580 MR EDMUND J. MILLS’S 
the air, and kept always in a vertical position, would always exhibit a descend- 
ing zero; the descent in the first stage being due to expansion and set as before, 
but, in the later stages, simply to the weight of mercury acting upon a bulb now 
rendered partially plastic by heat. I cannot find that other investigators have 
left any record of such an experiment. The following trials were made with 
thermometer 20, which had been open for five years, and whose zero had been 
long undisturbed :— 
TABLE IX. 
Zero (Therm, 20). Scale, Depression. 
Before experiment +'262 C. 0-000 
After 50° 134 128 
B 102 —°058 320 
a 150 203 "465 
4 200 360 622 
i 250 547 809 
= 310 ‘494 “756 
i 350 570 832 
The first seven observations were made consecutively on one day, the eighth 
nearly two days later. The temperatures are not corrected. With the excep- 
tion of the seventh number, which comes out a little too low, this experimental 
series is in accordance with theory. 
IV. PoGGENDORFF’s CORRECTION. 
So long ago as 1837* PoacEenporrFF showed that the mercurial thermometer 
requires a correction to the following effect. The instrument is completely 
heated to 100° in order to obtain the upper point. At any other temperature, 
say 50°, the stem is no longer at 100°, and therefore one-half of the original 
distance (0°—100°) does not correspond accurately to 50°, but to a somewhat 
higher temperature. If 8 be the coefficient of cubical expansion of glass, the 
correction is made by multiplying the observed reading ¢ by the fraction 
1+ft_. 
1+f100 
The effect of this correction is by no means inconsiderable. At 300° it amounts 
to nearly 2°+| When, moreover, this correction is employed, the maximum 
difference over the range 0°—100° between the mercurial and air thermometers 
is found to lie, not at 50°, as has been hitherto supposed, but near to 34°,—that 
is, sixteen units nearer to the axis of differences. 
* Pogg. Ann. xii. 472. {+ RecknaceEL, 7bid, exxiii. 130 (1864). 
