The magnetic radiometry technique relies on the principle that nearly
all materials generate a secondary magnetic field (induced magnetism)
when exposed to a strong primary magnetic field.  The Earth's magnetic
field represents this strong primary magnetic field; thus, all
materials on Earth with a non-zero magnetic susceptibility exhibit
some degree of induced magnetism.  A magnetometer is an instrument
that can be used to measure the sum total of the Earth's magnetic
field plus the induced magnetic fields of all the surrounding
materials. Differences in magnetic susceptibility (related to
different materials, i.e., stone, mud brick, loose sand, or soil) will
result in varied strengths of induced magnetism that can be measured
with a magnetometer.

A magnetic radiometer is a unit with two independent magnetometers
that may be placed one above the other in the vertical gradient mode.
The signal from the first (upper) is subtracted from the signal of the
second (lower) to yield the magnetic gradient at each station.  This
has several advantages: (1) The magnetic "signal" from deeper (i.e.,
non-archaeological) sources tend to be reduced; and (2) The effects of
magnetic (solar) storms and diurnal changes in the Earth's magnetic
field are minimized.

The magnetic survey works by detecting magnetic susceptibility
variations among different materials in the subsurface.  The primary
materials in the survey area were loose sandy soil and sandstone
blocks used for building construction.  The typical magnetic
susceptibility of soil can vary depending on soil type and soil
moisture and is often hard to predict, but dry, sandy soil should be
between 10^-4 and 10^-5.  The sandstone blocks should have a typical
magnetic susceptibility around 3 x 10^-5. Therefore prior to data
collection an anticipated magnetic signal from the ruins was thought
to be possible, though not guaranteed.

A Geometrics Inc., G-858 MagMapper cesium-vapor gradiometer attached
to a non-magnetic cart was used for the magnetic survey.
The vertical separation of the magnetometers was 0.75 m for all zones
and the lower sensor was 30 cm above the ground surface.  Data were
collected every second while moving the cart along the profile line,
resulting in one data point roughly every 0.5 m along the profiles.
Measured marks were input to the data logger by the operator in order
to accurately locate the data points in space during subsequent data
processing.