8

Geophysical Surveying of
Archaeological Sites
JOHN W. WEYMOUTH and ROBERT HUGGINS

ABSTRACT
In this chapter the two most commonly applied geophysical methods for asses sing archaeological sites are described in detail.
Magnetic surveying responds to contrasts in the magnetic properties of soils, which can be brought about by, among other causes, human
activities such as burning, by humic decomposition, by compaction, and by the introduction of structures. Resistivity surveying responds to
differences in the electrical conductivity of soils, which can be brought about by, among other causes, alterations of the natural soil
profile through the construction of mounds or ditches, by compaction, or by the introduction of structures. Magnetic surveying is independent of
soil moisture but will not respond if a structure is composed of the same magnetic material as the surrounding soil. Resistivity surveying is
strongly dependent on soil moisture and on the contrast in porosity between a structure and its surrounding matrix.
Magnetic surveying is faster and easier to interpret but cannot easily be carried out near interfering magnetic sources, such as
modern buildings or power lines. Resistivity surveying is slower and somewhat more difficult to interpret but is free from the interference
of nearby buildings and power lines.
Like all remote sensing methods, those described here are nondestructive and are considerably more economical than test excavations, if they
are properly conducted. A geophysical survey carefidly coordinated with an archaeological program can provide valuable information for the
planning and execution of that program, as the examples given for both techniques demonstrate.
The last 25 years have seen many applications of geophysical survey techniques in archaeological site surveying. The methods usually emerged
from small-scale field experiments by physicists and geophysicists; they were then applied pri-

192 / J o in t W . Wey mo ut h an d Ro bert H u g g i n s

marily by European archaeologists. In the last decade, however, there has been
a growing awareness among American archaeologists of tile value of geophysio
methods in site location and mapping.
The various geophysical techniques for gathering information about subsu rface features all depend, in one way or another, on differences in electric
magnetic, or elastic (seismic) properties of rocks and sediments. The technique
may be classified as passive or active. In the first category, existing force fields a
measured directly without instrumentally generated signals and the results are
interpreted in terms of subsurface features perturbing the field. Magnetic
thermal, and gravity measurements fall in this category. In the second or active
category, instrumentally generated signals pass through the subsurface and are
then detected and recorded. Seismic techniques, electromagnetic technique
(including use oldie simple metal detector, the pulsed-induction metal detector
and the soil conductivity meter), earth resistivity measurements, and the re cently developed ground-penetrating radar are all active devices. This chapter
will deal with one technique in each category: magnetic and earth resistivity
surveying. For summaries of most geophysical methods see Aitken (1974) or Ti
(1972).
MAGNETIC SURVEYING
SUMMARY OF METHOD

Magnetic surveying (or prospecting), as practiced on archaeological sites, con

lists of measuring the magnitude of the earth's magnetic field at each point on
grid established over the site. Variations in the ma g net ic properties of the
subsurface material (sediments, rocks, or artificial materials such as brick) can
produce an observable variation (anomaly) in the measured magnetic field.
Anomalies may he caused by artificial structures such as walls, ditches,
foundations, fire hearths, pits, or even an area of more intensive habitation. The task
of interpretationto separate the results of human activity from geological
variations in subsurface materialsis guided by knowledge of the physics of so
magnetization and by the manipulating and displaying of the data in various
ways so as to reveal significant patterns. At any site, successful application
of the method depends on the magnetic properties of the local subsurface, the
external and nature of the human activity, the burial depths of artificial and
natural features, and, filially, the care taken in field measurement and analysis.

HISTORICAL DEVELOPMENT

For several decades geophysicists have used magnetic surveying In the search for
minerals, but until 20 years ago the instruments available were not sufficiently
sensitive for archaeological applications. The development of the proton
magnetometer provided just the needed sensitivity. Belshe (1957) seems to have
been

Geophysical Surveying of Archaeological Sites / 193

the first to experiment with a proton magnetometer in an archaeological context.
He was followed by Aitken, Webster, and Rees (1958), and soon the Oxford
University group under Aitken was obtaining results of archaeological
usefulness. Subsequently, groups developed in Germany (Scollar, 1961), Italy
(Lerici, 1961; Linington, 1964), and France (Hesse, 1962) . The literature in this
field is now fairly extensive. Reports of technical aspects, until recently,
have been c onc en trated in f ou r journa ls : Archa eomet ry, Pro spe zion i
Arche olo giche , Revue darchometrie and Archaeo-Physika. The
literature on applications is scattered throughout a number of geophysical
and archaeological journals; extensive references may be found in Aitken (1974)
and Tite (1972).
In the United States, an early practitioner of magnetic surveying was the
NIASCA group at the University Museum, University of Pennsylvania, under
Rainey (Rainey and Ralph, 1966) and Ralph (1964; Ralph, Morrison, and O'Brien,
1968). Their applications for the most part were in the Old World, as was the
survey work by Rapp and Henrickson (in McDonald and Rapp, 1972). To our
knowledge the first recorded application on this continent was at Angel Mounds
by Black and Johnston (1962). Subsequent applications in this hemisphere are
documented in Ezell et al. (1965), Morrison, Clewlow, and I leizer (1970),
Greiner and Coe (1972), Arnold (1974), Nashold (1977), and von Frese (1978).
The MASCA group has conducted some magnetic surveys in the eastern United
States (Bevan, 1975).
The present writers recently started magnetic surveying on sites in the
Central Plains (Weymouth, 1976) and, in conjunction with the Midwest
Archaeological Center (National Park Service) and other agencies, have
surveyed or analyzed data from surveys covering approximately 33 hectares in
about 10 states and a few sites outside the hemisphere (Weymouth and
Nickel, 1977; Weymouth, 1979; Bleed et al., 1980).
In the last few years several groups and individuals throughout the United
States have started using magnetic surveying on sites. In one case a group from
Michigan State University has covered a fairly large area (Mason, 1981).

THEORY

The magnetic field at any point on the earth can be defined, for our purposes, as
the direction taken by a compass needle freely suspended there. The direction
can be specified in terms of declination, the angle between true north and the
horizontal component of the earth's field, and inclination (or dip), the angle
between horizontal and the direction of the total field. The field strength or
magnitude is proportional to the maximum torque exerted on the compass
needle by the field. In this chapter the unit of magnetic field strength used is the
gamma (1 gamma is also equal to 1 nanotesla, the SI unit). The earth's magnetic
field in the United States varies from roughly 49,5(X) to 59,500 gamma, and its
inclination varies from 60 to 75 below the horizontal.

194 / John W. Weymouth and Robert Huggins

Geophysical Surveying of Archaeological Sites / 195

conversion from one form to another are the significant factors of soil magnetization.
Two measures of the response of a material to magnetization are its
magnetic susceptibility, which is the ratio of magnetization (dipole strength per
unit volume) to magnetic field strength, and its specific susceptibility (dipole
strength per unit mass per unit field). This latter quantity is measured in emu per
grain. (For definitions of units see Aitken, 1974:140.)
Typical values for specific susceptibility are as follows (all in units of 10'
einu/g).
Limestone, some unbaked clays
Subsoils
Topsoils
heated soils, fired clays

10
50-100
100-1,000
1,000-2,500

Another aspect of the magnetic field with which we must be concerned is its
variation with time. In a regular diurnal variation the magnitude decreases
during the middle of the day by approximately 20 or 30 gamma from higher
morning and evening values (fig. 8.1). During magnetic storms larger variations occur
over time periods of a few hours to days.

Natural and anthropogenic activities can cause a conversion from hematite to

magnetite or maghemite, thus resulting in a greater susceptibility. Such conversion processes, explored by Le Borgne (1955, 1960), occur when hematite is
reduced to magnetite either during heating, as in a hearth, or during anaerobic
decomposition in humic soil. Consequently, topsoils usually have a higher
susceptibility than subsoils and hearths, while burned houses and trash pits may
be even more magnetic. For a complete discussion on various aspects of soil
magnetization relating to magnetic surveying of archaeological sites, see Graham
and Scollar (1976).

