at Very Low Rates of Shear. 687 



cross-sections of the beakers. The tube and beakers were 

 immersed in a large water-bath, and the work was performed 

 in a suitable u constant temperature" room. 



If Ji be the difference of level of the liquid in the two 

 vessels at time t. the volume of flow in time dt from one 

 vessel to the other is. according to Poiseuille's law, 



If Aj and A 2 be the areas of the sections of the vessels at the 

 level of the" liquid, the change of the difference of level 

 between the two vessels in time dt is 



A 2 A 2 Sir) \A A 2 / j] 



whence it readily follows by integration that 



/<„ being the difference of level at time t , and li x that at 

 time £]. 



The flow must be slow enough for the motion to be recti- 

 linear. This condition, according to 0. Reynolds' criterion, 

 is amply satisfied in the following experiments. 



Sources of Error. 



The method is simple, but there are several sources 

 of possible error the effects of which may become very 

 large when very small differences of level are to be 

 measured. — (1) Surface tension. A sensible curvature of the 

 surface of the liquid between the vessel and the float would 

 be a source of pressure (positive or negative) on the liquid^ 

 and the magnitude of this could not be easily ascertained. 

 Also the float would probably be drawn to one side and would 

 not ri><* or fail vertically. Both sources of error were elimi- 

 nated by choosing large vessels 10 cm. in radius and floats 

 4 fin. in radius, so that the surface-film between vessel and 

 float was about <> cm. wide. The vessels first tried were 

 much -mailer, and the error was considerable, the results 

 showing no consistency. — (2) Change of pressure of lever on 

 float. If the centre of gravity of tin; lever be not in the 

 plane of the ends of the legs, any change of tilt of the lever 

 will cause a redistribution of pressure of the lever on the 



2 Z 2 



