158 Scientific Proceedings^ Eoijal Dublin Societi/-. 



the two flasks change in temperature at the same rate, owing to alteration in 

 temperature of the surroundings, errors will be introduced should the 

 external temperature change during the course of an experiment. In our 

 later experiments on tlie inversion of sucrose the external temperature 

 remained constant within about 0'1° C, so that the error introduced was 

 negligible. But owing to the sensitiveness of the temperature measurements, 

 it was found that slight inequalities in the radiation to which the two flasks 

 were exposed gave rise to considerable errors. To avoid this, the apparatus 

 was enclosed in a heavy copper cylinder, covered with a copper lid. This 

 was placed in a wooden cask, and packed around with cork dust. 



The water-equivalent of the experimental flask was found to be 9"5 grams. 



Radiation Correction. 



The radiation correction was found to he affected very largely by the 

 method adoj^ted for stirring the flasks. In our earlier experiments the 

 glass tubes of the thermocouple, with the capsules attached, passed loosely 

 through holes in the oorks which closed the necks of the flasks, and were 

 attached to a cross-piece which was free to slide on a vertical rod between 

 the two flasks. A light piece of cane was attached to the cross-piece, and 

 passed through a hole in the copper lid of the containing vessel. By ' this 

 means the thermocouple and capsules could be moved up and down, thus 

 stirring the liquid in the flasks. 



This method of stirring suffers from the disadvantage that the radiation 

 error is very greatly increased owing to loss of heat from the wet surfaces 

 of the tubes when they come out of the flasks, and was therefore abandoned. 

 The method finally adopted was to clamp the glass tubes between split corks 

 in the necks of the flasks, and to effect the stirring by shaking the flasks. 

 The cask containing the apparatus was slung in a liorizontal position from 

 a bar of wood which rested on a central pivot, and could thus be rocked 

 backwards and forwards. By this means the rate of cooling during stirring 

 was decreased to less than one-third of its former value. 



The correction for radiation was then determined. In a preliminary 

 experiment 315 c.c. of water, at about 90" C, was put into each flask, and a 

 sensitive thermometer inserted through the cork which closed the neck. 

 The flasks were then placed in a shaking-machine, and the rate of cooling 

 determined. Assuming Newton's Law of Cooling, the rate of cooling at any 

 temperature = the excess of that temperature over the surroundings x a 

 constant. This constant, which we may call h, was calculated from the 

 formula : — 



- A; log e = -- log 



