390 SCIENTIFIC RECORD FOR 1882. 



gen found the tension at 0° = 0.015°"", Eegnault. 0.02 ; at 20^, Ilagen, 

 0.021, Eegnault, 0.037; at lOOo. Hagen, O.Gl, Eegnault, 0.75; at200o, Ha- 

 gen, IG^ Eegnault, 19.9™™. Though the values now found have no claim to 

 absolute accuraey (owing to the difBculty of taking readings with the 

 €atlietometer through round glass), tliese experiments at least make 

 certain that the Eegnault values for the vapor-tension of mercurj', which 

 have passed into all text-books, are considerably too large. {Ifature, 

 XXVI, p. 108.) 



Professor Eucker, in a paper " On the calibration of mercurial ther- 

 mometers, by Bessel's method," stated that the late Mr. Welsh, of Kew 

 Observatory, described to the British Association, in 1853, the methods 

 which he introduced of making and correcting mercurial thermometers. 

 The correction with which the author dealt was that due to the variations 

 in the bore of the tube. Mr. Welsh's method of making this correction, 

 which is still employed at Kew, is lesfj theoretically perfect than others, 

 and has been unfavorably criticised abroad. The author, in conjunction 

 with Professor Thorpe, has recently corrected a number of thermome- 

 ters with great care by Bessel's method. One set of three thermometers 

 was made for them at Kew and calibrated according to Welsh's method. 

 Afterwards the measurements necessary for the application of Bessel's 

 method were made by the Kew authorities, the calculations being per- 

 formed by the author and Professor Thorpe. The Kew thermometers 

 were thus subjected to the most rigorous possible test, and they were able 

 to announce that in one instrument the errors left after the apiilication 

 of Welsh's method were not greater than four-thousandths of a degree 

 centigrade, and in no case did they exceed one-hundredth of a degree. 

 As it is impossible to read on these thermometers less than a hundredth 

 of a degree with certainty, Welsh's method, as applied at Kew, is prac- 

 tically perfect. {Nature, September, 1881, xxvi^ p. 407.) 



The alteration of the zero of thermometers after undergoing sudden 

 changes of temi^erature is a well-known phenomenon, as is also the 

 gradual rise in the zero in thermometers during the first few months 

 after they have been made. M. Pernet has lately examined the ques- 

 tion whether the distance between the "boiling i)oint" and the "freez- 

 ing point" of a thermometer is constant at all tliS'erent stages of secular 

 alterations in volume of the bulbs, and finds that this is so, provided 

 the freezing point be deterinined immediately after the boiling point. 

 On the other hand, if the boiling point be determined and a long inter- 

 val elapse before the zero is determined, there is considerable error. 

 Suppose a thermometer to be (owing to recent heating or to long rest) 

 in any particular molecular state. In this state its reading will i)rob- 

 ably be in error ; but this amount (so far as due to the above cause) 

 imay be ascertained by immediately plunging the thermometer into 

 ice, and observing the error of the zero reading. In order that a ther- 

 nioiiieter should read rightly at any particular temperature it should be 

 exposed for a considerable time to the temperature for which exact 



