HALL AND AYRES. — HEAT CONDUCTION IN IRON. 291 



whole disk, and therefore the mean difference of temperature of the two 

 faces of the iron. There is doubtless some inaccuracy in this conclusion. 

 Strictly, somewhat greater weight should be given to the indications from 

 the outer circle of junctions than to those from junctions nearer the centre ; 

 for an inspection of Figure A will show that each of the outer junctions 

 represents a somewhat greater area than one of the inner junctions. 

 The multiple arc method of operation makes no allowance for this fact, 

 but the error from this cause was probably very small. It is to be ob- 

 served, moreover, that an error of 1%, for example, in the absolute 

 value of the thermal conductivity of a particular piece of iron is of no 

 great consequence in the present state of investigation, provided the 

 change of conductivity with change of temperature can be determined 

 with some degree of accuracy. 



Before the experiments upon conductivity were made, a number of the 

 fine wires leading from the faces of the disk having failed, a complete 

 new set of wires, from the same piece as the first set, was put in by the 

 same method and in the same positions as before. The apparatus was 

 then set up once more, in its former condition as nearly as possible. 



The strength of the electric current coming from the disk was meas- 

 ured bj' means of an astatic galvanometer, the sensitiveness, or figure of 

 merit, of which was determined frequently by sending through it a known 

 fraction of a current measured by a good tangent galvanometer. The 

 resistance of the circuit containing the disk and the galvanometer being 

 known, the thermo-electromotive force producing |he current from the 

 disk was found. But before this e. m. f. could be translated into differ- 

 ence of temperature between faces of the iron, it was necessary to deter- 

 mine by experiment the e. m. f. arising from some known difference of 

 temperature between two junctions made of copper and of iron like the 

 iron of the disk. For this test a piece about 10 cm. long and 0.16 cm. 

 in diameter was cut from the same slab of cast iron from which the disk 

 had been taken ; and to each end of this slender bar a copper wire, from 

 the same piece as the wires attached to the coatings of the disk, was 

 fixed by electrolytic deposit of copper. The bar was set in a hard rub- 

 ber holder, about 2.5 cm. projecting at each end, and the whole was 

 mounted between two brass tubes in such a way (see Fig. 4) that water 

 flowing through either tube would flow over one end of the bar. Thus 

 water entering at Ai ran past the bulb of the thermometer 7\ along the 

 end /i of the iron, and out at Ei. An alternative exit for the water is 

 indicated by the dotted lines below I^. A coating of shellac was used to 

 protect the iron and the copper from the chemical action of the water. 



