LORD RAVLKI(4U OX A NEW MAXO.METER 
' 2-20 
pressures of Tables 111.— \MI. tlio mean error is less than ‘0004 millim. And it 
must be remembered that tlie numbers to which these errors relate are the means of 
two observations only. 
As a means of dealing with very small pressures, the sloping manometer has 
proved itself in a high degree satisfactory, the performance being some tvrenty-five 
times better than Amaciat’s standard. It could hardly have been expected that the 
mean error would prove to Ije less than one wave-length of yellow liglit.'^'' Considered 
as a pressure, the mean error corresponds to the change of barometric pressure 
accomi)anvino; an elevation of 4 niilhms. 
1^/0 
On hydrogen more than one series of observations have been carried out. The 
S 2 )ecimen that will be given is not in some respects the most satisfactory, hut it is 
chosen as having l)een pursued to tlie greatest rarefactions. The gas was dried 
carelully with phosplioric anhydride and was introduced into the apparatus as already 
described. It is tliought sutHclent to record oidy numbers corresponding to the 
three last columns of Tables I.—YIl., the hrst column giving the pressure in 
minims, of mercury, the second the deviation of pr from the mean value of the 
set taken as unity, the third the eriur in p from what would Ije rerpilred to make 
2 )U a])Solutely constant. 
* I had at one time contemplated an apparatus from which a fnrilier ten-fold increase in accuracy might 
he expected. Two beams of light, reflected nearly perpendicularly from the mercury surfaces, would be 
brought to interference by an arrangement similar to that used in investigating the refractivity of gases 
(‘Proc. Roy. Soc.,’ vol. .59, p. 200, 189G; vol. 64, 23- 97, 1898). Preliminary trials proved that the 
method is feasible; luit the delicacy is excessive in view of the fact that according to Hertz the jiressure 
of mercury vapour at common temperatures itself amounts to '001 millim. 
