Viscosity of Water. 519 



(o) The values obtained for m were constant over a big- 

 range of pressure; and at a very high pressure there was an 

 indication of an abrupt change in the value of m, or in the 

 nature of the flow. The velocities at this pressure were much 

 below the critical velocities for the various tubes, but were 

 all above the lower limit of critical velocity. 



(6) Consistent values for the viscosity of water at 50° C. 

 were obtained in cases where the kinetic energy correction 

 was as high as 60 per cent, of the viscosity. 



I have much pleasure in acknowledging my indebtedness 

 to Professor Sir J. J. Thomson, Cavendish Laboratory, Cam- 

 bridge ; Professor Lyle, Melbourne University ; Professor 

 Pollock, Sydney University : and Mr. G. H. Knibbs, 

 F.R.A.S., Federal Statistician, formerly Director of Technical 

 Education, N.S.W., and Lecturer in Surveying, University of 

 Sydney, for valuable assistance during the progress of this 

 research, which was commenced at the Cavendish Laboratory, 

 Cambridge, and completed at the Sydney University. 



Note on the Viscosity of Solutions. — The viscosity of certain 

 lithium chloride solutions was determined with the glischro- 

 meter described in the previous paper. The only novel 

 feature of the measurements was the automatic recording of 

 the time of flow. The inner platinum wires at b and c 

 (fig. 1, p. 503) were connected by insulated wires, also the 

 inner wires at a and d. Wires were fastened to the outer 

 wires at a, b, c, and d, and were connected to four plugs on 

 a double reversing key. The two remaining plugs were 

 joined by wires through a battery and one of the electro- 

 magnets already described. With the key in one position, 

 there was electrical communication between the battery and 

 electromagnet and the outer c on the one side, and the outer 

 b on the other side. The circuit was complete only when 

 the solution filled the spaces at both c and b. With the key 

 reversed, the battery was connected to the outer d and the 

 outer a, and the circuit was complete when the solution 

 filled the spaces at a and d. By regulating the amount of 

 solution in the glischrometer the signals could be made as 

 short as necessary, at the beoinnmo- and end of the flow from 

 h to L, or in the opposite direction, and the time of flow 

 could be read off accurately on the tape. 



The following set of readings will be sufficient to illustrate 