Magnetic Properties of Soils

Measurement of the Magnetic Field

In the presence of a magnetic field, material such as soils, rocks, and ferrous
objects can become magnetized. Such a magnetization is said to be induced. In
addition to induced magnetization-, which vanishes when the applied field is
removed, some materials exhibit remanent magnetization, magnetization that
persists in the absence of an applied field. Baked clays and some rocks retain a
thermoremanent magnetization after being heated to several hundred degrees
centigrade and then cooled in a magnetic field. Remanent magnetization can also
arise from chemical change or from the settling of small particles in a magnetic
field (see pp. 240-43).

The three instruments primarily used for archaeological surveys are the
proton free -precession magnetometer, the fluxgate magnetometer, and the
cesium or rubidium magnetometer. The proton magnetometer is the least
expensive and by far the most wid ely used. Its sensor consists of a coil surrounded by a hydrogen -rich liquid (water or kerosene). A polarization current
through the coil creates in the liquid a magnetic field many times more intense
than the earth's. This field partially polarizes the hydrogen's nuclear protons,
which are spinning magnetic dipoles. The polarization current is then quenched
and the protons precess (gyrate) in the field of the earth. For the few seconds that
the protons precess coherently, a voltage is induced in the coil. This voltage is
amplified, the frequency measured, and the results displayed in gammas. In
older commercial units the total cycle time may be 7 seconds, but this can be
shortened to 3 or 4 seconds with no loss of sensitivity; the normal sensitivity is 1
gamma, which can be increased to 0.25 gamma in the usual portable models.
Commercial units arc now available with cycle times of 1.5 seconds and sensitivities to 0.1 gamma.

A range of time responses exists between the extremes of permanent magnetization and the very rapid component of induced magnetization. Because the
time response depends on particle sizes in the soil, parts of soils can become
magnetized very rapidly while other parts change their magnetization very
slowly. (This phenomenon of viscous magnetization is explained in detail on pp. 243,
245.)
The compounds in soils which are important in causing magnetization are
hemat ite (a Fe 2 0 3), mag netite (Fe 3 0 4 ) and maghemite Fe90 3 ). The h itter two compounds are much more strongly magnetic than the first, their
saturation magnetization being approximately 200 times that of hematite. Since
soils contain a few to several percent iron oxides, these compounds and their

The fluxgate magnetometer measures the component of the vector field along
the axis of a coil and is thus strongly direction-dependent. This disadvantage can
be overcome by combining two instruments as a gradiometer, which is direction-

196 / John W. Weymouth am! Robert Huggins

Geophysical Surveying of Archaeological Sites / 197

independent. A gradiometer measures the magnetic gradient or difference in

magnetic field at the two sensors. The requirement for parallel alignment of the
detector units is stringent; therefore, the cost of a differential fluxgate magnetometer is high.
In cesium or rubidium magnetometers atomic electrons of these two elements (in a vaporized state) replace the nuclear protons of the proton magnetometer; otherwise the operational principles are comparable. Both fluxgate and
cesium or rubidium magnetometers have a high sensitivity and also provide
continuous measurements of the field, but their costs are greater than that of a
proton magnetometer.
Anomalies Produced by Local Features
An isolated magnetic feature (termed a source) whose dimensions arc small
compared with the sensor distance produces the simplest, so-called dipole
anomaly. A normal dipole anomaly results from induced magnetization (such as in
a small pit), where polarization will be in the same direction as the earth's field.
However, dike magnetization is permanent (such as in a piece of iron, a burned
rock, or a fire hearth), the polarization may be in a different direction than that
of the earth's field and the resulting anomaly is termed a nonnormal dipole
anomaly.
The total magnetic field in the neighborhood of a normal dipole anomaly is
the combination (vector sum) of the uniform, downward- pointing field of the
earth and a weak, dipole field from the source feature. The profile of magnetic
values measured along a south-north line is represented in figure 8.2. Three
characteristics of this anomaly type may be noted:
I. The maximum intensity of the magnetic profile is displaced to the south of
the source by about one-third the sourcesensor distance.
2. The full width of the profile at half maximum is about equal to the
sourcesensor distance.
3. The negative region, due to the source dipole field opposing the earth's
field, is about 10 percent of the maximum intensity. As in the first characteristic,
this is true at midlatitudes.
Deviations of a magnetic anomaly from these characteristics imply a nonnormal dipolethat is, a source with permanent magnetization. Long, narrow
pieces of iron or burned rocks in particular produce anomaly profiles with large
deviations from those of normal dipole anomalies. Thus, if the minimum is not
north of the maximum, or if the minimum deviates appreciably in magnitude
from 10 percent of the maximum, one can conclude that the source is not a
feature resulting from induced magnetization.
The magnitude of the anomaly depends strongly on the source sensor
distance, decreasing proportionally with the inverse (Witte cube of that distance.
The magnitude also depends on source volume, V, and magnetic susceptibility
contrast, k. For further discussion see Breiner (1973), Tite (1972), Aitken (1974),
or Weymouth (1976).

A filled ditch oriented east -west is like a row of parallel dipoles, with a
negative anomaly north of the maximum. For other shapes ditches, plates,
etc.see Aitken and Alldred (1964), Breiner (1973) or Linington (1972).

APPLICATION OF THE METHOD

The method to be used in surveying a site depends on the information sought. If

one wants information on possible linear features, such as ditches or walls of
known orientation, or if the location of small features (such as fire hearths) in
buildings of presumed location are known approximately, then one or a few
magnetometer traverses may suffice. If surveying is done over a short period of
time, it is not necessary to use a second or reference magnetometer; plotting
profiles of such traverses may be sufficient to reveal the information desired.
If simple traverses do not suffice, then the problem is to seek patterns of
anomalies in a two-dimensional mapping of the magnetic field over the site.

198 / John W. Weymouth and Robert Huggins

Mapping is accomplished by measuring the field on a grid of points over the site.
When more than a few minutes are needed to sur vey the area, some method
must be used to correct for the temporal variations of the earth's field. Without
such corrections, the resulting magnetic map will be distorted and spurious
anomalies will appear, particularly along traverse rows. The basic idea behind all
corrections is the assumption that the earth's field changes in time simultaneously everywhere over a region considerably larger than the site being
mapped. To carry out the corrections, several approaches can be used.
If only one magnetometer is available, it alone can be used to correct for
temporal variations. The operator simply intersperses repeated readings at a
single reference station between groups of grid readings for instance, after
each row of points. The reference readings are then plotted against time and a
curve drawn through them. In this way reference values corresponding in time
to each grid value are estimated and subtracted from the grid values. This
procedure, however, can still result in incomplete corrections and spurious
anomalies, particularly linear anomalies along traverse lines.
To survey a site properly, particularly if it is more than a few meters in area,
two magnetometers in either of two modes must be employed. In the differential
mode, one magnetometer is kept at a fixed reference point while the other
measures values at the grid points. Because the instruments are operated
simultaneously, any difference between the two readings represents the total
field magnitude, corrected for time variation, at that point. In t he gradiometer
mode, two sensors are kept a fixed distance apart -1 in, for instance in a
vertical array. Both magnetometers are again operated simultaneously, and the
difference in readings is recorded. These values are gradient magnitudes, the
vertical spatial change of field magnitude. As in the differential mode, the
temporal variations are canceled, but in addition any long-range trends affecting
both sensors equally are canceled, whereas local anomalies (which will more
strongly affect the lower sensor) will be recorded. The writers prefer the differential mode for any extensive survey.
Before starting the field survey of an extensive archaeological site, all visible
geological and archaeological features must be examined. Also, the nature, size,
and depth of features, as well as the possible existence of iron artifacts, should be
ascertained, as this information will shape the field strategy and aid in the final
interpretation of the data. Those conducting the survey should work closely with
the archaeologists, so both parties share an understanding of the archaeology of
the site and the geophysical method.
Some evaluation of the survey's possible success may be obtained by measuring the magnetic susceptibility of typical soil samples from the site. These
samples should be taken at representative stratigraphic levels, as well as inside
and outside any archaeological features of the site. By measuring the susceptibility
of the samples before and after heating in a reducing atmosphere, it is possible to
evaluate expected anomaly sizes and to get some information on the extent of
anthropogenic hematite-magnetite conversion.
The size of the survey grid unit is determined by the size of the features

