The luminous source employed is a gas burner with reflector. This is
placed in a walnut box in order to prevent any projection of heat upon
the balance. This burner, thus isolated, is lighted for but one or two
minutes at a maximum, at the end of each weighing. So, on fixing a
thermometer in the cage, we find that no variation, ever so slight,
occurs in the temperature. In order to effect a weighing, the gas being
turned down to a taper, we proceed as with an ordinary balance until the
extremity of the needle no longer emerges from the lower dial. Then we
count the difference of the number of the divisions made by the needle
to the right and left of zero. This difference, multiplied by the
approximate value, in milligrammes, of each division of this dial (value
given by the instrument) immediately gives the number of centigrammes
and milligrammes that must be added to the weights already placed upon
the pan of the balance in order to obtain an equilibrium, to about a
half division of the lower dial.

The value of each division of this dial varies from 3 to 10 milligrammes
according as the balance shows 0.1 or 0.5 milligramme. As the dial has
10 divisions on each side of the central mark, we thus estimate, without
tentatives, the three last centigrammes or the last decigramme,
according to the sensitiveness.

At this moment the doors of the cage are closed, in order to prevent
draughts of air, the gas is turned on by means of a regulating cock, and
the balance is manipulated by first lowering the beam and then bringing
the pans to a standstill. We then read the difference of the divisions
traversed to the left and right upon the luminous dial through the image
of the reticule. The images are reversed upon the dial, but practice
soon causes this petty difficulty to disappear. This number of divisions
indicates the number of milligrammes and fractions of a milligramme by