of its difference from the first of these two stated diameters, to 
their own difference. The length of the cone by this process is 
4035700 inches, a very small fraction too short (when the cury- 
ature is considered) for perfect compensation; but the resulting 
errors are inappreciable. 
In the proposed compensated syphon, the interior capacity and 
mercurial contents are intended to be the same as if the tube 
chamber will accomplish this object by imparting the required 
elevation at the expense of diameter. But this modification of 
€ upper end of the long arm changes its relation to the pre- 
viously cylindrical short arm of the syphon, and A apes that 
he short arm should be an inverted fac simile of the conical 
am 
dergo any change of vertical dimensions. In effect, the full 
cubic contents of the conical vacuum chamber will occupy the 
being equalled by the space occupied by mercury in the short 
arm of the syphon. 
he tendency of the surface of the mercury to assume a 
spherical form with increased elevation may require a trifling 
adjustment. Another adjustment may also be necessary on ac- 
As ; 
conical space which least diminishes its sphericity. 
At the temperature 32° F. the lower limit of utility of the 
Inches, the upper limit being 32 inches. he point at which 
the tube Sant be self-compensating, if cylindrical, corresponds 
to the pressure 27-483489 inches at 32°. Above that point the 
error has the same sign as in the ordinary barometer; below 
- The point of self-compensation for a cylindrical tube corres- 
ponds to the section of mean diameter, or rather, diameter of 
mean sectional area of the conical cha 
J. Lewis on the Barometer. 237 