Geophysical Surveying of Archaeological Sites / 199

expected. Since the number of magnetic field measurements will be proportional to the square of the grid spacing, this choice is a compromise between
detail sought and time available. Generally speaking, the grid spacing should be
comparable with, or somewhat smaller than, the linear dimensions of expected
anomalies. Features of a meter or two in size near the ground surface can be
surveyed with a 1-in grid. If the features are deeper, then the spatial extent of the
anomalies will be larger (as well as weaker) and a larger grid spacing can be used.
On historical sites, linear features such as walls or cellars can be picked up with a
larger unit. One approach on large sites is to use a coarse grid (such as 2 in) and
then concentrate with a 1- or 1/2-m grid in areas of greater magnetic activity.
In choosing sensor height above the surface, two considerations must be
noted. First, since the width of an anomaly increases with the sourcesensor
distance, a greater sensor height will result in less resolution between anomalies
from neighboring sources. An approximate rule is to have the sourcesensor
distance no greater than the intersource distances that one wishes to resolve.
This suggests a sourcesensor distance equivalent to or less than the grid
spacing.
Second, the sensor height must be selected so as to reduce the relative
contributions from surface noise arising from variations in surface-soil magnetization. The noise contribution relative to the signal will decrease with increasing
sensor height. Probably the best compromise is to set sensor height at
between 40 and 60 cm for a 1-in grid.
A necessity in all archaeological work is the location of some permanent
reference points. This is no less true in the case of geophysical surveys. When
surveying a site, particularly if it is large, the area should be subdivided into
squares or blocks, each of which is treated as a unit. We have found that a block
20 grid units on a side is convenient. This area will have 21 x 21, or 441, grid
points and can be surveyed in about 70 to 90 minutes, depending on site
conditions. Including the time needed to lay out blocks and set up equipment, it is
possible to survey three to five blocks ill a day. If possible, these blocks should be
oriented along magnetic north, since this orientation aids in the interpretation of
anomalies. For a large site the blocks can be arranged in hectares, five blocks along
an edge. The coordinates of grid points in each hectare can be designated I north, J
east, where I and J) run from I to 101. This is a convenient size array to be handled
as a single matrix of values in computer programs.
OPERATIONAL DETAILS
The following discussion assumes survey operation in the differential mode, with one
magnetometer sensor moved from grid point to grid point and one kept
stationary for reference readings. We thus speak of moving sensor values,
reference sensor values, and their differences.
Our technique for locating the grid points within a block is as follows. Stakes are
placed at the four corners of the block and two ropes marked with grid units are
placed, one at each end of the block, from stake to stake. Another rope marks the
grid points along a traverse row from end rope to end rope. To avoid

200 / John W. Weymouth and Robert Huggins

confusion and aid systematic data recording, all traverses should be conducted in
the same direction, either south to north or west to east.
A systematic form for recording data in the field should be carefully
planned. The writers use data sheets calling for the following information:
site, block, data, grid unit, and names of persons operating the
magnetometers. In addition, the height of the sensor and the sensitivity of
magnetometers should be noted. Each sheet has space for four rows with 21
grid points and columns for the moving-sensor value, stationary-sensor value,
and their difference, the last in case it is desirable to hand-calculate these
values. The time of day is recorded at the start of each row. This is desirable if
problems of analysis arise and one wishes to reconstruct events at the time;
recorded data can also then be correlated with magnetic storm information.
Errors can arise in the hand-recording of data, so it is preferable to
machine- record data with a data-collecting circuit connected to the
magnetometer. Such a system involves expense and design time but
considerably improves reliability.
Before taking measurements, it is important to evaluate the local magnetic
environment. Vehicles can produce a shift of roughly 1 gamma at 30 in. Such
disturbances are acceptable only if they remain stationary throughout the
survey. Larger and closer amounts of iron, if stationary and not too large,
can be treated by mathematical filtering techniques applied to the data.
Power lines, moving trains, and other nonstationary sources must be avoided.
The person holding the moving sensor must be carefu lly checked
before starting. Steel shoe tips, belt buckles, bank cards with magnetic strips,
and even eyelets in hats can cause trouble. Repeated readings should be
taken with the person assuming several positions relative to the sensor.
Variations in readings should be random and not greater than one or two times
the "least count" of the instrument (the smallest possible change in value
displayed).
Even if the magnetic field is absolutely stationary in time, a random
scatter or noise in reading values of about 0.5 times the least count will
be observed. When two instruments are recorded at the same time and the
difference taken, this random scatter becomes -0.7 times the least count. In
actual practice, the reproducibility of difference values (taken by repeated
readings at the same grid point, repositioning the sensor each time and
calculating the standard deviation) is more like three to five times the least
count. This variation in the difference values is a measure of the least
amount of noise to be expected. Anomalies smaller than this variation may be
lost in this noise unless they are observed on several grid points.
The reference magnetometer should be close enough to the grid points so
that communication between the operators is not impaired but not so close
that the two sensors interact. Two or 3 in is sufficient. It is important not to locate
the reference sensor in a strong magnetic gradient. In such a gradient the
instrument can lose sensitivity and give erratic readings. The location should be
checked this possibility by taking repeated readings, with the sensor
moved slightly between readings. Variations no greater than random noise
indicate lack of a gradient.

Geophysical Surveying of Archaeological Sites / 201

One final and very important field procedure is the measuring of common
rows. Relocating the reference magnetometer between recording two blocks
will result in a constant shift in the difference values between the two blocks.
This may be corrected by measuring one row common to the two blocks twice,
once with each block. The average shift in the differences on the common row is
then the correction factor that is added to all the values of one of the blocks to
bring them to the same "level" as the other block. If the standard deviation in the
distribution of shifts in differences on the common row is greater than 3 to 5 times
the least count of the magnetometers, then the correction factor may not be
reliable. This can arise by careless positioning of the sensor or by passing over a
strong gradient due to an anomaly on the common row. In this latter case small
and unavoidable shifts in sensor positions can cause large shifts in recorded
values. Either a subset of values on the common row that avoids the anomaly
should be used or other common rows should be measured. These problems arc
particularly troublesome with matching blocks recorded on several days on a
large site.
INTERPRETATION
The first requirem ent in the analysis of magnetic data is to produce a matrix of
magnetic field values corrected for diurnal variations to be used in all subsequent
mapping and profiling. Since the number of values from a site can be quite large,
computer processing becomes desirable or even necessary. The following discussion assumes the use of a computer.
Usually, results are presented in various ways so that individual anomalies
and patterns of anomalies can be identified. The simplest treatment is to plot
profiles along traverse rows, this is useful in seeking linear features of known
orientation.
For any more complicated situation it is necessary to examine the areal (twodimensional) patterns of anomalies. Various forms of magnetic contour maps
are generated in which the magnetic field strength is treated as a "height'' on a
map of the site. Although simple contour maps can be hand -drawn by
recording the magnetic field values on a grid of points and drawing lines of equal
magnetic field magn itude through the grid of values, this process can he very
tedious and is not likely to be often repeated if the map needs to be redrawn with a
different scale or contour interval. Thus it becomes necessary to produce
contour maps by a computer.
The most convenient types of computer contour maps are those produced by a
line printer. Such maps are quickly generated and easily interpreted because of
the visual advantage obtained through use of the shading variations of print
characters. They have the disadvan tage of relatively low resolution determined
by the print character size. In the simplest map type the magnetic field values at
the grid points are sorted according to predetermined intervals (usually no more
than 10) and print characters of different shades arc printed at the grid points
according to these intervals. The result is effective for sites of more than one
hectare.

