52 Mr. S. R. Milner and Prof. A. P. Chattock on 



the disadvantage of introducing a high thermal resistance not 

 independent of the temperature, and must therefore be thin 

 as well as plane. After some unsuccessful attempts to get 

 the bottoms of glass beakers ground flat inside and out, we 

 adopted thin disks of microscope cover-glass 12*7 cm. in 

 diameter and 0*051 cm. thick cemented onto the end of a 

 glass cylinder 11*5 cm. high, which formed the sides of A. 



To equalize the distribution of heat-flow the glass disk was 

 covered on its upper side with a circular sheet of tinfoil 

 9 cm. diam., cut by two concentric grooves about a millimetre 

 wide each into three pieces to prevent radial conduction 

 through it, the inner circle thus formed being equal in size to 

 the inner circle «]8 of the heating-coil. The heating-coil 

 itself comes immediately above the tinfoil ; it is insulated from 

 it by two circles of thin paper boiled in paraffin-wax ; and 

 the whole is then imbedded in wax as described above. Above 

 this is the single junction J of a german- silver-iron pair made 

 of very thin wires soldered to the centre of a disk of sheet- 

 zinc 2 cm. in diam. Above this is the cotton-wool W, then 

 the balancing-coil E (german-silver, 25 ohms, on a mica disk) 

 with another thermopile-junction F just below it similar to J. 

 Over E is another layer of cotton-wool, and finally a flat zinc 

 box P through which tap- water is passed in parallel with C. 



The leading wires to the thermopile-junctions J, F, K are 

 taken immediately into a trough of oil in which their junctions 

 with copper leads are immersed. From here the copper wires 

 lead to a mercury commutator by means of which J at the 

 heating-coil can be joined up in opposition to either K or F 

 for the measurement of the E.M.F. of the couple. 



The two vessels R.H. and L.H. are fitted up in all respects 

 alike. 



As will be seen later, the indications of these thermopiles 

 are used in determining small correction-terms in the ex- 

 pression for conductivity. It was therefore necessary to 

 calibrate them with thermometers ; but high accuracy was 

 not required. The piles in use were read by the deflexion 

 they produced on a mirror galvanometer, and the difference 

 of temperature corresponding to 1 cm. deflexion was found 

 to be 



R.H. JK . . . 0-516° 0. 



JF . . . 0-588 

 L.H. J'K' . . . 0-541 

 J'F' . . . 0-602 

 The constant of the galvanometer was tested each day, but 

 did not vary appreciably. 



