302 Mr. C. C. Paterson and Dr. N. Campbell on the 



battery well insulated from earth, E an electrostatic volt- 

 meter, and C a condenser connected across the terminals of 

 the voltmeter. A separate condenser C is not always in- 

 serted ; if it is not, the capacity of C represents simply the 

 capacity of the voltmeter and its connexions. A is shown 

 connected to the valve so that the negative peak potential is 

 measured ; if the positive peak potential is required, the leads 

 from the valve to A and E must be interchanged. 



The following are the conditions necessary for accuracy: — 



(1) The valve must pass no reverse current. 



(2) The insulation between the terminals of E must be so 

 great that the potential between them does not fall appre- 

 ciably in the interval between successive occurrences of the 

 peak potential. 



The necessity for these two conditions is obvious. 



(3) Condition (2) could always be fulfilled by making C 



sufficiently large and sufficiently well insulated. But the 



insulation must fulfil a further condition ; it must be so 



great that the amount of electricity which the source of 



potential has to supply at successive occurrences of the peak, 



in order to make up the loss occurring between them, is not 



so large as to alter the peak potential obtained. Suppose, 



for instance, that the source of potential can be shunted by a 



resistance of 1 megohm without affecting appreciably its 



peak potential ; then, if the interval between successive 



occurrences of the peak potential is n times the duration of 



the peak potential, the least resistance which can be placed 



as a shunt across E will be of the order n megohms and not 



of 1 megohm. Thus, when the source of potential was a 



magneto, it was found that a resistance of 10,000 megohms 



across E reduced the peak potential recorded by nearly 



10 per cent., although a resistance of 1 megohm placed 



across the terminals of the machine did not affect it more 



than 2 per cent., and although the potential recorded by E 



did not fall by as much as ^ per cent, between successive 



occurrences of the peak potential. 



It may be noted that it is a sufficient criterion that con- 

 ditions (2) and (3) are fulfilled that the peak potential 

 recorded is independent of the temperature of the filament 

 of the valve. It will always be found that there is a lower 

 limit to the temperature below which the peak potential 

 recorded decreases : but unless (3) is fulfilled, there will be 

 no well-defined upper limit above which the temperature of 

 the filament is without effect on the recorded potential. 



(4) The capacity C must be a sufficiently large multiple 

 of the capacity of the valve. The necessity for this condition 

 appears to be often overlooked. 



