ON THE SPECIFIC HEAT OF WATER. 215 



In the above formula; for MI and M a the coefficients of mi and of m a and m* 

 remained constant over a long period of time, but small changes in the constant 

 1 1 THIS took place owing, perhaps, partly to contraction of the glass and partly to the 

 working out of minute bubbles of air. In order to avoid errors arising from these 

 causes the resistances were constantly re-calibrated at points in the neighlxmrhood of 

 those at which the resistances were actually in use. This re-calibration was usually 

 carried out immediately after the conclusion of an experiment in which the resistances 

 were used, which had the additional advantage of eliminating any correction for the 

 effect of changes of lalx>ratory temperature on the exposed stem of the thermometer 

 tube of the resistance. Examples of such re-calibration may be seen in the details of 

 experiments set out in Appendices A and B. 



Definition of the Mercury Heater. One of the difficulties involved in any system 

 of calorimetry by electric heating is the definition of the limits of the heater. The 

 heater necessarily is connected with leads which are also heated by the passage of the 

 current, and as the leads conduct heat as well as electricity, it is always difficult to 

 say at what precise point the resistance is to be reckoned as belonging to the leads 

 and not to the heater. There must always l>e a certain debatable portion at the 

 junction of leads and heater, and the endeavour must be to make the resistance of 

 the debatable portion negligible in relation to that of the heater within the limits of 

 accuracy desired. 



For the purpose of heating from one temperature to another the problem is slightly 

 different from that of heating for continuous flow. For instance, in the design of the 

 present heater the diameter of the mercury leads was chosen so that with a current 

 sufficient to heat the contents of the calorimeter at the rate of 1 C. per minute the 

 heat developed in the mercury leads should be about sufficient to raise the tempe- 

 rature of the leads at the same rate. The resistance of a portion of the leads could 

 be then included or not in the resistance of the heater, the only effect being to alter 

 the capacity of the calorimeter. 



But for the purpose of continuous-flow experiments the leads remaining at a steady 

 temperature might still serve as a source of heat to the contents of the calorimeter. 

 Referring to fig. 1, showing the position, of the obturator and the water level in 

 continuous flow, it was decided to include the portions of the mercury leads above the 

 water and up to the baffle plate F (each portion being about 1 cm. in length) as part 

 of the heater. The heat generated in this portion and the portion next above it 

 would be disposed of partly in the air space and partly by conduction to the water. 

 As the resistance of the mercury leads was only 0'00025 per centimetre of length, and 

 as the total length of the portions of the leads above the water and up to the baffle 

 plate was about 2 cm., their total resistance would be about 0'0005 on a total of 

 about 10 ohms. We estimate that the possible error due to the inclusion or exclusion 

 of the debatable portion cannot exceed 1 in 10,000. It may be noted in passing that 

 the element of heating due to conduction of heat from the obturator along the 



