36 XOTES ON SOME BEMAEKABLE EEEOES 



diiference increased to 13 "3 degrees, and then fell again. On the 

 3rd May, and again on the 7th of May, two sudden jumps occurred, 

 and the difference rose to 13 "7. Since then it has gradually fallen, 

 except a slight rise May 21, 22. {See diagram.) 



In the first instance the barometer was steady at about 30 

 inches, with some minor oscillations, and there has been no baro- 

 metric oscillation during the whole time which might be supposed 

 equal to produce such effects, and even if there had been, it should 

 have affected all the thermometers alike. At present I can see 

 no satisfactory clue to the explanation of these thermometer 

 excursions. Once before when using a standard, in April 1872, 

 to test another thermometer, the temperature of the water was 

 raised to 210°, and it was found that subsequently, for four days, 

 the standard itself read too low by several tenths of a degree 

 (see Appendix B), and then recovered its normal condition. 

 Here the cause was evident ; the glass had expanded, and did not 

 contract as fast as it cooled. "We often find differences also in the 

 readings given by thermometers of known good quality, but these 

 are attributed to difference in sensitiveness arising from thick- 

 ness of glass, or other causes ; where the changes are sudden, as 

 in thermometers on the grass exposed to the effects of radiation, 

 the difterences sometimes amount to several degrees. (See 

 Appendix C.) That common thermometers give results difiering 

 by several, and in some cases as much as ten degrees, is well 

 known to all who have had much to do with them, and that 

 the glass continues to contract for 3'ears after it has been 

 melted is beyond question ; for the makers keep thermometers 

 two or three years before graduating them, and even then in 

 many cases they go on contracting. But all the thermometers 

 about which I have made these notes are of the best description, 

 and the comparisons made in England before they Avere sent out 

 have not been taken without re-examination. All the ther- 

 mometers we use are compared in air with the standard here. 



On examination with the microscope, the dry bulb in question 

 presents two features which must be mentioned ; one is a small 

 piece of coloured glass as if lead had been reduced in melting the 

 bulb ; the other is a Jittle patch that looks like water inside the 

 bulb. As the bulb is blown by the workman's lungs, this maybe 

 condensed water in the bulb, from that source. "Whether this 

 can have anything to do with its uncertain readings I do not 

 know, but it is possible that some action may take place between 

 mercury and water under a vacuum. 



One other point must be mentioned. Fine glass, such as 

 that used for lenses, if kept untouched for some months and then 

 examined with a microscope, will be found covered with oily- 

 looking specks, which evidently ooze out of the glass, or are 

 formed by the action of moisture on some of its constituents. 



