for some Pure Metals, 



219 



magnetized to saturation by flashing through a very great 

 number of turns of fine wire round the needles as heavy a 

 current as the wire would carry. They were then sus- 

 pended, and the stronger gently stroked with a weak magnet 

 till the desired sensibility was obtained. The system was 

 quite stable in the earth's field, and the time of swing was 

 about 10 seconds. The mirror was concave, with a focal 

 length of 8 inches. A half-millimetre scale ruled on glass 

 was placed about 60 inches from the mirror, between it and 

 a well-lighted frosted window. The image of this scale was 

 very sharply defined and permitted magnification by an eye- 

 piece magnifying 10 diameters, the lines still being sharp 

 and clear, and tenths of a division could be estimated. An 

 electromotive force of ten absolute units produced a deflexion 

 of one scale-division, and one absolute unit might be detected. 

 A Thomson galvanometer of 10,000 ohms to be equally 

 sensitive would have to indicate 10"^ 12 amperes. 



The method of observation may be learned from the accom- 

 panying diagram. E is a Clark cell joined in series with an 



adjustable high resistance R, a known low resistance r, and a 

 key K x ; as a shunt on r there is the thermal junction e, the 

 galvanometer Gt, and a key K 3 . A copper wire w, of the same 

 resistance as the junction, is put across its ends in such a way 

 that by means of the three-way key K 3 either this wire or the 

 junction may be put into the gavanometer circuit. In making 

 an observation w is put into the circuit, K 2 closed, and the 

 deflexion, if any, observed. This comes from the unequally 

 heated junctions forming the circuit ; and so great was the 

 sensitiveness of the galvanometer that even when there was 

 nothing but copper and brass in circuit, and it had not been 

 interfered with for some hours (the first thing in the morning 



