458 Mr. Owen Glynne Jones on 



the test is simply to send down a minute sphere and observe 

 its speed for short lengths all along the tube. This speed is 

 constant if the temperature is uniform. In accurate esti- 

 mations of absolute viscosity the variation in speed should not 

 exceed 1 in 1000 along the tube. 



It is probably preferable to read the temperature by a 

 thermometer inserted in the viscous liquid, using a burette 

 of sufficient bore to render the insertion safe. But good 

 results have been obtained with the thermometer in the 

 water-jacket, tied close to the inner tube ; this arrangement 

 serves quite well when the water-jacket is fairly constant 

 in temperature. 



An observation with glycerine cannot be called good if the 

 temperature-variation exceeds '03° C, and any that are made 

 with greater variation must be discounted. This conclusion 

 is forced upon one when the periods of falling are taken for 

 the same sphere twenty or thirty times in succession. The 

 curve plotted, connecting the temperature with the time of 

 falling, is a straight line within a range of 1° C, and it is 

 generally found that the points farthest from the straight line 

 correspond to the greatest variation of temperature during the 

 fall. Such a curve supplies us with the means of determining 

 the correction necessary to reduce a speed at one temperature 

 to that at another. 



To ensure that the falling sphere shall remain in the axis of 

 the burette, the burette and its jacket are mounted vertically 

 on a stand fitted with levelling-screws, such as that of an 

 ordinary Jolly's balance. The mercury is let down into the 

 burette by a small funnel. The speed is determined by 

 observing the time taken to pass from one mark to another. 

 The distance taken was generally about 50 centim., measured 

 correct to ^ millim., the time taken to traverse that distance 

 being from 100 to 200 seconds. The time was measured 

 with an ordinary watch beating fifths-seconds, or more 

 recently with a Siemens chronograph. In either case the 

 masses of mercury were so chosen that the probable error in 

 a single observation was 1 in 1000. 



The following results are for glycerine at temperatures of 

 18°*28 and 4°*35 C. The former agrees closely with a value 

 obtained by interpolation from Schottner's results. The 

 latter would appear to be considerably higher than the 

 corresponding interpolation. 