202 / John W. Weymouth and Robert Huggins

The map next in complexity is based on interpolation between grid points and
prints several characters per point. A commercial program package that makes
such maps with a wide range of options is SYMAP (1975). A simpler program to
obtain some of these same results is described by Davis (1973).
The traditional type of line contour map can be programmed to output on a
plotter. Such a map ca n have much higher resolution but lacks the immediate
visual impact of a shaded map. Color contour-mapping equipment is also available, but it is expensive.

EXAMPLES OF MAGNETOMETER SURVEYS

The Knife River Indian Villages National Historic Site
The objective of this National Park Service project is to assess and develop into a
national historic site a collection of Mandan, Hidatsa, and other Native American
village sites grouped at the confluence of the Knife and Missouri rivers north of
Bismarck, North Dakota (Weymouth and Nickel, 1977). The Midwest Archaeological
Center (MWAC) of the National Park Service has been gathering magnetic
survey data at Knife River since 1976, with the present writers analyzing them.
Two magnetometers in the differential mode have been used to collect data,
which since 1978 have been recorded automatically with an electronic data logger.
One of these villages Big Hidatsa (32M E12), occupied in the late eighteenth and early nineteenth centuriescovers several hectares and is marked
today by many depressions left by earth lodges. These show up clearly on the
magnetic record, partly due to the , topography, but also due to differences in soil
susceptibility inside and outside the lodges: some lodges are visible on the
magnetic maps but are not visible in the topography. The most characteristic
anomaly of these lodges is their central fire hearths, which lie at depths of 40 to
120 cm below the ground surface and produce anomalies of 20 to 50 gamma. In
addition, anomalies associated with smaller interior or exterior fire hearths are
visible. Midden areas between houses produce magnetic high regions. These
correlations have been established by soil probes and by test excavations.
Figure 8.3 is a grid-point map of data taken over two seasons (1977 and 1978) at
Big Hidatsa. Each print character corresponds to data on 1 grid point, with an
interval of two grid points between different characters. The map illustrates a
way in which one can organize a large survey. Each set of blocks obtained on one
day has been balanced against other sets by using common rows. Each hectare is
handled as a separate matrix in the computer processing.
Figure 8.4 is a topographic map of Sakakawea Village (32M E 1 1), another
village of the same time period at Knife River. It may be compared with figure
8.5, a SYMAP magnetic map of the same region with a 9 -ganuna interval
(Weymouth and Nickel, 1977). Figure 8.6 is the southeastern part of this map
with a contour interval of 5.6 gamma. The anomalous regions at N21 -31,
E80-89 is due to a house not visible on the topographic map. Two excavations in

Geophysical Surveying of Archaeological Sites / 207

this area established that the anomaly at N36, E69 between the houses was
caused by an exterior fire hearth about 60 cm in diameter and 40 cm below
surface and that the smaller anomaly just to the south inside the house was
caused by a group of burned rocksa presumed sweat lodge.
Figure 8.7 is a magnetic map of house 6 in block 0 in the northwest quadrant,
taken with the sensor at our usual height of 60 cm. Figure 8.8 is a map of the same
area but based on readings taken with the sensor 120 cm high. Some of the
smaller and sharper anomalies produced by near-surface sources disappear in

208 / John W. Weymouth and Robert Huggins

the second map, while the lire-hearth anomaly produced by a larger and deeper
source persists.
Fort Union National Historic Site
This site, a trading post in North Dakota from about 1830 to 1865, has undergone
considerable excavation. The data for the survey were obtained by the MWAC
and analyzed by the present writers.

Geophysical Surveying of Archaeological Sites / 209

Figure 8.9 is a line contour of the total survey area. A number of anomalies
can be seen, particularly along the position of the western range of houses. Some
of these are due to previous excavations and some arise from original wall
foundations and fire hearths. The central area, a parade ground, is relatively free
of anomalies except for the strong anomaly in the center due to a modern flagpole
base. This survey illustrates two points. A strong anomaly at N 19, F.70.5 in the
southeast quadrant of the map (see fig. 8.10) was actually located earlier by
personnel from MWAC who ran a series of seven 20-m-long traverses using a 1-m
grid unit and one magnetometer. On a map of these data (fig. 8. 11) the large

210 / John W. Weymouth and Robert Huggins

anomaly and a number of smaller anomalies are reproduced. Such an

operation is not usually successful on a larger area. Figure 8.12 is a magnetic
line contour map resurveyed in part of the southwestern corner using a
grid interval of 50 cm. Comparison with figure 8.9 shows several anomalies
duplicated, as well as more detail.
Ward and Guerrier Trading Post

The Ward and Guerrier Trading Post is located adjacent to the Fort
Laramie National Historic Site, Wyoming. As part of a program to
construct a parking lot and visitors' center, it was necessary to determine
the location of die trading post that was known to be in this vicinity around
1858. Excavations in 1963 had

Geophysical Surveying of Archaeological Sites / 213

revealed a small, square Inundation (presumed to be a smithy), several hits and

pieces of foundations, a row of portholes, and a sizable trash pit containing iron
material that was not removed at that time. The data for the 1963 excavation,
however, had been lost and only the approximate position of the site was known. In
1976 the MWAC opened up three excavation units and surveyed the area with two
magnetometers, as shown in the magnetic map of figure 8.13 (Weymouth, 1979).
Three observations from these excavations and from the survey led to the conclusion
that the previous excavations had been found again and determined the position
of the main structure of the trading post.
First, the excavation unit and survey information in the easternmost block
established this to be the smithy area with the now rediscovered small foundation. Second, the large anomaly at N31, E31 was assumed to be the iron- filled
trash pit. Third, a row of sharp anomalies (marked by the dotted line in fig. 8.13)
was believed to mark a wall of the main building of the post. The length of this
line is approximately 100 feet, which agrees with what is known of the post. This
interpretation is further supported by a row of post molds at the end of the line
uncovered in the 1963 excavation and indicated in the excavation trench at about
N21, E5. Figure 8.14, a 3 -1) view of the results, again shows the line of
anomalies, the strong anomaly due to the trash pit, and the group of anomalies in
the smithy area.
The Dolores Archaeological Program
The final example is drawn from the Dolores Archaeological Program (DAP), a longterm study of the archaeological resou rces in the Dolores River valley,
southwest Colorado, before its inundation by dam construction. As part of the
resource assessment a large number of sites, each encompassing two to ten
blocks, have undergone magnetic surveying.
Because most surveys in this program have been followed by excavations or at
least blading of the plow zone, it has been possible to obtain rapid checks on the
archaeological significance of magnetic anomalies. This has improved our ability to
interpret data and make predictions. Following the 1979 season, a study was
made of the correlation between predictions made from the magnetic maps and
the results of the various excavations. The pit structures and other large architectural features were of sufficient volume and extent to prod uce anomalies over
several meters that could be readily detected; of 26 such "high -priority" anomalies, excavation revealed that 23 were caused by cultural features.
We show as an example of the larger anomalies results from site 5MT2193,
surveyed in 1978. Figure 8.15, a magnetic map of this site, shows in the western
half two large anomalies that were presumed due to two pit -house structures.
These anomalies extend over several meters but have a maximum of only about
15 gamma. Use of quarter-gamma sensitivity here was justified, but a 2-m grid
spacing could have been used if only this type of anomaly was being sought.
Upon excavation, the sources of the anomalies were shown to he pit-house
structures originally about 1.5 m deep that subsequently were filled in with

216 / John W. Weymouth awl Robert Huggins

burned-roof fall and soil. The shapes of the anomalies followed the shapes of
the sources fairly faithfully, including the air vent in the southern
structure. As a further test of the source-anomaly relation a model
calculation was made, representing the southern pit house by a box
with a hearth on the floor and a sphere for the air vent to the south.
This model is represented in figure 8.16, along with a three dimensional representation of the magnetic results. The susceptibility
values had to be multiplied by four in order to obtain a magnetic map
(figure 8.17) comparable with the original map, which was plotted in
quarter-gamma units.

