1022 
pressure exerted by the liquid inside the tube on the “window”, 
should be borne by the side-walls We therefore gave the window 
the shape of a truncated cone, the basis turned to the side of the 
liquid, and the smaller plane parallel to the basis quite free and 
turned to the observer. The conical wall must be ground as carefully 
as possible into a steel cylindre with conical opening, which is 
screwed into the steel tube of observation. Fig. 2, where B is such 
a steel cylindre, and A the glass cone, will probably make this 
sufficiently clear. If the cone A is ground with sufficient care, and 
then fastened in B with a little cement, an absolutely tight closure 
is obtained in this way; we have never experienced any trouble 
owing to leakage between glass and steel, nor has any of the glass 
cones ever burst in consequence of the high pressure, in such 
a way that the liquid could be pressed through those windows. 
We did meet, however, with other difficulties. First of all the 
difficulty of getting pieces of glass from which the required glass 
cones could be obtained, without too much loss of time. We 
first tried to start from thick plates of plate glass (8 cm. thick), 
but it appeared impracticable to saw or cut off such small pieces 
(1.5 to 2 em? area) *), that they could serve for further preparation. 
We then applied to the “Stichtschen Glashandel” at Utrecht, which 
prepared octogonal rods for us, about 6 em. long and ofa diameter 
of about 1:5 em. from the best plate glass. After these rods have 
been cut off doubly conical (two cones of 3 em. the bases resting 
1) When the conus is ready the basis turned to the liquid has a diameter of 
12 m.m., the other end-plane a diameter of 10 mm. 
