46 REPORT— 1840. 
A. Thermometers* . 
23. Fired Points.—Legrand+ has studied once more the 
vexed subject of the rise of the thermometric zero point. He 
finds that the pressure of air has no influence, thus confirming 
the observations of Bellani; the molecular change in the form of 
the bulb is therefore the remaining cause, and this is confirmed 
by finding that it depends on the nature of the glass; not oc- 
curring (according to the author) in Crystal. Now if by Crystal 
is meant, as I believe to be the case, Flint Glass, this obser- 
vation is at variance with that of other observers, flint glass 
being generally used for thermometers in this country. Since 
writing the last report, I have myself found the following dis- 
placement of zero in thermometers all warranted standards by 
the following makers—Troughton and Simms(two), Adie, Crich- 
ton, Collardeau :— 
+ 0°°56, + 0°°33, + 0°41, + 0°54, + 0°35, Fahrenheit. 
24. Correction of Scale. When the points are fixed and 
the thermometer graduated, the degrees may be examined by 
the method of Bessel{. Rudberg has proposed another founded 
on similar principles§, and I have given an account of a method 
employed by myself, and attended, I think, with considerable 
practical advantages, in the introduction to a paper on the Tem- 
perature of Hot Springs]. 
25. Every one must have noticed the difficulty of reading ther- 
mometers quickly and correctly when the tube projects much 
in front of the scale, owing to an evident error of parallax, which 
it requires some experience to avoid. A particularly ingenious 
method of correcting it has been communicated to me by M. 
Valz, the eminent astronomer of Marseilles, which though un- 
published, I trust he will forgive me for mentioning. By plun- 
ging two-thirds of the diameter of the tube into the material of 
the scale so that the plane on which graduation is made is ad- 
vanced in front of the mercurial column to the amount of one- 
third of the radius of the tube, it may be shown that the error 
of refraction will exactly correct the error of parallax. 
26. For self-registering thermometers for maximum tempe- 
ratures, such as are now in demand for experiments on deep Arte- 
sian Wells, we have the overflowing principle of Cavendish and 
* See last Report, p. 208. Mahlmann, p. 25. 
t Comptes Rendus (Paris), iv. 173. 
t Berzelius, Jahresbericht, xv. 70, quoted in Poggendorff's Annalen, xxxvii. 
376. 
§ See a full illustration of Bessel’s method in Kupffer’s * Instructions,” p. 5. 
|| Philosophical Transactions, 1836, p. 571. 
