Messrs. G. J. Stoney and R. J. Moss on Crookes's Force. 71 



With the improved apparatus repulsion at first appeared to 

 exist at all tensions in hydrogen, the distance between the disks 

 being 1 millim. This repulsion was not perceptibly increased by 

 alterations in tension, until the latter was reduced to about 200 

 millims., when a slight increase of repulsion took place ; and further 

 reductions of tensions were in each case followed by an increase of 

 repulsion. 



It was observed that \\-hen repulsion was detected at ordinary 

 tensions the glass disk occupied the upper portion of the containing 

 tube, and that, when the disk was made to swing in the lower por- 

 tion of the tube, instead of being repelled it was attracted towards 

 the pith when illuminated. These results are obviously due to 

 convection-currents. By carefully balancing the glass disk as 

 nearly as possible in the centre of the tube the effects are greatly 

 reduced ; but it Mould be very difhcult, if not impossible, to balance 

 the disk in a perfectly neutral position. 



In an atmosphere of hydrogen at ordinary atmospheric tensions 

 there is no indication of attraction or repulsion when the distance 

 between the pith and glass disks is 100 miljims. and the time of 

 illumination 15 seconds, a period which experience has led us to 

 adopt. The first immistakable indications of pressure on the 

 swinging disk at this distance occur when the tension is about 

 50 millims., at which tension there is a very feeble repulsion. As 

 the tension is reduced the repulsion increases. An extensive series 

 of experiments haAe teen made for the purpose of determining 

 the ratio in which the repulsion increases for given reductions in 

 the tension of the residual gas. The accompanying Table exhibits 

 the results of one set of experiments : — 



T = tension of the residual gas. 



D = distance between the disks in millims. 



P=pressure on the swinging disk, as represented by the space, 



in degrees of the scale, through which the index passes 



in 15 seconds. 



of the molecules of the gas and the surface of a solid with which they come 

 into contact are at another temperature, then the Cixx)tes's force whicli arises 

 may be either normal to that surface like the pressure of a gas, or tangential 

 to it like friction, or in any way compounded of these two, being in each ease 

 in the direction spoken of above. 



Accordingly the forces that act upon the containing ressel and the vanes of 

 radiometers are in general partly tangential and partly normal ; so that in 

 estiruating the intensity of these forces the matters to be considered, along with 

 the directions of the motions in the intervening gas, are the proximity and 

 extent of the opposed surfaces, and not the degree of their parallelism. It is 

 probable that all the phenomena of radiometers with crumpled or inclined 

 vanes, which Mr. Orookes has lately investigated, admit of explanation by these 

 considerations. 



It is manifest that in the apparatus described in the text the chief part of the 

 reaction on the containing vessel, when the disks are at a considerable distance 

 asunder, consists of tangential forces acting on the inside of the tube. 



