360 
NATURE. 
| [ April 13, 1882 
Now, the carbon being a substance of high resistance, 
a very small change in its length will tell on the potential 
of the returning current: its effective length, however, 
varies with the level of the mercury, and the object in 
view is to measure the movement in the barometer by 
the potential of the returning current. And, in the first 
place, what is the theoretic sensitiveness to be expected ? 
Taking the conductivity of copper as 100, that of car- 
bon is about 0'07 ; and supposing eight miles of copper 
wire in circuit (barometer being four miles from observa- 
tory), and that a wire of one-eighth inch diameter be 
used; supposing, also, that the carbon filament be of 
one-fiftieth inch diameter ; then the following is the result 
ment used at D indicates the barometric height at A when 
the galvanoscope is brought to zero. 
Before describing the apparatus used at D, it is neces- 
sary to explain how the question of temperature is dealt | 
with, Copper has its resistance increased by about o-4 
per cent. for each rise of 1° C. above 20° C. ; and as the 
temperature along the four miles traversed by the wires | 
is wholly unknown, some means must evidently be found 
for allowing for errors from this source. The problem is 
to do this without necessitating extra wires to baro- 
meter A. 
It is obvious that if the barometer could be thrown out 
of circuit before each observation, and the resistance of 
the eight miles of circuit independently balanced in the 
bridge at D, then, restoring the barometer, a second 
arrived at:—For a rise of the mercury of one-fiftieth of 
an inck, the resistance is lowered 1-455th. Closer read- 
ings would probably be questionable, owing to capillarity. 
I would observe, also, that the 0’07 applies to graphite in 
general; I do not know what exactly may be the resist- 
ance of the carbon thread lately come into use. 
In order to measure these changes of potential in terms 
of the barometric height, the whole circuit is treated at 
the observatory as one resistance in a Wheatstone’s 
Bridge. Thus, in Fig. 1 let A be the distant barometer, 
B and C battery and galvanometer in the observatory, D— 
also in the observatory—a means of altering ad 6. the 
resistance in the second circuit of the bridge. The instru- 
Fic. 3. 
determination made at D would yield results which might 
be dealt with quite independently of the resistance of the 
wires. 
This can be effected by sending a reverse current 
| through the circuit immediately before each observation. 
The apparatus shown in Fig. 2 explains this. The figure 
is an elevation showing the upper part of the barometer. 
The iron wire I and the carbon C are shown in position. 
The galvanometer placed immediately in front of the tube 
|is contrived to deflect the current from the barometer 
when the current traverses it in one particular direction. 
It will be best understood, if the action of the current be 
considered in detail throughout the operation of reading. 
We desire, in the first place, to find the resistance of 
| the circuit independently of the barometer. 
