close

Вход

Забыли?

вход по аккаунту

?

Geophysical Investigation at the Falling Creek Ironworks an Early Industrial Site in Virginia.

код для вставкиСкачать
Archaeological Prospection
Archaeol. Prospect. 8, 247–256 (2001)
DOI: 10.1002/arp.173
Geophysical Investigation at the
Falling Creek Ironworks, an Early
Industrial Site in Virginia
GEOFFREY JONES*
Archaeo-Physics, LLC, 1313 5th Street SE, Minneapolis, MN 55414, USA
ABSTRACT
A geophysical investigation was conducted at the site of the Falling Creek Ironworks (1619–1622),
the first iron production facility in North America. Electrical resistance and magnetic field gradient
surveys were conducted over the site of the seventeenth century ironworks. Additionally, groundpenetrating radar was used to investigate areas under an existing roadway. Linear and rectangular
anomalies that appear in the resistance data may be caused by architectural features, possibly
shops or domestic structures associated with the ironworks. Several low-amplitude magnetic
anomalies that appear in the data appear to be cultural in origin, and may be caused by features
associated with the ironworks. A number of high-amplitude magnetic anomalies appear in the
magnetic field gradient data that are thought to be associated with iron production, including
extensive slag deposits and the possible location of the blast furnace. Ground-penetrating radar
data shows an anomalous reflection in the vicinity of the suspected blast furnace, which is partially
beneath the modern roadway. Copyright  2001 John Wiley & Sons, Ltd.
Key words: Falling Creek Ironworks; Virginia; industrial archaeology; magnetic field gradient;
resistivity; ground penetrating radar
Introduction
During the years 1619 to 1622, the Virginia
Company attempted to establish an ironworking
facility on Falling Creek, in Chesterfield County,
Virginia, now on the outskirts of the city of
Richmond. The Falling Creek Ironworks was the
first iron production facility in North America.
The ironworks were to produce iron from local
ore deposits. According to Virginia Company
records, the ironworks was able to produce some
quantity of iron, although it is not clear whether it
had begun full production. In 1622, war with the
Powhatan Confederacy of tribes put the operation
of the Falling Creek Ironworks to an abrupt end.
An attack by Native American forces left all but
*Correspondence to: Geoffrey Jones, Archaeo-Physics, LLC,
1313 5th Street SE, Minneapolis, MN 55414, USA.
E-mail: jones@archaeophysics.com
Copyright  2001 John Wiley & Sons, Ltd.
two colonists at the Ironworks dead, and the
facilities destroyed (Higgins et al., 1995).
Previous archaeological investigations had
identified a terrace located at the foot of the
lower falls of Falling Creek as the site of the
1619–1622 ironworks (McCord, 1964; Higgins
et al., 1995). This site was given the site designation 44CF7. These investigations concluded
that the site contains significant archaeological
resources. Limited testing during the course of
these investigations found, ‘no conclusive evidence of structures or domestic areas’, however,
‘thick slag and charcoal deposits associated with
the ironworks were identified adjacent to the current access road near the southern boundary of
the sites’ (Higgins et al., 1995). The location of the
blast furnace is thought to be under the existing
access road, adjacent to this slag and charcoal
deposit.
Current archaeological investigations at the
Falling Creek Ironworks are under the direction
Received 2 March 2001
Accepted 1 June 2001
G. Jones
248
of Lyle E. Browning, Browning and Associates,
Ltd working through the Chesterfield County
Department of Parks and Recreation. The objective of this geophysical investigation was to
evaluate the entire site using non-destructive
geophysical techniques prior to subsurface testing. This evaluation hoped to locate and identify
evidence of possible iron works structures such
as a blast furnace, slag and charcoal deposits,
bloomeries, forges, as well as any related structures, such as living quarters or other facilities.
The investigation consisted of electrical resistance
and magnetic field gradient surveys over the suspected location of the ironworks. These methods
have proven to an effective means of evaluation
for iron working sites (Vernon, 1998), as well
other historic and prehistoric archaeological sites
(Clark, 1990; Scollar, 1990). Additionally, groundpenetrating radar (GPR) was used to survey areas
under the existing roadway.