Geophysical Surveying of Archaeological Sites / 217

RESISTIVITY SURVEYING
SUMMARY OF METHOD

Resistivity surveying on archaeological sites indicates spatial differences in

sediment moisture: the presence of features, architecture, activity areas, and other
archaeological remains can be detected if the amount of moisture they retain is
different from that retained by surrounding sediment. Location of these
anomalies or contrasts involves careful measurement of the sediment resistivity at

Geophysical Surveying of Archaeological Sites / 219

discrete points on the surface along traverses or on a grid of points. The collected
data are usually displayed as profiles or as electrical-resistivity contour maps.
Because of the amount of data involved, a computer is an invaluable assistant.
The data are then computer-enhanced (if necessary) and interpreted, and areas
of potential archaeological interest are located on the basis of a clear understanding of the principles discussed in the following sections.

HISTORICAL DEVELOPMENT
Measuring the resistivity of the subsurface has long been used as a method for
exploring geologic structure. A practical measuring technique was first introduced by Wenner (1915) and significant improvements were made by Schlumberger
(1920). Resistivity measurements are still widely applied in geophysical exploration
fir mineral deposits and gravel beds. An early archaeological use of the method
was in 1947 by II. Lundberg (De Terra, 1947:162- 64), to locate fossil human
remains at Tepexplin, Mexico. Since that time resistivity has been utilized
successfully by Atkinson (1952), Clark (1969, 1975), Carr (1977), Lithe, Schneider,
and Carr (1976), and Ginzburg and Levanon (1977), among others, in a variety of
archaeological contexts. For an exhaustive treatment of the historical development of
the technique, refer to Van Nostrand and Cook (1966).

THEORY
In order successfully to conduct and interpret a resistivity survey, a grasp of basic
electrical theory is necessary, beginning with the nomenclature. Electric current
is defined as the rate of flow of charge passing through a cross section of a
conducting medium for a specific length of time. To cause charge to flow, a
voltage (also known as potential difference, a measure of the energy used to move
the charges) must be applied. When a voltage is applied and a current flows, a
resistance is encountered to the moveme nt of the charge, which is dependent on
the characteristics of the medium in which the charges are moving. These three
physical quantities are related by Ohm's law,

Resistance is measured on Ohms (SI), voltage in volts (V), and current in amperes
(A). In a conductor of length L and cross section area S time voltage difference per
unit length can be thought of as the moving force, the current as the quantity that is
moved, and the resistance as the opposition encountered by moving the
current. From Ohm's law we can develop the concept of resistivity by incorporating into equation (1) the geometry of the medium. Resistivity is a more
useful quantity than resistance in the examination aim archaeological site since

220 / John W. Weymouth and Robert Huggins.

it is specific to the medium and independent of the geometry of the material
being surveyed. Resistivity (p) is defined as

where V/L is the change of voltage with distance in the direction of current flow
and J is the current density in the medium in which charge is flowing. The basic
unit of resistivity is the ohm-meter or ohm-centimeter (1 -m = 100 -cm ). If a
specified current is flowing in a known geometrical shape, we can deduce the
resistivity of the material, providing the voltage difference is known. The inverse
of resistivity (1/p) is known as the conductivity, althoug h in the discussion
following, we will consider only resistivity. More complete discussions of these
concepts are available in most basic physics texts, such as Resnick and Halliday
(1966).
Resitivity Measurements
The concept of subsurface resistivity measurements can be illustrated in an
actual field situation. Current is induced in the ground by inserting in the earth
two metal probes that are connected to a battery. In this idealized case, distribution of voltage and current in a uniform earth is well understood (a model is
shown in fig. 8.18). Also shown are current- and voltage- measuring devices to
indicate both the amount of charge flowing between the current probes and the
voltage in the area of interest between the two potential or voltage probes.
By calculating the volume affected by the flow of current, we can derive an
expression for the average resistivity within the measuring probes:

which is easily calculated because the distance between the probes (a) is known
and the current (I) and voltage (V) are measured quantities. The resistivity is
correct only 16r this particular probe configuration (or probe array), as other
probe geometries change the volume of earth affected by the current flow. In
archaeological applications, the prune concern is not the absolute value of the
resistivity at any one point but the change between readings.
We can alter this simple model to illustrate how the current and voltage are
affected by some inhomogeneity in the uniform earthfor instance, a trash- filled
pit. The voltage and current deviate from the normal pattern and the resistivity
measurement of the earth between the two voltage probes changes (fig. 8.19).
If these measurements are continued by moving all fo ur probes from grid
point to grid point, we can generate a series of readings indicating the lateral
variations in electrical resistivity to a depth approximately equal to the separation between the voltage-measuring probes. By increasing the spacing between
the probes for any given survey, one can examine a greater volume (and therefore depth) of material.

222 / John W. Weymouth and Robert Huggins

Geophysical Surveying of Archaeological Sites / 223

ELECTRICAL PROPERTIES OF SOIL SEDIMENTS

In the following paragraphs only sediments on which soil profiles have developed are
considered, although resistivity measurements may be made in other sediment
types (sand, till, mud) as well. The conduction of current in soils is largely an
electrolytic phenomenonthat is, moisture in a soil containing free charged particles
is responsible for the current flow. The resistance to currents flowing in all soil
types depends directly upon the following variables:
1. Soil moisture content, which at archaeologically significant depths is
usually generated by rainfall, with occasional contributions from areas having
high water tables or from nearby streams. In general, soils receiving little rainfall
have a high average resistivity and conduct electricity poorly. Seasonal variation in
the total amount of rainfall also affects the resistivity (Al Chalabi and Rees, 1962;
Clark, 1975). The amount of water that the soil can contain is determined by soil
porosity, which exhibits wide spatial variation according to soil type, shape of
the constituent grains, and amount of compaction.
2. Permeability; although a soil might have a high water content, current
cannot flow unless connections exist between its interstitial pores.
3. Ion content; the ions responsible for conduction in the soil come from
dissolved salts, such as calcium and sodium carbonates. They may be derived
from a variety of cultural and noncultural sources: from the soil itself, underlying
geologic strata, rainwater, modern agricultural fertilizers, or compounds generated by cultural processes. Figure 8.20 illustrates the pronounced effect of the
addition of small amounts of dissolved salts on soil resistivity (Tagg, 1964).
4. Temperature affects resistivity, particularly when freezing of the groundwater
takes place. Fortunately, most field surveys can be performed when the temperature
is above 0C, where daily variations in temperature are not sufficient to affect
the resistivity in an archaeological context.

SOIL RESISTIVITY
The variables outlined above show wide spatial variation depending on climatic,
geologic, and edaphic conditions. Consequently, the resistivity of different
archaeological sites changes dramatically as well. Typical values of resistivity
different soils are given in the following table (Tagg, 1964).

Similarly, wide variation in resistivity can be encountered on a small scale on an

individual site.

