Jjicilit on Electrical Tension in Metals. 83»3 



constant, with the exception of very slight variations due to 

 chanocs of temperature. These variations were so gradual 

 that it was impossible to confound them with any of the phe- 

 nomena here described; and so much the less as the direction 

 of the deflections due to the behaviour of the metallic plate 

 could be and was altered at pleasure. The apparatus was so 

 sensitive that with a single thermoelectric couple of iron and ' 

 German silver a deflection of 300 millimetres was obtained on 

 warming the junction with the hand. In this case the distance 

 of the mirror from the scale was about 2\ metres. In com- 

 mencing operations each of the plates, already described, 

 was covered with a black cloth and left to itself until any 

 possible deflection of the galvanometer had vanished. This 

 precaution was necessary, owing to the fact that, when the 

 different parts of the apparatus had been recently connected, 

 differences of temperature had probably arisen, through the 

 ti^ditenino' of bindino;-screws, the warmth of the observer's 

 body, and other causes. 



As soon as the galvanometer indicated no change for re- 

 versed positions of the commutator, the plate was uncovered, 

 and was then exposed to diffuse daylight, when a small deflec- 

 tion was sometimes observable. This deflection immediately 

 increased when the metallic side of the plate was illuminated 

 by magnesium-light. The true deflections, i. e. the semidiffe- 

 rcnces of the extreme positions of rest of the needle corre- 

 sponding to reversed positions of the commutator, amounted, 

 when magnesium-light was employed, to from one to four 

 millimetres. After exhibiting this phenomenon, each plate 

 w^as then warmed by means of a spirit-lamp at the junction 

 of the two metals, and a deflection was thereby produced 

 wdiich was invariably opposite in direction to the previous 

 one, and, in general, greater than it. The plates were usually 

 warmed on the glass side ; but there was no difference in the 

 effect when, as v^as done in a few cases, the flame of the 

 spirit-lamp was brought into actual contact with the metallic 

 surfaces. The connexions between the plate and the commu- 

 tator were repeatedly changed, in order to make sure that the 

 electrical currents o])serv8d on exposure to light originated 

 in the metal plates themselves, and not in any other part of 

 the apparatus. Under these circumstances it was clearly 

 proved that the current generated by the light had always the 

 same direction in each particular plate, and consequently must 

 have originated within it. 



Of the five metals here mentioned, nine different com])ina- 

 tions of two were examined, and several specimens of each 

 combination were prepared ; so that, on the whole, thirty plates, 



