under the Influence of the Electric Discharge. 363 



3. For the determination of the number of discharges the 

 very accurate heliometric method employed by Wiedemann 

 and Biihlniann * could not be used, on account of the great 

 number of discharges. It was difficult to attach a mirror to 

 the axis of the machine itself, in consequence of the mode of 

 its construction; and it made too few revolutions. In par- 

 ticular cases satisfactory results were obtained by a self- 

 registering method. The end of the wire coming from the 

 galvanometer usually connected to earth, was placed opposite 

 to a cylinder covered with tin-foil and blackened; a tuning- 

 fork was also made to record its vibrations on the blackened 

 surface. The lamp-black is removed at each discharge. But 

 this method is only applicable when the interpolation of a 

 spark in the circuit is of no consequence. 



The following method has been found satisfactory : — An 

 ordinary Greissler's tube was placed in front of and parallel 

 to the discharge-tube A, which was to be investigated 

 thermically or spectroscopically. Both were then covered 

 with paper so that the covered portion of the one corre- 

 sponded to the uncovered portion of the other; and both were 

 then observed in a revolving mirror. The discharge of a 

 small induction-coil w T as passed through the Geissler's tube, 

 the primary current of which was made and broken 100 

 times a second by means of a self-acting electromagnetic 

 tuning-fork. The reflection in the revolving mirror showed 

 a series of bright lines of light, at distances corresponding 

 to hundredths of a second. By observing how many dis- 

 charges of the tube A lay between two of the Geissler 

 tubes, the number of discharges in yj q of a second, and con- 

 sequently the number per second, was ascertained. It was 

 possible to make a tolerably accurate estimate when there 

 were not more than 4 or 6 discharges in the t Jq of a second. 

 But when the machine was in more vigorous action, as was 

 always necessary when thermic measurements were to be 

 made, the number was much greater (up to 60 times). 



The number of discharges could, however, be obtained in 

 this case by allowing the machine to run more slowly, and 

 observing the deflection of the galvanometer which corre- 

 sponded to a given number of discharges. In order to find 

 the number of discharges during the measurements, it was 

 only necessary to multiply the number observed in the first 

 case by the ratio of the galvanometer-reading in the two cases. 

 It is true that in doing so we make the assumption that 

 the number of discharges is proportional to the current- 

 strength ; but this assumption is justified if between two 

 * Wiedemann and Ruhlmann, Pogg. Ann. cxlv. p. 242, 1872 



