DISPLACEMENT INTERFEROMETRY. 105 



the latter conditions it may happen that the particles actually stop and then 

 begin a retrograde movement, soon to be accelerated in turn. During this 

 period of transition, particles may be seen also to rise and fall, but with rela- 

 tively great slowness as compared with the usually swift horizontal motion. 

 Some of the arrows are somewhat oblique to the horizontal. Under rare 

 conditions I noticed a line of light instead of shadow. Breadths differ greatly 

 and would naturally depend on focussing. 



Usually the motion persists with apparently undiminished swiftness for 

 hours, so that it much outlasts one's patience. Often a single particle can be 

 observed for a minute or more ; but after 10 or 20 hours all particles disappear 

 and the spectrum is clear. From this I concluded that the diffractions are 

 not due to local difference of density, etc., of the solution, as I first supposed, 

 but actually originate either in minute solid particles (or, in case of other 

 liquids, in minute air-bubbles) entrapped in the liquid. The slow subsidence 

 and persistence of particles indicate this state of things. 



Moreover, I eventually found that the motion of particles as a whole from 

 red to blue or blue to red could be controlled by rotating the cylinder G on 

 its axis, a, either counter-clockwise or the reverse, respectively. Brownian 

 motions are excluded, since these are promiscuous and since the magnification 

 is inadequate. It is difficult to conceive how the angular momenta impressed 

 on this solution can persist for hours within it, after the solution is apparently 

 quite at rest, even if the solution is of large density (dense enough to float 

 glass) . Probably, since the internal friction of liquids vanishes with the rela- 

 tive velocity of layers, and since the apparent motions are magnified, there is 

 eventually no friction torque left to absorb whatever angular momenta may 

 be renewed or survive. The occurrence of direct and retrograde motions at 

 the same time, separated sharply by a plane of demarcation, is suggestive of 

 vortices. Above this plane, particles move with about the same speed in one 

 direction; below the plane with a very different speed in the opposite direction. 

 A particle which happens to be in the plane in question does not move at all. 

 After a long interval the direction of the motions above and below a plane of 

 demarcation may be found to have reversed, respectively. If a solution is 

 cleared of particles by the lapse of a sufficient time for subsidence, they may 

 be restored by brisk rotation. The number, size, density of color, and speed 

 of the particles increases with the violence of rotation. Gentle rotations in 

 opposite directions leave the particles in a curious state of indecision, after 

 which the definite direction red to blue is adopted. 



I have also tried the method where there is symmetrical reflection within 

 the cylinder (as in the case of the rainbow) . The results are similar, but less 

 luminous. 



To conclude: After the cessation of the initial disturbances, the liquid, 

 left to itself and owing to the presence of motes, shows a persistent motion of 

 its middle layers in the general direction of the impinging beam of light, while 

 the motion of the relatively thin layers at the top or bottom (one or both) 

 is usually persistently retrograde, but slow in comparison. This continues 



