760 Mr. H. D. Arnold on Stokes's Law for 



watch was "used, and thus considerable chance of error was 

 introduced in the case of short fall times. In general the 

 times may be considered accurate to the nearest fifth of a 

 second, although the occasional errors must be much greater 

 than this. 



Density. — The densities of the oils were found by the 

 pyknometer method at the temperature at which they were 

 used. The methods of determining the densities of the 

 spheres will be referred to later. It is important to notice 

 that it is the difference of the densities of the sphere and of 

 the liquid that occurs in the formula. If, therefore, we 

 attempt to obtain small velocities by making (a — p) small, 

 we must make the density determinations with great pre- 

 cision, and the question of the homogeneity of the spheres 

 becomes of importance. 



Radius. — Radii measurements were made with a micro- 

 meter microscope just before each sphere was placed in the 

 fall tube. With spheres of the size used there is little chance 

 of error unless the surface is highly reflecting, as is the case 

 with mercury. 



Viscosity. — A very simple form of Poiseuille's apparatus 

 proved sufficiently accurate for this purpose. A glass tube 

 2'2 cm. in internal diameter and <15 cm. long was bent at 

 right angles near one end, and fitted with a perforated 

 stopper through which a capillary tube could be inserted. 

 With the capillary clamped in a horizontal position, the 

 vertical tube was filled to a height " h " with the liquid under 

 investigation. By placing a beaker in contact with the tube 

 near its outlet the liquid flowed steadily down the side, 

 obviating any error which might be introduced by the for- 

 mation of drops. Experiments with different heads of the 

 liquid showed the back pressure to be negligible. To guard 

 against errors due to irregularities in the capillary bores two 

 tubes were chosen, their average internal diameters were 

 found by mercury columns, and they were then checked 

 against each other by a determination of the viscosity of 

 colza oil. The Table I. (p. 761) shows that Ihe values agree 

 as closely as could be expected from the probable errors in 

 height and temperature measurements, and justifies the use 

 of either tube separately. 



Hose Metal in Colza Oil. 



The Rose metal spheres were made from metal cast a 

 number of years ago for another purpose. It had been 

 turned up in two cylinders, and hence its density could be- 