Magnetic field gradient survey was used to
locate and map induced magnetization anomalies expected from high magnetic susceptibility,
iron-enriched soil associated with iron working facilities, as well as any large thermoremanent magnetic fields that could be expected
from in situ furnace or hearth features. Electrical
resistance survey was used to locate and map
structural remains (foundations, wall trenches,
pits, etc.) associated with the iron working facilities. Ground-penetrating radar was used to investigate areas that lie beneath a modern access road
for evidence of buried archaeological features.
Fieldwork was conducted from 25 September
through to 27 September 1999.
Site setting and field conditions
The survey area is located on a terrace of Falling
Creek, at the foot of the lower falls and about
3000 feet upstream of its confluence with the
James River (Figure 1). This location would have
provided both waterpower and boat access for
the seventeenth century ironworks. The terrace
is fairly level, although somewhat dissected by
erosion. It is located on the south (right) bank
of Falling Creek, and bounded on the south
by a steep bank several metres high. A modern
asphalt and gravel road runs along the base of
Copyright  2001 John Wiley & Sons, Ltd.
this slope. Soils are composed of Fluvaquents,
a mixed alluvium of loose, poorly consolidated
sands and gravels. Vegetation consisted mainly
of deciduous trees. The survey area had been
covered with a heavy growth of brush and vines,
but had been cleared prior to the commencement
of fieldwork. The terrace had been flooded briefly
the week before the survey by high tides owing
to a hurricane, but the soils were well drained.
A later forge was operated by Archibald Carey
from 1749 to 1782 (when it was burned by the
British troops). This forge traditionally is thought
to have been located on the opposite (north) bank
of Falling Creek, and may have occupied a ruined
structure that is still visible opposite the project
area. It is not known what impact the Carey forge
may have had on the south bank of the creek,
but it should be considered as a possible origin
for both ironworking and architectural features
(L. E. Browning, personal communication, 2001;
Higgins et al., 1995).
Survey design and methods
The project area was divided into 20 ð 20 m
survey grids (Figure 1). Several 5 ð 5 m and 10
ð 10 m grids were used to cover irregular spaces
at the edges of the project area. An effort was
made to clear the project area of modern metal
debris, although many metal objects, both large
and small, were not seen or were buried in the
loose soils.
Resistance survey
The resistance survey was performed with a
Geoscan RM-15 resistance metre and a PA-5
probe array. The instrument was operated in the
twin-electrode mode. Data were collected along
north–south traverses across each survey grid,
with a 1 m traverse interval and 0.5 m sample
interval (two samples per square metre).
A 0.75 m electrode spacing was selected. At
this mobile probe separation the data measured
is representative of the volume average resistance
to a nominal depth of 0.75 m. A current of 1.0 mA
at a frequency of 137 Hz and a potential of 40 V
was used to log data at a 0.1 ohm sensitivity level
in the medium integration mode.
Archaeol. Prospect. 8, 247–256 (2001)
Geophysical Investigation at Falling Creek
249
(a)
Delaware
Maryland
Chesape ake Bay
Richmond
s
me
Ja
Virginia
r
Falling Creek X
R iv
e
Atlantic
Ocean
North Carolina
(b) 70
F a ll
0
0
60
in g
C re
ek
North (metres)
50
44CF7
40
30
3
20
GPR com
posite (Fig
10
.4)
6
CULVERT
ROA
D
9
0
0
10
20
30
40
50
60
70
80
East (metres)
Elevations in metres above high tide level
90
100
110
120
Figure 1. (a) Location of Falling Creek. (b) Map of project area.
Resistance data were processed with Geoplot
software. Isolated outliers (greater than š3 standard deviations) were removed and replaced
with the local mean. The data were expanded in
the y direction (transect spacing) using [(sin x/x]
interpolation, resulting a uniform number of data
points per metre in both the x and y directions. In
order to enhance small, subtle features and suppress large-scale anomalies of geological origin,
Copyright  2001 John Wiley & Sons, Ltd.
the resistance data were subjected to a Gaussian weighted high-pass filter with a 5 m radius.
The processed data were exported for display to
Surfer mapping software.
