(3) Temperature correction chart. Barometers ML- 1 0 2-D, ML-
1 0 2-E, and ML-3 1 6/TM have a temperature correction plotted for
each instrument individually. Specific instructions for the applica-
tion of temperature corrections to Barometer ML102E are given in
paragraph 11. Specific instructions for the use of the temperature curve
or Barometer ML-102 -D or ML-316 /TM are given in paragraph 13b.
c. Scale errors. (1) The dials of Barometers ML-102 -B, ML102 E ,
and ML-102-F have linear graduations; that is, the spacing between
each of the 0.02 inch graduation lines is the same, The aneroid cells
used in these barometers do not expand and contract linearly with
changes in pressure, but by selection of cells and individual adjust-
ment of the lever system the movement of the indicating pointer may
be made approximately linear with changes in pressure. There is
always some residual scale error, however, which varies slightly at
different arts of the scale. Small differences in the characteristics of
individual aneroid cells cause variations in the value of this scale error
among different barometers. It is necessary, therefore, to calibrate
barometers individually to eliminate scale error.
(2) Barometer ML-102- E has been calibrated for scale error and
the corrections are included in the temperature correction curves
fastened to the back of each barometer (fig. 5) . Thus no separate
correction can be applied for scale error; it is automatically included
when temperature corrections are made.
(3) Barometers ML-102-B and ML-102-F do not have either tem-
perature or scale error correction curves, hence pressure measurements
made with either of these instruments probably are less accurate than
those made with Barometer ML-102 -E, ML-102 -D, or ML-316 / T M .
The error in indication introduced by neglecting scale and temperature
corrections will vary among instruments and at different pressures. On
the average, this error probably does not exceed 0.7 millibar.
( 4 ) The dials of Barometers ML- 1 0 2- D and ML- 3 1 6/ T M are
individually calibrated for each instrument. This reduces scale error
to an amount which need not be considered in making pressure
measurements.
d. Drift. (1) Another error to which the barometer is subject is drift
(often called creep), which is due to changes in the aneroid cell occur-
ring slowly over a long period of time. Drift is caused by molecular
changes in the metals of which the cell is made, and by alterations
in the shape of the cell, due to the tendency of all materials to assume
a new permanent shape when placed under steady stress. In meteoro-
logical textbooks, drift is sometimes referred to as "secular change,"
for it is an error taking place over a long period of time.
(2) Drift is manifest by a gradual increase in the difference between
the indications of the aneroid barometer and a mercurial barometer
with which it may be compared. Errors of drift can be largely elimi-
nated by setting the barometer to indicate the pressure shown by an
accurate mercury barometer.
(3) Some barometers drift because of very minute leaks in the
evacuated cell. Usually a leaking cell is detected before the instru-
ment leaves the manufacturer's plant, but occasionally a leak can
develop in a barometer that has been in use some time. The indica-
tions of a leak are progressively lower readings than are normal for
18