ON THE MEASUREMENT OF TEMPERATURE. 
iniiiation the normal scale of temperature of the Comite International 
Kammerliugh Onnes (‘Leiden Communications,’ 1901, No, 60) gives ^ 
0-0036627 as the value of the coefficient, the actual numbers being ^ 
0-0036624, 0-0036628, 0-0036628. The result obtained by Regnault, p 
0-0036613, need not now be considered. ^ 
It has hitherto been assumed that the pressure coefficient for ^ 
hydrogen is independent of the initial pressure, and does not, as in ^ 
the case of nitrogen, decrease and at length attain a limiting value. 
As important deductions have l^een based on tliis statement, we 'p 
decided to measure the coefficient for our standard gases at initial 
pressures corresponding to 350, 500, and 700 millims. of mercury. 
An approximate measurement of the pressure coefficient of helium 
was made by Kuexen and Randall in 1895 (‘Phil. Mag.’). In 
employing the helium thermometer for the measurement of low 
temperatures, Olszewski assumed that the coefficient is the same 
as that of hydrogen. 
4. Accuracy of the Results. 
In order to determine the fifth significant figure in tire pressure 
coefficient, the total error must not exceed 1 part in 40,000, a degree 
of accuracy which it is practically impossible to attain in individual 
experiments. As Avill lie seen later, the greatest error in our 
observations may be attributed to inaccuracy in the scale wliicli, o\'er 
the whole length, may be considered accurate to at least O'Ol millim., 
or to 1/25,000 of the increase of pressure on the gas between 0° and 
100° C. at an initial pressure of 700 millims. 
5. Apparatus employed in the Research. The Barometer. 
Since at the boiling-point of water a change of pressure of 0-1 millim. 
of mercury corresponds to a change of temperature of 0-0036° Cl, or 
1/20,000 of the difference between the boiling and freezing-points of 
Avater, it was sufficient to obserA-e the atmospheric pressure Avith this 
degree of accuracy. The liarometer employed is shoAvn in fig. 1 ; it Avas 
constructed by us specially for tliis research. The upper chamber A 
Avas 20 millims. in diameter and 200 millims. in lencrth. It was sealed 
to the stem B, AAdiich Avas draAvn out at the loAver end C to a capillary 
tube of an internal diameter of about 1 millim. In filling the barometer, 
the capillary end of the tube B aaxis sealed to a T tube (fig. 2), of Avhich 
the branch D led to the mercury pump, and the brancli C projected 