Differences in soil moisture, dissolved salts, and like factors also are responsible for producing the culturally formed resistivity contrasts that are detected at
archaeological sites. Linear features, such as fortification ditches a few meters
long, are the most susceptible to detection when the surrounding medium and
climatic conditions allow a change in their moisture content. Stone alignments
and foundations may exhibit detectable resistivity contrasts because of their
marked difference in water retention. I louse floors or other compacted activity
areas are visible occasionally due either to decreased porosity or to moisture that
has accumulated on their surfaces. Midden areas, which are often high in soluble
ions and have a larger volume of interstitial-pore space, can show distinct
resistivity contrasts. The resistivity contrast of filled pits, although more subtle
than the other, more extensive features, are sometimes discernible.
Although the phenomena responsible for conduction of current in soils and
archaeological sites are understood fairly well, one cannot easily predict which
archaeological features will be detectable by resistivity surveying or whether the
soil noise will confuse or mask cultural resistivity contrasts. Since the state of the

224 / John W. Weymouth and Robert Huggins

physical remains depends on environment and cultural history, a feature which is
easily located by resistivity surveying in one area may be imperceptible in another.
INSTRUMENTATION AND FIELD PROCEDURES
Probes
The probes used for the transfer of current or measurement of voltage usually
consist of mild steel rods pointed on one end, with suitable hand holds on the
other to facilitate insertion in the soil. Ideally, the probes should act as point
sources, but since they must be inserted a finite distance into the ground this is
not the case. The actual value of resistivity will change if the insertion depth is
varied for one or all of the probes, so steps must be taken to ensure uniformity.
This can be accomplished by the use of an adjustable ring on the probe that will
arrest insertion at the desired depth. Field conditions usually dictate what is
reasonable, although to maintain adequate precision in determining the resistivity
the insertion depth should be less than 20 percent of the distance between the
nearest adjacent probes (Aitken, 1974; Van Nostrand and Cook, 1966). For a 1-in
spacing, the writers have found that insertion to 5 -10 cm provides adequate
contact in all but the driest soils.
Care must also be taken in situating the probes relative to one another. For
example, when using small spacings on the order of 30 cm in a colinear array such
as the Wenner or double dipole, errors in the measured resistivity can be as
large as 15 percent for the misplacement of a probe by 3 cm (Aitken, 1974).
Fortunately, archaeological applications seldom require spacings smaller than 50 cm,
as this seems to be the minimum size of detectable features. At 50 cm
between probes, one should attempt a positional accuracy of 1 cm.
Probe Configurations
For simplicity, only the four-in-line probe configuration, commonly known as the
Wenner array, was shown in the previous discussions. In practice, a variety of
arrays are effective in an archaeological context, the choice among them
depending on the terrain, the size of expected features, and the experience or
familiarity of the operator with one array or another. Figure 8.21 shows a plan
layout of the more common types. Descriptions of each array follow:
i. The Wenner array is the most often used and has several advantages. It
produces the largest percentage change over most lateral soil variations. Most
instruments are set up to accommodate this array. With the use of a fifth probe
and a rotary switch, the last probe can be deactivated and "leap-frogged" to the
front of the array while the instrument operator takes a reading using the other
four probes. This allows each additional reading to be taken with the insertion of
only one probe instead of moving the entire array. The main disadvantage of the
Wenner array is that it tends to produce subsidiary peaking in the data, typified by
large excursions in the readings before and after a resistivity contrast occurs.

226 / John W. Weymouth and Robert Huggins

This gives rise to M- and W-shaped anomalies, as illustrated in figure 8.21i.

ii. The double-dipole array is an attempt to reduce the subsidiary peaking

that occurs in the Wenner array while retaining similar sensitivity and mobility. As
shown in figure 8.21ii, only a single peak is encountered over a feature of
interest, making interpretation of profiles and gridded data less complex (Clark,
1975). A five-probe system can b e used in a fashion similar to the Wenner
system, thus speeding data collection. The apparent disadvantage of the doubledipole array is the increased fall-off with depth, which means that deeper
features contribute less to the readings.

Geophysical Surveying of Archaeological Sites / 227

polarized during current flow because build-up of charge near the probes alters
the measured resistance. Finally, measuring probes can detect small, naturally
occurring currents flowing directly through the earth, which introduce discrepancies. All of these error- producing effects can be minimized by using an
alternating current with a judiciously chosen frequency, typically between 10
and 500 liz. Most instruments available today are designed in this manner and
many are calibrated to read resistivity directly for a specific array.

iii. The twin array (fig. 8.21 iii) devised by Aspinall and Lynam (1970) used one
fixed current -voltage pair (II and VI ) that remains fixed throughout the survey
while the other pair (/ and V)) is used in the measuring procedure. The motivation
for this alteration is economy: the roving pair can 1w mechanically linked
together, resulting in the least amount of actual probe movement of any of the
configurations. The twin array also has the interpretational advantage of
producing single-peak anomalies. Its main disadvantage is reduced sensitivity,
producing the least percentage change from background of any of the other
arrays. However, in regions where resistivity contrasts are large the twin array is
probably the ideal choice. It should be noted that the value of n should always be
at least 30 or confusing effects will occur.

Field Procedures

iv. The square array designed by Clark (1968) was au attempt to reduce
difficulty in the field and in interpretation (fig. 8.21iv). Clark configured the
probes in such a way that they can be used as the legs of a table while the
instrument and other recording equipment rest on top. This seems like a
compact and suitable solution in many instances. Its main disadvantage is the
fixed distance between the probes,- usually at a maximum of 1 in, with the effect
that the array can be used only in the detection of features that are less than 1 in in
depth.

Establishing the electrical properties of the site before systematic surveying

aids in choosing the array and spacing. If the location of any suitable feature is
known, a trial traverse can be run using either the Wenner array or the doubledipole or both. The array spacing for this test and the later complete survey
should be contrived to allow current flow to encompass as much of the feature as
possible. The depth measured by the array is about equal to the distance
between the voltage probes. In other words, a volume roughly equal to a
hemisp here of diameter a is measured. It is permissible to have the probe
spacing at most equal to the width of the feature, providing its width and depth
are roughly the same. A higher degree of confidence in the existence of an
anomaly is obtained if three or more readings can be taken over the feature. If its
width is greater than its depth, the array spacing should be reduced accordingly.
For example, if some horizontal feature, such as a house floor, is encountered at a
depth of 30 cm with a width of 4 m and exhibiting a resistivity contrast of 5:1, the
following anomalies (expressed in arbitrary units depending on soils) would be

v. The Schlumberger array (fig. 8.21 v) is more commonly used in large-scale

geophysical applications but has been used successfully in an archaeological
context (Rees and Wright, 1969). Its sensitivity is comparable to that of the
Wenner array, but subsidiary peaking is again a problem. The distances between II
and VI or I 2 and V2 should always he much larger than a.
Choice of the correct array for any given field situation is not a straightforward
procedure. Despite the problems with peaking, the present writers generally
use the Wenner array, sometimes verifying a particular anomaly with the doubledipole or the more expeditious twin array, both of which yield single peaks.

Lateral variations in soil resistivity are measured by using linear traverses or a

regularly spaced grid, the choice depending primarily on required resolution
and economy. If the expected archaeological features are ditches or palisades, Or
features measuring several meters on a side and lacking interior detail, then
traverses are likely the best choice. However, if features on the order of a meter
or two in diameter are sought, then more definition can be obtained by measuring with a gridded regular interval. A locational system similar to that described for
magnetometer surveying is adequate, providing a nonconductive material, such
as dry cloth tapes or ropes, is used.

Instrumentation
In the direct- current cases presented up to this point, a simple ammeter and
voltme ter wou ld be adequate to measure the c urrent and voltage
needed to calculate the resistivity. In actuality, several problems arise
with this method. F ir s t, s ma ll , c he mic a ll y d e r iv ed v o lta ge s de v e lop
b e t w e en th e p r o be s a nd the ground, causing measurement errors. Second,
probes gradually become

The test traverse should be extended 10 or 15 in on either side of a prospective feature to ascertain ground-noise conditions. It is often advantageous to

