Rood — Experiments on [Ugh Electrical Resistance. 95 



is to be diluted with Chinese vermilion, when good substan- 

 tial layers can still be used. If the amount of the oxide of 

 manganese be reduced as low as twenty per cent, as high 

 resistances can be obtained as any that have been handled by 

 me. The thickness of oxide of manganese giving a resistance 

 of only one megohm was not measured, but must have been 

 less than one millimeter. Some of these low units, of from 

 one to a hundred megohms resistance, were given to Mr. H. C. 

 Parker, who is engaged in testing their constancy ; he finds 

 that they exhibit no polarization, neither have I, experiment- 

 ing with higher units, detected anything of this kind, by 

 which is meant that in neither case does the resistance increase 

 while the current is flowing. 



As already indicated, the best results were obtained by the 

 use of an insulating material of rosin with a small percentage 

 of yellow wax. Paraffin was also tried, but even with low 

 units exhibited a tendency to break down and lower its resist- 

 ance, temporarily ; plain yellow wax was better, but both of 

 them were liable to break down rather suddenly when the 

 resistances were quite high. Plain peroxide of manganese on 

 cobalt glass, not insulated by immersion in any bath, answered 

 admirably in dry weather for rather low units (under 50,000 H) 

 but in damp weather became quite unreliable. 



The units employed by me were 10 cm long, 2 cm broad, each 

 brass terminal having a length of 2 cm . For the higher resist- 

 ances they might be made two or three times as long with 

 advantage. 



Battery. — In the earlier experiments the street current was 

 employed with an E.M.F. of 112 volts, but was found to be 

 unsatisfactory owing to its small variations ; an arrangement of 

 200 cells or tubes with copper and zinc and very dilute acid is 

 preferable, but the so-called dry cells are far better, as with 

 such high resistances their electromotive force is almost per- 

 fectly constant, and they need no care. 



Leakage. — It would appear tolerably certain that when the 

 electrometer, with or without a condenser, is first employed, a 

 small amount of electricity is, so to speak, absorbed by it, so 

 that with a given unit, the time of a stroke becomes in a few 

 minutes somewhat shorter and then remains constant. This 

 leakage reduces itself to a minimum, but there is reason to 

 believe that some always remains. For example, if two equal 

 units are joined in a series, and matters are arranged so that 

 the time of the stroke does not exceed thirty or forty seconds, 

 then this joint time will be equal to the sum of their times 

 used separately. If, however, this joint time is allowed to rise 

 as high as one or two minutes, then it always exceeds the sum 

 of the separate times by anything from four to ten per cent or 