Magnetic field gradient survey
The magnetic survey was performed with a
Geoscan FM-36 fluxgate gradiometer. Data were
Archaeol. Prospect. 8, 247–256 (2001)
G. Jones
250
collected along north–south traverses across each
survey grid, with 1 m traverse interval and
0.125 m sample interval (eight samples per square
metre). The instrument was operated in the 1
nT sensitivity range. At this setting the dynamic
range of the gradiometer is š2047 nT.
Gradiometer data were processed with Geoplot
software. The data were expanded in the y
direction (transect spacing) and reduced in the
x direction using [(sin x/x] interpolation to
two data points per metre in both the x and
y directions. The magnetic data were found to
contain both very high amplitude (exceeding the
dynamic range of the instrument) and very subtle
anomalies that were believed to be of cultural
origin. Therefore, two sets of magnetic data were
created, one clipped at š3 standard deviations,
and another clipped at š0.5 standard deviations.
These dual data sets are intended to display the
dominant high-amplitude magnetic anomalies,
and to display the more subtle magnetic features
respectively. The processed data were exported
for display to Surfer mapping software.
GPR survey
The GPR investigation was performed using a
pulseEKKO 1000 ground-penetrating radar in
fixed offset reflection mode. A frequency of
225 MHz was used at an antenna separation of
0.5 m. Data were collected systematically in a
series of transects 20 m long along the entire
northern edge of the existing access road and
portions of the southern edge of the roadway.
Data were collected within a time window of 100
ns. Each transect used a step-size of 0.05 m and
a stack setting of 32 with a sampling interval of
400 ps. An electromagnetic wave velocity of 0.10
m ns1 was estimated.
The GPR data were processed and analysed
by applying appropriate time gains, temporal
and spatial filters, image plotting and statistical
analysis. An exponential gain was applied to
compensate for signal attenuation at increasing
travel times. Bandwidth was reduced through
the application of a temporal bandpass frequency
filter. The signal-to-noise ratio was improved by
applying a temporal median filter down each
trace and a spatial median filter from trace to
trace. Strong horizontal banding was reduced
Copyright  2001 John Wiley & Sons, Ltd.
by applying a spatial high-pass background
subtraction filter with a window size of 3 m.
The profile image was created using processed
data and smoothed using temporal and spatial
low-pass filters.
Visual analysis of the profile image was augmented by calculation of the statistical variance
of data 10 to 60 ns from time zero. This time
range represents approximately 0.5 to 3 m below
surface using an estimated propagation velocity
of 0.1 m ns1 . Examination of the statistical variation within a GPR trace, referred to as activity
analysis by Barker et al. (1998), is an effective analytical method used to numerically detect areas
of ‘interest’ or ‘activity’.
Results and interpretations
Survey results
Electrical resistance survey was performed over
a total area of approximately 3900 m2 . Measured
resistance values in the raw data ranged from
32.8 ohms to 204.7 ohms (the limit of the instruments dynamic range), with a standard deviation
of 37 ohms. The high degree of variance in the
raw data was a result of the variation in grain size
and moisture content of the soils. In the processed
data, values ranged from 36 to 36 ohms with a
standard deviation of 12 ohms. (Although data
values are given in units of apparent resistance,
rather than absolute resistivity, they may be taken
as an approximation of the relative resistivity of
the soil matrix across the site.) The much smaller
range and lower standard deviation of the processed data result in greatly improved display
characteristics and improves the signal-to-noise
ratio, making subtle archaeological features more
easily detectable. The results of resistance survey
are presented graphically as a greyscale image
map (Figure 2a). Resistance values close to the
mean are displayed as shades of grey, the lighter
shades representing areas of relatively low resistivity and the darker shades representing areas of
relatively high resistivity. Extreme high and low
values are displayed as white and black, respectively. In Figure 2b, interpretations are overlaid
on the resistance survey results. Interpretations
of anomalies that may be associated with the
Archaeol. Prospect. 8, 247–256 (2001)
Geophysical Investigation at Falling Creek
251
(a) 70
35 Ω
60
25 Ω
North (metres)
50
15 Ω
40
5Ω
−5 Ω
30
−5 Ω
20
−25 Ω
10
−3 Ω
0
0
10
20
30
40
50
60
East (metres)
70
80
90
100
110
120
70
80
90
100
110
120
(b) 70
60
North (metres)
50
40
30
20
ROAD
10
0
0
10
20
30
40
50
60
East (metres)
Figure 2. (a) Electrical resistance survey results. (b) Electrical resistance survey results with interpretations.
ironworks are discussed in descending order of
interest below.
