1903 - 4 .] Thermal Expansion of Solutions of Hydroxides. 283 
it is desired to obtain a value for the relative density of a water, 
which shall be as nearly correct as possible to the fifth decimal 
place, the use of a counterpoise for automatically eliminating 
certain incidental errors is absolutely essential.” 
From calculations I made, based on a paper by Dr G. J. Parks * 
“ On the Thickness of the Liquid Film formed by Condensation at 
the Surface of a Solid,” it was found that had the pyknometer 
had maximum deposition of moisture in the one case and none at 
all in the other, the difference between two weighings of the 
pyknometer empty could not exceed - 004 per cent. Parks found 
that the thickness of the film of moisture deposited on the surface 
of the glass after 16 days’ exposure, when the maximum was 
attained, amounted to 13'4xl0 -6 cm. This moisture if all 
present would increase the weight of my pyknometer by ‘0008 
gms., which is equivalent to -004 per cent, of the weight of the 
pyknometer empty. 
The difference I am dealing with is not the absolute amount of 
moisture deposited, but the change in the amount of moisture that 
may occur from experiment to experiment, and therefore is much 
less than that calculated above. 
To find whether it was necessary to use a counterpoise or not, 
when I wished an estimation of density which should have no 
greater error than 5 in the fifth decimal place, I determined the 
specific gravity of a solution at various temperatures, both with 
and without the counterpoise. 
I took two pyknometers of the Sprengel-Ostwald type, of the 
same kind of glass and of nearly the same external volume. I 
weighed each one, reducing the weight to weight in vacuo. One 
of the pyknometers was then sealed by closing the end of the 
tube of large bore and melting the end of the tube of small bore 
till it was almost closed. 'I 'lie whole pyknometer, except about a 
quarter of an inch of the capillary tube, was immersed in a beaker 
of water, and the beaker covered with layers of paper to prevent 
the heat of the sealing flame reaching the water. The pyknometer 
was left in the water till the air inside had reached the temperature 
of the water, and the capillary end was sealed with a fine small 
flame. Knowing the temperature of the water, the height of the 
* Proc. Phys. Soc. Lond., 18 , 410 , 1903 . 
