it enclosed, the cor- 

 rection for which was not included in the conductivities assigned last year. The 

 final results in the present list are, notwithstanding this correction, somewhat 

 inferior to those formerly observed with the thin palladium-iron thermopile, 

 whether its attachments were solid or effected by moist cements ; but in re- 

 peated experiments with that instrument on quartz and other rock-sections 

 offering only small resistances, the results arrived at were in general so high, 

 and at the same time so irregiilar, from the predominance, in measuring their 

 small temperature-differences, of the local currents, that for this reason only 

 a partial dependence on its generally higher indications can be placed. The 



* An accidental iujui-y -nliieh haiDpened to one of the solclerings was thus immediately 

 detected; and the fault having been found and repaired, when the instrument was re- 

 .gradnated, it was found to afford, with the above resistances in its circuit, exactly the 

 same proportional values of the Bcale-iudications as before. 



