a Sphere in a Viscous Fluid. 521 



the movement of a single tooth corresponds to one revolution 

 of the tly- wheel, the period of rotation of the latter is \^ sec. 

 Hence the interval of time between two successive flashes is 

 12 x 45 sec * or '02037 sec, very nearly g 1 ^- sec. 



The second requirement, which is far more difficult to meet,, 

 is that the image of the ball should show a sharply-defined 

 outline. To ensure this result the light should issue from as 

 smnll an opening as possible so as to avoid the formation of 

 a penumbra or region of partial illumination, and at the same 

 time the duration of the flash should be so short that no' 

 appreciable motion of the ball takes place. In practice this 

 means that both the fixed and rotating slits should be as 

 narrow as possible. 



On the other hand, if the slits are made too narrow the 

 amount of light transmitted will be insufficient to give a 

 developable image. 



As the work proceeded considerable improvements were 

 made in order to overcome these difficulties, and the later 

 results show much greater sharpness than the earlier ones,, 

 although in the later photographs the balls were moving with 

 greater velocities. 



Several sources of light were experimented with, but that 

 which proved efficient, with the additional advantage of sim- 

 plicity, was a strip of burning magnesium ribbon held in 

 position by a fixed clip. 



The plates used were Cadett Lightning, developed with 

 hydroquinone and intensified when necessary. 



In order to find the actual and not merely the relative 

 velocities of the falling spheres, it is necessary to determine 

 the relation between the distances measured on the photo- 

 graphic plate and the distances traversed by the spheres im 

 the liquid. This was done by photographing a glass scale 

 suspended in the liquid in the same vertical plane as that 

 passed through by the spheres in their descent. Then the 

 actual velocity of a sphere must bear to the velocity obtained 

 by direct measurement of the plate the same ratio as the dis- 

 tance between two fixed points on the scale bears to the 

 distance between their images. 



The plates were measured by means of a travelling micro- 

 scope, the vernier of which read correctly to the tenth of a 

 millimetre. 



The main features of the photographic apparatus are shown 

 in section in the diagram, drawn approximately to a scale of j 1 ^. 

 The light produced by burning a strip of magnesium ribbon 

 M, inside the lantern, is brought to a focus by the lens L on 

 an adjustable slit K. It passes through one of the slits in the 



