42G Mr. H. M. Tory on the 



5. Method of Heating. — In order to get an accurate com- 

 parison, only steady temperatures were used. For this 

 purpose, the pyrometers, enclosed in a porcelain tube, were 

 heated in an iron pot of molten tin, and the gas-supply regu- 

 lated by means of a constant-pressure gas-regulator. The 

 gas also flowed through a graduated gas-supply regulator, 

 so that, with a little patience, almost any temperature between 

 400° and 1000° C. could be obtained. The furnace used was 

 one whose maximum was 1000°, so that the observations were 

 not taken beyond that point. The molten tin was continuously 

 stirred by means of an air-engine and stirrer. It was found 

 necessary to make the stirrer of fire-clay or tool-steel. When 

 iron was used, the action of the tin at its surface, on the iron, 

 soon rendered it useless. To prevent oxidation of the tin, 

 which took place very rapidly when the surface was exposed, 

 and as a result soon filled the pot with tin oxide in a semi- 

 viscous condition, various expedients were used. The one 

 which succeeded best was to keep the surface covered with a 

 stratum, half a centimetre thick, of ground arc-light carbon. 



The greatest difficulty experienced was due to the action of 

 the tin on the poi'celain tubes. They would not stand repeated 

 heatings in the tin bath, but became very rotten. Several of 

 the observations were spoiled by the tubes breaking at the 

 moment of removing them from the molten tin. It was 

 sometimes difficult at first to remove the pyrometer-tubes, 

 because of the oxidized tin on the surface. After the difficulty 

 of oxidation was overcome, less trouble was experienced, but 

 the tubes were never used more than two or three times with- 

 out becoming so affected at the junction with the surface of 

 the tin, as to make it undesirable to use them again. In taking 

 the melting-point of silver, the tubes which had been pre- 

 viously used for only one set of observations in tin, generally 

 broke when removing them from the melted silver, though 

 the greatest care was exercised. 



6*. Results of Observations. — Four different samples of 

 ordinary platinum wire were compared with a very pure 

 specimen in general use in the laboratory as a standard, be- 

 tween 400° 0. and 1000° C, the two extremes of the furnace 

 used. The. temperature-coefficient of the standard wire was 

 found to be slightly different in different pyrometers. These 

 variations were due doubtless to the lack of absolute uniformity 

 in the wire, but were not sufficiently large to affect the read- 

 ings of temperatures at 1000° by more than 1/10 of a degree. 

 The difference-coefficient d of each wire in the formula (2) 

 was calculated from an observation of the boiling-point of 

 sulphur in the usiial way. 