228 / John W. Weymouth and Robert Huggins

traverse a nearby area devoid of cultural features to determine what natural

resistivity variation one might encounter. If the overall noise level appears high
compared with the anomaly produced by the feature, it is probably best to
continue using the double -dipole or Wenner array; if significant change is
detected, one might choose the Clark or the twin array to speed data collection.
The choice of the array is usually made on the basis of the operator's experience
and familiarity with an array and the affinity of the instrument used to one probe
configuration or the other. When little is known about the geometry of the
features on a site, a probe spacing of I in using the Wenner or the double-dipole
array is a judicious choice.
If traverses are used for field measurement, they should extend well away
from the region of archaeological interest so that additional background information can he determined. If a survey takes place over an extended time and
periodic rainfall occurs, steps should be taken to adjust the traverses relative to
one another to correct for the increased soil moisture. This can be accomplished
by running a common traverse before and after a rainfall and musing the average
difference as a constant for adjusting subsequent measurements.
Where the ground is level, easily penetrated, and clear of excessive vegeta tion, the writers have taken between live and seven readings per minute or,
considering the repositioning of measuring ropes involved, 150 to 190 readings
2
per hour. When a I -m grid spacing is used, an area of about 1,000-1,200 m can
be covered in a day, although such large surveys are arduous. Hard ground slows
the work, as the probes must be pounded to ensure proper contact.
Field notes should be maintained with observations on variations in soil type,
ease or difficulty in probe insertion, change in the density of vegetation, and
noticeable topographic features. These factors affect the resistivity and can often
be of valuable assistance during interpretation. A log of the rainfall for the period
of the fieldwork is also a valuable aid, since large amounts of rain can drastically
affect the magnitude and even the polarity of resistivity anomalies (Clark, 1975;
AI Chalabi and Bees, 1962).
Several simple field techniques can be used to distinguish more readily the
resistivity anomalies due to background variation in areas where noise from
surrounding soil is high. When the validity of a reading is suspect, the traverse
should be rerun for verification. W ith a simple alteration, instruments can
quickly read both the double- dipole and Wenner arrays at a single station
without repositioning the probes. In agricultural regions where cultivation has
truncated features, the confusing effects of disturbed topsoil caul be reduced and
the relative contribution from features increased by using the Barnes Layer
Method (Barnes, 1952, 1954; Carr, 1977, 1982). By running two or three traverses over the area of interest and using increased probe spacings for each
traverse, sets of data are generated that encompass greater soil depths. These
data sets can be treated to reduce contributions from specific layers to enhance
resistivity contrasts from cultural features or to reduce noise from topsoils.

Geophysical Surveying of Archaeological Sites / 229

INTERPRETATION
Once data are collected, they are processed and displayed in a manner that will
enhance the resistivity contrasts. Little alteration of data is needed before
preliminary display, since many instruments produce readings that are already
converted to resistivity and some compensate for the array being used. The only
other preliminary processing that might be necessary is the addition of an
empirical climatic factor, used to compensate for rainfall during a lengthy survey.
Because individual traverses are utilized more often in resistivity than magnetic surveying, profiling the data tends to be a more common display technique.
Profiles can be plotted readily by hand and on occasion should be done in the
field, in order to check noise levels or the feasibility of a particular array or
spacing. For large surveys, where an appreciable amount of data is collected, a
computer is a necessity.
Where data have been collected on a grid system, contour plotting yields
better resolution and more information. In these instances computer maps such
as SYMAP or line contours can be used.
Because the signal-to- noise ratio is difficult to quantify in resistivity surveying, in archaeological contexts the data are interpreted qualitatively. Essentially
two types of noise are involved: correlated noise, caused by the contributions
from natural soil variation, and uncorrelated noise, the sum of instrument
variation, difference in probe spacing and depth, and the occasional poor contact
of a probe with the ground. Both sources of noise are sufficient to mask contrasts
from archaeological features, so steps should be taken to recognize and to
minimize their occurrence.
Correlated noise from natural soil variations can be as large or larger than
the signal and in these instances can be distinguished only when the noise
signature has a different shape that that of the signal. Traverse data from an area
lacking buried cultural remains as the best preliminary clue to recognizing
correlated noise. If no specific correlated-noise anomalies are recognizable from
natural soil variation, then the standard deviation of all the readings on the test
traverse provides a reasonable background level above which to look for cultural
anomalies.
Uncorrelated noise levels can be minimized by field procedures. If a measurement on a traverse is greater than the preceding one by a predetermined
amount (usually the standard deviation of the test traverse), then the probe
contacts should be checked and the reading retaken. Recognition of both correlated and uncorrelated noise can be aided by the use of alternate probe configurations.
Figure 8.21 shows typical anomalies for features similar to pits or ditches.
Anomalies are similar in shape for most other archaeological features of that size, but
the responses will vary in magnitude. The feature in figure 8.21 has a higher
resistivity than the surrounding media such that 2/1 > 1. The anomalies

230 / John W. Weymouth and Robert Huggins

Geophysical Surveying of Archaeological Sites / 231

would be inverted if the resistivity of the feature were lower (i.e., p 2/p, < I).
Resistivity contrasts between larger features may be manifest as an average
change in the measurements, rather than as an individual, symmetrical anomaly.
If the resistivity contrast is large, subsidiary peaking may he present at the
boundary. Models of features have been used to aid in interpretation, but the
mathematical calculation for the expected anomalies quickly becomes formida ble, even for simple geometric shapes (Cook and Van Nostrand, 1954; Grant and
West, 1965; Telford et al., 1976).

visible difference in soil in the river bank where the site is partially truncated,
several radial traverses were run from the center of the village across the
surrounding depression. The first traverse, which is located north of the area of
figure 8.4, is shown in figure 8.23. When a marked area of low resistivity was
encountered using p robes spaced 1.0 in apart, the traverse was rerun with a
spacing of 1.5 m, which shows a reduced contrast. From this we can infer that the
source is of limited depth, probably no more than 1 in. Because the other
traverses revealed no anomalous responses in this region, it is likely that the
source of this anomaly is localized, indicating no detectable fortification ditch.

EXAMPLES

House 6

The following examples are all drawn from a survey at the Knife River Indian
Village National Historical Site in the vicinity of Sakakawea Village (32ME11 ).
The survey was undertaken in cooperation with the Midwest Archaeological
Center, National Park Service, to assess the potential responses of experimental
features and to examine what cultural areas were amenable to detection using
resistivity.

Using a sample spacing of 1 m, we collected two matrices of data over house 6

(block 0, fig. 8.4) by running traverses first eat to west and then north to south.
This method provides more information on the geometry of features and assures
more accuracy in each reading. In figure 8.24 the average of the two profiles is
displayed, a technique that reduces contributions from subsidiary peaking for
features of certain geometries and reflects some of the larger trends on the site.
In the center of the map is an oblong region of lower resistivity corresponding to
the center of the depression (see fig. 8.4). We can infer a region of higher
moisture and likely a reduced amount of compaction from this data. Surrounding
this central low region is a ring of higher resistivity that appears to lie well within
the depression. This area is apparently more compact, thus having a reduced
moisture content and a higher resistivity. A particularly notable portion of the
high ring is a crescent- shaped anomaly on the southeast edge. Wilson (1934)
mentions that livestock were kept in the interior of lodges and penned close to
the walls. It is conceivable that this high anomaly defines the boundaries of a
horse corral or other area of excessive traffic.
The midden region surrounding the depression has mostly a low resistivity
due to the reduced compaction and higher ion content. It is interesting to

Sterile Region
On the periphery of the village an imitation "cache pit was excavated and
refilled with moistened earth. Profiles across this anomaly using the Wenner and
the twin array are shown in figure 8.22. The Wenner array produces the classic Wshaped anomaly, whereas the twin array produces a narrower, single peak.
The magnitude of the twin anomaly is uncharacteristically large; it would
normally be quite a bit less than the magnitude of the Wenner anomaly.
Suspected Palisade
To test for the existence of a fortification ditch inferred by the presence of a

232 / John W. Weymouth and Robert Huggins

Geophysical Surveying of Archaeological Sites / 233

District U. S. Corps of Engineers, the Dolores Archaeological Program (Bureau

of Reclamation), the Oklahoma Archaeological Survey, and the Arkansas Archaeological Survey. In particular we wish to thank the Midwest Archaeological
Center, National Park Service, and F. A. Calabrese and R. Nickel of that agency
for long and continued support, encouragement, and collaboration.

