Il8 THE SCIENTIFIC PAPERS OF 



due to its own diameter, and to the correct number of rotations 

 ri per second. 



it is 



2m 

 and also n '=^ 2ffh ' 



2pir 

 if p is the radius of lower aperture of cup ; therefore 



2rir 



Or *' = V--298p> ; 



and in substituting this value for h' into the first formula, we have 



n = 



2rir 

 or 



for our corrected formula to determine the velocity of cup with 

 continuous overflow). 



In applying this formula to the clock, taking for p its value of 

 8 millimetres and the other values also in millimetres in order to 

 avoid fractions, we find 



19600 * 881+ 



400 



-16-05) 



= 7*4 revolutions per second, 



or '1 revolution per second less than the actual speed. This 

 remaining excess of the actual over the calculated velocity of the 

 cup is rather more than what may be fairly attributed to error of 

 measurements, and appears to be due to adhesion of a film of 

 liquid (which may be estimated at nearly '25 millimetre thickness) 

 to the inner sides and edge of the cup, whereby its effectual dimen- 

 sions are proportionately reduced. For a uniform speed this error 

 must be a constant quantity, which cannot affect the working of 

 the instrument injuriously, so long as the other conditions are such 

 as to produce uniform rotation. The discrepancy is diminished 

 by increasing the dimensions of the cup, and its amount is such 



