600 PROF. J. J. THOMSON AND MR. O. F. C. SEARLE ON THE RATIO OF THE 
A 500 c.c. flask was filled with water and weighed, its contents were then trans¬ 
ferred to the cylinders, and it was then weighed again. The difference in weight 
gives the weight of water transferred to the cylinder. This process was repeated 
until the space between the cylinders was full. 
The weights of the full and empty flasks were determined to 1 centigrm. 
The 500 grin, weight used to balance the water was compared with the standard 
500 grm. weight of the Laboratory and found too heavy by '055 grm. This has been 
allowed for. The equality of the arms of the balance was also tested. 
The two cylinders were fastened down to a flat metal plate with a thin layer of 
cement so as to be quite wader-tight. To get any accurate estimate of the volume of 
water required to fill the space it was necessary to provide some means to ascertain 
when the space was exactly full. The effects of capillarity, grease, kc., preclude any 
very accurate result being obtained when there is no top or cover fitted to the top of 
the cylindei’, and as it was necessary to see whether any air bubbles were left inside a 
glass top had to be used. Two holes were bored through the glass, and tubes were 
fixed into these. The water was introduced through one tube, and the air escaped 
through the other. Although no difficulty was experienced in makmg the joint at 
the bottom quite water-tight with any of the cements employed, it took us several 
days to make a satisfactory joint at the top. The ease with which tightness at the 
bottom w^as secured was probably owing to the great weight of the cylinders. 
The top gave us all the more trouble, because we could not tell whether it was 
watertight or not until we had almost completed the filling in of the water. If, then, 
the joint proved bad the whole of the time spent in weighing the water poured in 
was wasted. 
We tried Front’s elastic glue, then gutta-percha dissolved in benzene, but both of 
these failed. The water seemed to loosen the hold of the glue upon the glass, so that 
although the system seemed air-tight it would not remain water-tight for more than a 
minute or two. 
We finally tried some red wax which had been sent to the Laboratory by Professor 
Threlfall, who obtained it in Germany, and this answered very well. It looks 
somewhat like a mixture of bees’-wax and sealing wax, and as it nev^r gets quite 
hard it never cracks. It possessed another property which was also useful for our 
purpose, viz., that of melting at a comparatively low temperature. To apply the 
other cements in a sati.sfactory manner the glass had to he heated to a somewhat 
high temperature, and this frequently cracked it. 
The wax when melted became very fluid, so that only an extremely thin layer was 
included between either the top or bottom plate and the cylinders. 
At first the water was poured in through a funnel inserted into one of the tubes, 
but with this arrangement it was fimnd impossible to get rid of the last air bubble, 
since fresh quantities of air were continuall}^ carried down the tube. After some 
experiments wdth different arrangements for filling, we finally adopted as the means 