REFERENCES
Aitken, M. J. 1974. Physics and archaeology. Oxford: Clarendon Press.
Aitken, M. J., G. Webster, and A. Rees. 1958. Magnetic prospecting. Antiquity 32:
270 - 71.
Aitken, M. J., and J. C. Alldred. 1964. A simulator-trainer for magnetic prospection. A
rchaeomeiry 7:28-35.
Al Chalabi, M. NI., and A. I. Rees. 1962. An experiment on the effect of rainfall on
electrical resistivity anomalies in the near surface. Bonner Jahrbucher 162:
266-71.
.
Arnold, J. B. 197 1. A magnetometer survey of the nineteenth century steamboat Black
Cloud. Bulletin of the Texas Archaeological Society 45:225-30.

compare figure 8.24 with figure 8.7, which shows the magnetic field over this
same house. The general outline of the lodge is present in both, but each method
shows distinctly different properties of the same area.
Acknowledgments
We wish to express our thanks to the following agencies, whose contracts have
permitted us to develop our techniques and gain valuable experience: the Tulsa

Aspinall, A., and J. T. Lynam. 1970. An induced polarization instrument for detection of
near surface features. Prospezioni Archeologiche 5:67-75.
Atkinson, R. J. C. 1952. Methodes electriques de prospection en archeologie. In La
decouverte de passe, ed. A. Laming, 59-70. Paris: Picard.
Barnes, 11. E. 1952. Mapping and subsurface exploration for engineering purposes.
Michigan State Highway Department Research Board Bulletin, 65.
______ . 1954. Electrical subsurface exploration simplified. Roads and Streets 97:81-84.
Belshe, J. C. 1957. Recent magnetic investigations at Cambridge University. Advances in
Physics 6:192 - 93.
Bevan, B. 1975. A magnetic survey at Les Forges du Saint-Maurice. MASCA Newsletter
11:1.
Black, G. A., and R. B. Johnston. 1962. A test of magnetometry as an aid to archaeology.
American Antiquity 28:199-205.
Bleed, P., M. Yoshizaki, W. Hurley, and J. W. Weymouth. 1980. A Preliminary Report
on the 1978 Excavation at the Yagi Site, Japan. Technical Report No. 80 -14.
Division of Archaeological Research. University of Nebraska.
Breiner, S. 1973. Applications manual for portable magnetometers. Sunnyvale, CA:
Geometries.
Breiner, S., and M. D. Coe. 1972. Magnetic exploration of the Olmec civilization.
American Scientist 60:566-75.
Carr, C. 1977. A new role and analytical design for the use of resistivity surveying in
archaeology. Mid-continental Journal of Archaeology 2:161 - 93.
. 1982. Handbook on soil resistivity surveying. Evanston, IL: Center for American
Archeology Press.
Clark, A. J. 1968. A square array for resistivity surveying. Prospezioni Archeologiche
3:111-14.
. 1969. Resistivity surveying. In Science in archaeology, 2d ed., ed. D. Brothwell and
E. S. Higgs, 695-708. London: Thames and Hudson.
. 1975. Archaeological prospecting: A progress report. Journal of Archaeological
Science 2:297 - 314.

234 / John W. Wey mouth and Robert Huggins

Cook, K. L., and R. G. Van Nostrand. 1954. Interpretation of resistivity data over filled
sinks. Geophysics 19:761-70.
Davis, J. C. 1973. Statistics and data analysis in geology. New York: John Wiley and
Sons.
De Terra, H. 1947. A preliminary note on the discovery of fossil man at Tepexpan in the Valley
of Mexico. American Antiquity 13:40-44.
Ezell, P., J. R. Moriarity, J. D. Mudie, and A. I. Rees. 1965. Magnetic prospecting in
southern California. American Antiquity 31:112-13.
Ginzburg, A., and A. Levanon. 1977. Direct current resistivity measurements in archae ology. Geoexploration 15:47-56.
Graham, 1. D. C., and I. Scollar. 1976. Limitation on magnetic prospection in archaeology
imposed by soil properties. A rchaeo-Physika 6:1-125.
Grant, F. S., and G. F. West. 1965. Interpretation theory in applied geophysics. New
York: McGraw-Hill.
Hesse, A. 1962. Geophysical prospecting for archaeology in France. Archaeometry
5:123- 25.
Le Borgne, E. 195.5. Susceptibilite magnetique anormale du sol superficiel. Annales de
geophysique 11:399-419.
. 1960. Influence du feu sur les proprietes magnetiques du sol. Annales de
geophysique 16:159-95.
Leith, C. J., K. A. Schneider, and C. Carr. 1976. Geophysical investigation ofarchaeological
sites. Bulletin of the International Association of Engineering Geology 14: 123-28.
Lerici, C. M. 1961. Archaeological survey with the proton magnetometer in Italy.
Archaeometry 4:76-82.
Linington, R. E. 1964. The use of simplified anomalies in magnetic surveying. Archaeometry 7:3-13.
. 1972. A summary of simple theory applicable to magnetic prospecting in archaeology.
Prospezioni A rcheologiche 7/8:9-60.
Mason, R. 1981. Large-scale archaeomagnetic surveys of the Barton and Vinton Townsites. Paper delivered at the Fourteenth Annual Meeting, Society for Historical
Archaeology (New Orleans).
McDonald, W. A., and G. Rapp, Jr., eds. 1972. The Minnesota Mes.senia expedition.
Minneapolis: Univ. of Minnesota Press.
Morrison, F., C. W. Clewlow, and R. F. Heizer. 1970, Magnetometer survey of the La
Venta Pyramid. Contributions of the University of California Archaeological
Research Facility 8:1- 20.
Nashold, B. W. 1977. An archaeological magnetic survey at Cahokia. M.A. thesis,
University of Illinois at Chicago Circle.
Rainey, F., and E. K. Ralph. 1966. Archaeology and its new technology. Science 153:
1481 -91.
Ralph, E. K. 1964. Comparison ()fit proton and a rubidium magnetometer for archaeological
prospecting. Archaeometry 7:20-27.
Ralph. E. K., F. Morrison, and D. P. O'Brien. 1968. Archaeological surveying utilizing a
high-sensitivity difference magnetometer. Geoexplomtion 6:109-22.
Rees. A. 1., and A. E. Wright. 1969. Resistivity surveys at Barnsley Park. Prospezioni
A rcheologiche 4:121-24.
Resnick, R., and D. Halliday. 1966. Physics. New York: John Wiley and Sons.

Geophysical Surveying of Archaeological Sites / 235

Schlumberger, C. 1920. Analysis of electrical prospecting of the subsoil. Academie de

sciences (Paris), Comptes rendus 170:519-21.
Scollar, I. 1961. Magnetic prospecting in the Rhineland. Archaeometry 4:74-75.
___ . 1965. A contribution to magnetic prospecting in archaeology. Beihefte der
Bonner Jahrbucher 15:21-92.
SYMAP. 1975. Synagraphic computer mapping. Cambridge, MA: Harvard University, Laboratory
for Computer Graphics and Spatial Analysis.
Tagg, G. F. 1964. Earth resistances. London: Pitman.
Telford, W. M., L. P. Geldart, R. E. Sheriff, and D. A. Keys. 1976. Applied geophysics.
Cambridge, England: Cambridge Univ. Press.
Tite, M. S. 1972. Methods of physical examination in archaeology. New York: Academic Press.
Van Nostrand, R. C., and K. L. Cook. 1966. Interpretation of resistivity data. U. S.
Geological Survey, Professional Paper no. 499.
von Frese, R. R. B. 1978. Magnetic exploration of historical archaeological sites as
exemplified by a survey at Ft. Ottiatt non (12T9). M.S. thesis, Purdue University.
Wenner, F. 1915. A method of measuring earth resistivity. Bulletin, U.S . Bureau of
Standards 12, Scientific Paper 258, pp. 469-78.
Weymouth, J. W. 1976. A magnetic survey of the Walth Bay site, Midwest Archaeological
Center (NPS). Occasional Studies in Anthropology, no. 3.
______ . 1979. Magnetic surveying of archaeological sites. In Proceedings of the First
Conference on Scientific Research in the National Parks, vol. 2, 941-47.
Weymouth, J. W., and R. Nickel. 1977. A magnetometer survey of the Knife River Indian
villages. Plains Anthropologist 22:104 - 18.
Wilson, G. L. 1934. The Hidatsa earth lodge. Anthropological Papers of the American
Museum of Natural History 23:340-420.