T apprehend at eared cha: btt- acesdentalbiifet 0 barometers, made 
even by the same person but at different periods, or from any cause 
containing mercury of different ies; agreetl j in fhe: specific grav- 
ar as nearly as this. 
Finally, not to. make: too denig dintuasiies of anal point, it is 
- even rare to find two observers, in direct succession upon ihe 
same instrument, reading to a coincidence as close as the thou- 
sandth of an lipsli I thought myself therefore justified in dis- 
carding a graduation which, as I before said, multiplies the medians 
ment without increasing the certainty. 
Nevertheless, in limiting the tube-graduation to tenths of inches, 
Ihave not shut out the means ’of.subdivision to thousandths in a 
ready and unexceptionable. manner; whenever such subdivision 
should be requisite. These means consist in the application of 
a reading microscope or micrometer, whose position on the tube 
is held and: regulated by a spring and clamp. ‘The zero is ad- 
justed, first, to the now magnified image of the mercury surface : 
and then, by the motion of the screw, the space between said sur- 
face and the nearest division onthe tube below, i ‘is measured in 
hundredths of inches by the comb of the micrometer, and thou- 
sandths on the head of the screw in a well-known manner. 
6. A piece.of watch-spring, about th hes in length, is bent 
round at one end, so as to embrace three quarters of the cireum- 
ference of the tube: the other end, which will then project from 
the tube, has merely a light triangular, notch..made-im its ,upper 
edge. to catch the | loop of string or wire, 
‘mometer. in order to prevent any: unsteadiness or wriggling 
motion, w ly a strip, a 
seiabiaaniae th, from the : spring g, is fastened with a 
single rivet at right angles. to the former, so as to be vertical or 
peervee ai meee the poles is made ; jor, as was.done in the present . 
nstance is cut out of a wide clock-spring. - 
sil Nothing more need be said of the float than that it is of ivo- 
dd eeat: thin:and light. as i consistent ris its safety; and 
t surface and the parallel fidu- 
cial edge of the rectangular notch seen in it, must be exactly equal 
with the space between the zero mark on the tube and that other 
line next above the zero mark, which has been spoken of already 
in $4. It is obvious that in such case, when the image of said 
fiducial edge seen by reflection on the tube, coincides with the 
