RESEARCHES IN THERMOMETRY. 587 
that the effect of atmospheric pressure may count for about ‘2 in the zero’s 
ascent. 
Another mode of investigation consists in ascertaining the alteration of the 
zero after immersing the bulb in water boiling under different ordinary 
barometric pressures. I have made a large number of experiments with 
thermometers 2 and 3 in my collection, and have obtained the following 
mean relations :— 
Thermometer. 
i) 
3. 
Approximate value of 1 atmosphere in degrees, 1:31 0:84 
The change of zero due to fluctuation in barometric changes is so slight 
that I do not place much reliance on these numbers, and am not disposed to 
recommend the method. Ecen has also reported* the effect of artificial 
pressure upon a bulb, the bulb in this case lying in the interior of a flexible 
case, to which a tube containing mercury, and capable of adjustment at various 
angles for pressure, was securely attached. The following are his numbers :— 
Mean Elevation on Scale. 
Height of Mercury Ww 
. s epee Gh 
in inches. Found. Calculated. 
O15 26°6 30°9 
The ascent so far is proportional to the pressure. 
It appeared advisable to investigate this law under better conditions and a 
greater range of pressure. Fortunately for this purpose the well-known 
apparatus devised by ANDREWS for studying the effect of pressure upon gases 
is available without alteration for thermometers. All that is necessary is to 
substitute for the gas tube in a duplex apparatus the thermometer which it is 
desired to compress, the readings in the air tube (corrected as prescribed by 
ANDREWS) give the pressure. I am indebted to the kindness of Professor Sir 
WituiAM Tuomson, D.C.L., F.R.S., for a loan of this instrument. 
The thermometer (455) I employed had a cylindrical bulb. Its readings 
were corrected by its calibration table, and the subtraction of the mean scale 
reading before and after compression. The temperature rose only ‘05 during 
the experiments. The first eight numbers given below are the means of four 
closely accordant observations, two of which were made with increasing and 
* Loe, cit. dol, 
