202 ANNUAL OF SCIENTIFIC DISCOVERT. 



ordinary scales of temperature, but are of an arbitrary value, differing 

 for each instrument, and requiring separate tables for each thermome- 

 ter to convert the scale readings into degrees, the divisions at all 

 parts of the scale being equivalent to equal volumes, although their 

 length may vary very considerably. The freezing points were deter- 

 mined in the ordinary way, by immersion in well-pounded ice, from 

 which the water is drained off as it melts. The boiling points were 

 determined by the apparatus devised by M. Regnault, in steam, whose 

 elastic force is exactly equal to that of the atmosphere at the time, 

 a correction being made for the difference in the barometric pressure 

 from the adopted standard pressure. The boiling points, besides being 

 determined for the usual position of a thermometer with the stem 

 vertical, were also observed in a similar apparatus with the stem in 

 a horizontal position ; so that if the instrument should ever be used in 

 any other than the vertical position, the proper correction might be 

 applied. The difference between the boiling point of a thermometer 

 in the two positions, was found to be from 0-2 to 0-5 Fahr., accord- 

 ing to the thickness of the glass and the form of the bulb. After the 

 graduation of a thermometer had been completed, its accuracy was ex- 

 amined by a subsequent culibration with a longer column of mercury. 

 If the length of the column with reference to the scale divisions was 

 everywhere the same, the graduation was considered good ; but if any 

 alteration was found to exist, a more complete examination was made 

 by using columns of different lengths, each of which was nearly an 

 aliquot part of the range of the scale, the remaining errors being 

 deduced from these measurements by the method adopted by Mr. 

 Sheepshanks, for the thermometers used in connection with the na- 

 tional standard yard. It was, however, seldom that any appreciable 

 correction was found to exist. It had long been known that the freez- 

 ing point of a thermometer is not constant, but rises by a considerable 

 amount during the first year after its construction. There was, how- 

 ever, another peculiarity in the thermometers, which was less known. 

 If a thermometer, after having been for some weeks exposed to the 

 ordinary temperature of the air, were placed in melting ice, its freez- 

 ing point would be, for example, 32-2 ; if the bulb were then put for 

 two or three minutes, into boiling water, and soon afterwards again 

 placed in ice, the reading would no longer be 32-2, but would have 

 fallen to nearly 32-0 ; if, in a day or two, it were again placed in ice, 

 the freezing point would have risen a little, about 0< 1 ; and if again 

 tried after two or three weeks, the freezing point would be found to 

 have acquired exactly the original position of 32-2. This had been 

 found to be the case with every thermometer examined at Kew, what- 

 ever was its age : the difference in the freezing point before and after 

 boiling being about 0-17 Fahr., and varying inappreciably indifferent 

 instruments. This peculiar displacement of the freezing point seemed 

 to be owing to a temporary alteration in the dimensions of the bulb, 

 caused by a considerable change of temperature ; the glass, after hav- 

 ing been expanded by heat, requiring a week or two to contract to its 

 original size. It appeared, therefore, that the alteration in the freez- 



