; the Thermal Conductivity of Water. 59 



The amount to be added to the temperature T of the sink to 

 give the temperature at which k is measured is 



R.H. vessel 0-26(0! + 0/) -0-139 8, 



L.H. vessel 0-27(0! + 0/) -0-139 8. 



(6) The pressure of the water against the glass bottom of 

 -A alters with the thickness of the water layer; and as the 

 glass is thin it was feared that it might give. The reading- 

 microscope showed that it did so, but so slightly that the 

 observed value of k is only 0*1 per cent, too high. This error 

 has been allowed for in the final mean, but not in the tables 

 of results. 



Results. 



In taking the final readings, after the flow of water through 

 the lower vessels was turned on, the heating current and 

 balancing current were started, and after an hour's time 

 readings taken on the bridge-wire every 10 minutes until the 

 position of the contact-maker was steady during a quarter of 

 an hour. The readings of the thermopiles JK, JF were then 

 taken and one of the micrometer screws turned, altering the 

 thickness of the water in one or the other vessel. After about 

 three quarters of an hour the position of the contact-maker 

 was again observed at short intervals until no further change 

 took place, the thermopile readings being then repeated. 



The water thickness was then in most cases reduced to its 

 original value, and the readings on piles and bridge-wire de- 

 termined afresh ; their means with those originally obtained 

 being used in calculation*. 



Throughout each experiment the electrodynamometer was 

 observed every 10 minutes, and the current kept constant if any 

 small alteration had taken place by the adjustment of a 

 rheostat. The temperature of the water flowing through the 

 sinks was measured just before entering and just after leaving 

 the cast-iron containing box (it rose about a tenth of a degree 

 in the process) and the mean was taken as the temperature of 

 the sinks. 



To obtain a greater variation of temperature in our experi- 

 ments we tried heating the sink-water before it entered the 

 apparatus by passing it through a metal spiral connected with 

 a burner and thermostat. But the frequent slight alterations 

 of temperature resulting from this caused such large fluctua- 



* In certain cases marked with an asterisk in the table of results this 

 rerueasurement of the zero values was omitted. In taking the means 

 given in Table II. these are reckoned as half experiments. 



