6 C. Barns — Change of Heat Conductivity on 



could be measured independently ; or by soldering the german 

 silver wires together and coupling the ends of the copper wires 

 with the galvanometer, differences of temperature could be 

 measured. The latter method was adopted since this differen- 

 tial quantity enters the formulas. Great care must be taken to 

 see that the charge is electrically a non- conductor. This is the 

 case for thymol, but series of experiments which I made with 

 water proved worthless because of this discrepancy. To cali- 

 brate the thermo-couples in question, duplicate sets of identical 

 wires (metal and dimensions) were at hand. The galvanome- 

 ter of my own make showed about 3 scale parts per degree C, 

 and it was a periodic and nearly constant in sensitiveness. 



7. Heating and cooling. — There are two methods available 

 for heating the system of plates D and E uniformly through- 

 out, to be used respectively, when the environment is of high 

 temperature or of low temperature. 



(1.) To heat the plates to the temperature of the (hot) envi- 

 ronment, the stop-cock IT, is closed relatively to 6r, and. the 

 efflux water from f led by jacketed rubber tubing, through the 

 cork J? into (}' '. The water thus fills i^and issues at h. Its 

 temperature is read off at q. DD and. EE soon reach the same 

 temperature within -01° C. 



When the actual measurements are to be commenced, water 

 from the hydrant is first passed through H and K, until its 

 temperature is constant. The cock R is now loosened, and at 

 a given signal ZTis suddenly opened into G. The great bulk 

 of water now entering EE, forces out R, and floods the ther- 

 mometer reservoir below g. Here therefore temperature is 

 again registered by q. 



(2.) In case of the cold (2°-6°) environment, hydrant water 

 is passed directly through the tubes a to /', but the method of 

 heating D, Eis less simple. If warm water be passed through 

 R, as before, the temperature of E would even after long wait- 

 ing exceed that of D. For plates 'l cm apart this persistent 

 excess amounted to say '9° C, varying with the distance between 

 the plates, their internal and external conduction, etc. The 

 discrepancy may, however, be obviated by using two supplies 

 of warm water, one of which is a few degrees (3°-5°) hotter 

 than the other. The hotter water* is first passed through R 

 and EE. After the lapse of sufficient time the second hot 

 water supply of constant but lower temperature is made to 

 replace the other. Thus EEis slightly cooled at once, whereas 

 DD only cools very gradually, but must eventually fall to a 

 temperature below E Observations are therefore commenced 



* Fletcher's " instantaneous water heater," furnishes a satisfactory circulation, 

 for the temperature of this water need not be very constant. 



