(118) 
I have tried to express the values found by a formula, and for 
this I have chosen the following expression: 
PV =a + Bd + yd, }) 
where d== 1/V stands for the density of the gas with regard to that 
at O° C. and 1 Atm. at 45° northern latitude. By means of the 
method of least squares I have calculated @, (2 and y from 16 mean 
values derived from 107 measurements, where the equation which 
is obtained when V = 1 is taken to be absolutely correct and where 
the weight 1 is given to all measurements, while it was taken into 
account that the normal volume had been determined by means of 
REGNAULT’s coefficient of expansion a = 0.0036613, which does not 
agree with CHappuls’ coefficient of tension / = 0.0036626. If the 
latter is taken to be correct, we obtain the following table, for 
which the mean errors have been derived by means of the weights 
from the mean errors of the observations to be given subsequently. 
Value. | Weight Mean error, 
2 | 1.072,58 | 6,909. 0.000,003 
B | 0.00066 | 6,914. 0.000,003 
y |  0.000,000.98 10,421,200. 0.000,000,08 
| | 
The following table gives the densities measured at different dates, 
the product PV corrected for the correct coefficient of expansion and 
its deviation from the values calculated from a, /? and 7. 
1) As appears from a development in series borrowed from Amagar’s results, the 
term with d5 would be 0, while the term with d* has little influence below 60 atm. 
