34 J. E. Moore — Electrical Discharge from the point 



this pressure, the discharge producing visible effect in the gas, 

 is almost wholly confined to the concave surface of the elec- 

 trode. This discharge stream is seen to be affected by a mag- 

 net, causes fluorescence of the glass walls, and electrifies the 

 glass walls where it strikes, precisely as a discharge stream from 

 a cathode, in a "conduction tube"; and this, whether the 

 electrode be connected to one or the other side of the spark 

 gap. The above observations have been repeated, with the 

 uni-directional potential-difference described in an earlier para- 

 graph, with the same results as are here given. 



In the case of the heated spherical cup, previously consid- 

 ered, the quantity of heat conducted, in unit time, across any 

 isothermal surface limited by any unit heat flow tube, is, by the 

 kinetic theory, -JlSTrnv 2 ; N being the number of molecules cross- 

 ing the surface in one second, in the direction in which heat flows, 

 in excess of the number crossing the surface in the opposite 

 direction, v the mean velocity, and m the mass of the mole- 

 cules. But ISTm oo />, the gas density ; and -J/ov 2 = p, the pres- 

 sure of the gas. So that the heat-flow lines, shown in fig. 7, 

 may be taken to represent the normal component of the pres- 

 sure brought into existence in the gaseous medium by the 

 operations of heat conduction. Plainly a family of surfaces, 

 over each of which the gaseous pressure has a constant value, 

 can be described about the heated body. These surfaces cut 

 the pressure or heat-flow lines orthogonally, and appear, in a 

 meridian section, as lines everywhere normal to the heat-flow 

 or pressure lines. A few of these lines are shown in fig. 7. 

 The form of these surfaces is practically identical with that of 

 the " dark space," as it recedes from the cathode when the 

 pressure of the gas is reduced. 



From the description of the method of drawing the heat- 

 flow or pressure lines, shown in fig. 7, it will be clear that the 

 space rate of variation of pressure outwards, from the heated 

 surface, is a direct function of the angular divergence of the 

 heat-flow lines. Accordingly, as Professor Reynolds- 

 has pointed out, the reaction of the gaseous medium on the 

 convex side of a heated spherical cup ought to be greater than 

 on the concave side. Two such cups ought to rotate, with 

 their concave sides forward when properly pivoted in a rare- 

 fied gas, and brought to a higher temperature than the sur- 

 rounding gas. This they are observed to do in Crookes' cup 

 radiometer. 



It was thought worth while to determine whether or not 

 similar reactions could be observed, when the spherical cups 

 instead of being heated, were brought to a high electrical po- 



* Cf. note, p. 26. 



