OF ARTS AND SCIENCES. 47 



smaller interior glass vessel was packed with sawdust, the first outer 

 bath of salt and ice being retained. Instead, however, of filUng the 

 innermost vessel with the liquid nitrous oxide alone, the space was 

 first filled with fine copper turnings and filings. These occupied so 

 much of the volume as considerably to reduce the amount of the liquid 

 ^.,0 required, and at the same time by their good thermal conduc- 

 tivity tended to maintain a uniform temperature throughout the mass 

 without stirring. The device was found very satisfactory in operation. 

 The licpiid N.,0 was obtained by the rapid discharge of a cylinder 

 somewhat highly charged with the gas. The cylinder was one of 

 those in which this gas is commercially distributed for use as an 

 anaesthetic, but was more highly charged than is usual. 



In the comparison of the alcohol and air thermometers, the condi- 

 tions were necessarily rather unfavorable for high precision, but the 

 arrangements adopted are believed to have given mean results correct 

 to 0^5 C, or (probably) less at the lowest temperatures reached, as 

 high a degree of precision as can be relied upon with the alcohol ther- 

 mometer. The difficulty and expense of using a large bath of liquid 

 nitrous oxide precluded the total immersion of the thermometer in the 

 bath at low temperatures, and as it was thou2,lit desirable to work 

 under the same conditions as to stem exposure at all temperatures, the 

 same general arrangement was used throughout. The bulb and an 

 inch or two of the capillary were therefore the only parts of the 

 thermometers immersed, the remainder of the stem being surrounded 

 by a thin glass tube of about an inch in diameter, filled with water, 

 which, when properly stirred, rendered the temperature of the stem 

 uniform, and measurable (by an auxiliary thermometer). The two 

 alcohol thermometers passed side by side through the same stem bath, 

 and had their bulbs nearly in contact with each other and with the 

 air bulb in the cold bath. 



The scale of the alcohol thermometers was supposed to be normal, 

 i. e. of lines so spaced as to separate equal volumes of the capillary, 

 and to give approximate degrees by direct reading. It was also sup- 

 posed that the spacing of the degrees was determined by the use of 

 ice, and by comparison with a standard mercurial thermometer at one 

 or more temperatures, probably above 0° C, or by some equivalent 

 process, so that one degree of the scale at or near 0° C. corresponds as 

 closely as may be with 1° C. as ordinarily defined on the mercurial 

 thermometer, or, more precisely, with the degree as measured with 

 the Baudin standard mercurial thermometer. It was obviously of no 

 special interest, tlierefore, to study further the temperature value of 