(i) A series of linear high resistance anomalies,
some of which form rectangular features, are
outlined with solid black lines. Remains of
architectural features may be the source of
these anomalies, possibly buildings related
to the ironworks such as domestic structures
Copyright  2001 John Wiley & Sons, Ltd.
or workshops, or may be warehouses from
the eighteenth century Archibald Cary forge.
(ii) Pairs of distinct parallel linear resistance
highs appear in two regions. These are
centred at N40/E60 and at N26/E77, and
are outlined with a dashed black line in
Figure 2a. These anomalies have close spatial
correlation with linear magnetic anomalies
discussed in (iv) of the following magnetic
Archaeol. Prospect. 8, 247–256 (2001)
252
survey results section. The source of these
anomalies may be linear archaeological
features associated with the ironworks.
(iii) There also are numerous, less distinct, linear
resistance anomalies present in the resistance
data that were not outlined, particularly in
the area between E10 and E40. Any of these
linear anomalies may be associated with
archaeological features. A dam, which can
be seen on the banks and in the bed of the
creek, may extend south across the terrace.
The south end of the visible portion is at
N65/E25. Although it does not appear distinctly in the resistance data, this might be
the origin of some of these resistance anomalies. They also might be of geological origin
or the result of modern earth moving or other
activities. The extremely high resistance values seen throughout the northern edge of
the survey area appear to be the result of
changes in moisture content associated with
variations in soil type and degree of soil
consolidation.
Magnetic survey results and interpretations
A total area of approximately 4700 m2 was covered by magnetic field gradient survey. Magnetic
field gradient values ranged from š2000 nT,
with a standard deviation of š78 nT. The magnetic survey results are presented graphically as
greyscale image maps. Figure 3a displays magnetic data clipped at š3 standard deviations.
Figure 3b displays data clipped at š0.5 standard
deviations. Figure 3a is intended to display and
emphasize extremely high-amplitude anomalies,
whereas Figure 3b displays more subtle magnetic
anomalies. Anomalies of interest are outlined in
Figure 3c
The creek or its steep banks effectively limited
data collection on the north. A broken black line
indicates the northern limit of data collection.
Data beyond this line were recorded with the
instrument stationary at the limit of the survey
and are not valid. The high-amplitude anomaly
centred at N13/E43 is caused by a modern
drainage culvert that runs underneath the access
road. Interpretations of anomalies that may be
associated with the ironworks are discussed in
descending order of interest below.
Copyright  2001 John Wiley & Sons, Ltd.
G. Jones
(i) A high-amplitude, large wavelength bipolar anomaly centred at N18/E28 is outlined with a solid white circle (it is within
the larger circle outlining the anomaly
discussed in (ii) below). This anomaly
has strong positive and negative components (C280, 220 nT) aligned with magnetic north. The anomaly has a relatively
large wavelength in profile, indicative of
a greater depth below surface for the
anomaly source. The characteristics of this
anomaly make it a strong candidate for a
large, iron-rich, in situ buried thermoremanent archaeological feature such as a blast
furnace. Note: this anomaly is most readily
visible in Figure 3a. This anomaly has spatial correlation with anomalies in the GPR
data (below).
(ii) A large high-amplitude anomaly, centred
at N25/E35 is outlined with a dashed white
circle. This anomalous region is fan shaped,
approximately 30 m wide in the east–west
direction and 15 m in the north–south
direction. Magnetic field gradient values
range from approximately C225 to 100
nT. The source of this anomaly is probably
the slag and charcoal deposit identified
during previous investigations at 44CF7
(MacCord, 1964; Higgins et al., 1995). If
so, the source of the deposit probably is
up the slope directly to the south (see (i)
above). A pile of earth at N17/E45 that had
been excavated for a modern culvert was
observed at the time of survey to contain
blast furnace slag.
