234 SMITH : 



Preliminary Tests of the Accuracy and Sensitiveness 



OF the Method. 



Some blank tests were first performed to test the accuracy 

 and sensitiveness of the apparatus. The first of these was 

 carried out in shade in the laboratory, and was designed to find 

 out what reading of the galvanometer scale corresponded to 1° 

 C. difference in temperature between the two junctions : one 

 junction in air, the other in rubber tube in water; whole 

 apparatus in the laboratory. 



Table I.— January 16, 1906. 



Deflection 



Average deflection per degree, C •20. 



Thus, within a wide range of temperature, the readings of 

 the galvanometer agree in registering the temperature at • 20 

 scale divisions per degree C. of temperature difference between 

 the two junctions. An attempt was made to find out if the 

 junction itself when exposed to full sunlight was raised above 

 the temperature of the surrounding air by absorption of heat. 

 For this purpose readings were taken with the apparatus on 

 the verandah of the laboratory. It was set up as before, and 

 the whole apparatus exposed to direct sunlight. A piece of 

 pith, about J in. wide, was then interposed so as to cast a shadow 

 on the junction in the air, and the effect noted. The pith was 

 removed and replaced several times. 



In Table II. at first some shght difference (c/. readings at 

 11 . 45 and 11 . 50) of about • 5 C. seems to be caused by shading 

 the junction in air, but even this is probably due to the fact 

 that changes of temperature taking place in the water and the 

 air are not quite coincident in their effects on the junctions, 

 for when air temperature and water temperature have settled 

 down to a steady state, it is seen in both tables that the shutting 

 out of direct sunlight from the junction by the pith or vice 



