MARYLAND GEOLOGICAL SURVEY 445

than that of the single fibre. Of the 46 stations there was but one
station—Damascus—where the error due to uneliminated torsion
may amount to several minutes. But this station is in a disturbed
region and additional observations might well be added in this locality,
even if the accuracy reached is not more than one-tenth of a degree.
It would seem, therefore, that we have no reason for supposing that the
observation error of the magnetic portion of the determination of a
declination should in general have exceeded the reading error of the
horizontal circle of the magnetometer.

A consideration of the reduction error follows next. This consists
of several parts:

a. The diurnal variation.

b. The disturbance variation.

e. The secular variation.

d. The annual variation.

In reducing the observation made at a specified time to some other
time, all of these factors enter in, and all need to be taken into account,
with the exception of the last, which for stations in mid-latitudes has
a total range of only about 1'. The general method of procedure
amounts practically to reducing the observation to the mean of day,
making the necessary allowance in case the observation appears to
have been made at a magnetically disturbed period, and then applying
the correction for secular variation. To carry out this scheme as per-
fectly as possible it is necessary to have near the base of operation of
the magnetic survey a -magnetic observatory where by photographic
means a continuous record of the variations of the magnetic elements
is obtained. It is necessary that this observatory shall be sufficiently
close to the area surveyed so that it can be assumed that the diurnal
variation as observed at the observatory is practically the same over
the entire area. The diurnal variation progresses according to local
time, and hence it is not meant by the statement " that the diurnal
variation is to be assumed the same over the entire area, " that at the
same instant of absolute time the correction for diurnal variation is the
same over the whole area, but that at the same instant of local time
the correction is assumed the same. The disturbance variation on