Researches in Absolute Mercurial ^ Thermometry. 93 



the steam is slightly fluctuating, and that when it is accidentally high 

 the temperature rises, but when it is low, stiction prevents the thread 

 from falling, so that the ultimate effect is to make the thermometer 

 indicate too high by an amount equal to K. There is no evidence in 

 support of such a fluctuation. It is at least equally probable that 

 stiction is actually of smaller importance at the higher temperature, 

 where the distillation of mercury, which is known to take place from 

 the free surface, must assist the formation of the normal meniscus. 

 Mr. Sworn's method of reducing thermometric observations depending 

 on the complete disappearance of K at the boiling point, cannot there 

 fore be accepted without further evidence, but the matter is one well 

 worthy of careful investigation. Mr. Sworn was perfectly right in 

 saying that the three readings, viz., freezing point, temperature, and 

 boiling point, ought to be taken under like conditions of the meniscus, 

 but the proper way of accomplishing this is to alter the usual practice 

 of fixing the zero by substituting a method similar to that of deter- 

 mining the freezing points of solutions. If the water is first slightly 

 undercooled, and then brought to the proper temperature by the intro- 

 duction of a few ice crystals, a great improvement in zero point deter- 

 minations would be efl'ected. 



Mr. Sworn's comparisons between thermometers of different com- 

 position were carried out with great care, and may be considered 

 reliable, as far as the instruments used are concerned, but it is not 

 quite certain in how far diff'erent thermometers purporting to be made 

 of the same glass may diff'er. Thus the majority of the Jena glass 

 thermometers carefully studied at Berlin showed a difference of over 

 0*01° at 50° when compared with the French hard glass,* but one 

 instrument agreed throughout its range with the latter, while another 

 differed in the other direction. Marek, at Vienna, did not find 

 any systematic difference between the French and Jena glass, and 

 Mr. Sworn's thermometers also show a practical coincidence. Mr. 

 Sworn's evidence that his instruments were really made of the 

 16^^' Jena glass rests on the assurance of the maker (Gerhardt, of 

 Bonn), but the anomalous behaviour of two of the Berlin thermo- 

 meters leaves a doubt as to how far blowers are careful to guard 

 against accidental mixing up of different sorts of glass. It is not 

 pv ^sible, moreover, to compare directly the result of Mr. Sworn's com- 

 parison with that of other observers, on account of the difference in the 

 method of reduction, but, as far as I can see, the discrepancy would 

 have been greater if Mr. Sworn had reduced his observations in the 

 way adopted at Sevres and Charlottenburg. The same remark applies 

 with greater force to Mr. Sworn's reduction of the flint glass indica- 

 tions to those of the hydrogen thermometer. He uses Chappuis's 

 numbers for the relation between the French hard glass and hydro- 

 * * Zeits. f. Instruraentenkunde,' vol. 15, p. 43S (1895). 



VOL. LXVI. K 



