THERMO-ELECTRIC DIAGRAMS. 



303 



stroked with a small weak mag-net till the condition of instability 

 was approached as nearly as was desired. A silk suspension was 

 \ised for some time, but it was afterwards replaced by a quartz 

 fibre, the finest I could make, but not the finest I could wish for. 

 I put it in in hopes of doing away with the fatigue of the silk, 

 which was continually shifting the zero of the instrument, and 

 was somewhat disappointed in finding that the charge of zero 

 was still observed after a large deflection. The mirror was a 

 very fine concave one, of about Sin. focal length. Glass scales 

 Avere found far preferable to opaque ones, but with glass scales 

 one has again a variety of choices. Clear lines on black ojiaque 

 ground, black lines on clear grovmd, and clear lines on red 

 transparent ground (got by etching through the red "flashing" on 

 common red glass) were all tried, and each has some advantages 

 over the others, the red being delightful to work with in very 

 strong light. I decided to use a dark scale on clear ground, and 

 ruled a half millimetre scale accordingly. I fixed it between 

 the galvanometer and a frosted window of northern aspect, to 

 make sure of its always being well lighted. Its distance from 

 the mirror was about 4ft. The image fonned about 9in. from 

 the mirror was capable of being magnified about ten diameters 

 without sacrifice of distinctness, and a Ramsden eyepiece pro- 

 vided with cross wires was used to examine it ; tenths of a scale 

 division could be estimated, equivalent to measuring to 5" of arc in 

 the movement of the magnet. A deflection of one scale division 

 would be produced by an e.m.f. of 4 or 5 absolute units, according 

 to the resistance of the junction being measured. A Thomson 

 galvanometer of 10,000 ohms, to be of equal sensitiveness, would 

 have to indicate 10~^- ampere. 



The method of observation vised is diagrammatically shown 

 _ in the figure. E is a battery, R a 



high resistance, r a low resistance 

 (generally 1 ohm), G the galvano- 

 meter, e the junction whose e.m f . is 

 to be measvu-ed. R, is adjusted till no 

 current flows through the galvano- 

 meter. 



Then if C be the current through 

 the large circuit e is equal to the 

 e.m.f. at the ends of r, i.e , e z=. CV. 

 The current might be determined either 

 by using a constant cell and using 

 known resistances, or by measuring it 

 directly by a tangent galvanometer or 

 current balance. The former method 

 was used. For a constant cell the 

 choice was open for one or other of the various forms of Standard 

 Daniell or the Latimer Clark cell. I first tried Fleming's Daniel 



\ 



— jwwuwb 



py 



5^ 