(iii) Two low-amplitude, large wavelength
bipolar anomalies (centred at N34/E54
and N46/E84) are circled with a solid
black line. They have positive and
negative components that are aligned with
magnetic north, a relatively strong positive
component (C30 to 50 nT) to the south and
a relatively weak negative component to
the north (8 to 10 nT). Both anomalies
have a large wavelength in profile, making
them appear ‘diffuse’. Larger wavelengths
are often indicative of a greater depth
below surface for the anomaly source. The
characteristics of these anomalies make
them strong candidates for smaller buried
Archaeol. Prospect. 8, 247–256 (2001)
Geophysical Investigation at Falling Creek
253
(a) 70
60
275 nT
225 nT
North (metres)
50
175 nT
125 nT
75 nT
40
25 nT
−25 nT
30
−75 nT
−125 nT
20
−175 nT
−225 nT
10
−275 nT
0
0
10
20
30
40
50
60
70
East (metres)
80
90
100
110
120
(b) 70
40 nT
60
30 nT
North (metres)
50
20 nT
10 nT
40
0 nT
30
−10 nT
−20 nT
20
−30 nT
10
−40 nT
0
0
10
20
30
40
50
60
70
East (metres)
80
90
100
110
120
Figure 3. (a) Magnetic field gradient survey results clipped at three standard deviations. (b) Magnetic field gradient survey
results clipped at 0.5 standard deviations. (c) Magnetic field gradient survey results with interpretations.
in situ thermoremanent archaeological
features (bloomeries, chaferies, kilns, etc).
(iv) Two linear magnetic field gradient highs
centred at N29/E72 and N45/E72 are circled with a dotted black line. The southern
anomaly is a set of parallel linear features
separated by 1–2 m. These anomalies have
maximum amplitudes of approximately
15–20 nT. The source of these anomalous
Copyright  2001 John Wiley & Sons, Ltd.
regions may be linear archaeological features (walls, trenches, etc). These magnetic
anomalies have close spatial correlation
with linear resistance anomalies discussed
in (ii) of the previous resistance results
section.
(v) Numerous high-amplitude, short wavelength bipolar anomalies are circled with
a dotted white lines. These anomalies are
Archaeol. Prospect. 8, 247–256 (2001)
G. Jones
254
(c) 70
60
North (metres)
50
40
30
20
10
ROA
D
0
0
10
20
30
40
50
60
70
East (metres)
80
90
100
110
120
Figure 3c. (Continued).
aligned to magnetic north and have strong
positive and negative components (š50 to
100 nT). The high-amplitude and short
wavelength of these anomalies indicate
they are probably caused by ferrous iron
objects at or close to the ground surface, however, their north–south orientation suggests that they might be caused
by in situ thermoremanent archaeological
features.
(vi) Several subtle magnetic highs and possible
diffuse bipolar anomalies appear in the
generally noisy region east of E50. These
have not been circled because they are
poorly defined, but may be of archaeological interest.
(vii) Many other high-amplitude bipolar anomalies are present in the magnetic data that
appear to be randomly oriented, and highamplitude anomalies, which seem to lack
bipolarity. Many of these are the result
of modern iron or steel objects, many of
which were observed on the surface. It is
possible, however, that some of these relate
to archaeological features that have been
disturbed, or iron objects associated with
the 1619–1622 ironworks or the 1749–1782
forge.
Copyright  2001 John Wiley & Sons, Ltd.
(viii) The magnetic survey was extended into
the creek over the remains of a dam. The
dam can be seen in the magnetic data at
about E25, between N55 and N70. A large
number of modern iron and steel objects
washed from upstream among the rocks
of the ruined dam are the apparent cause
of the many high-amplitude anomalies in
this area. Data were not collected in the
deeper water around the dam. A broken
black line in Figure 3c indicates the limit of
data collection.
GPR survey results
A total of seven separate lines of GPR data, each
20 m long, were collected over the existing access
roadway. Five lines were collected along the
northern edge of the roadway, starting at E10
and finishing at E110. Two lines were collected
along the southern edge of the roadway, starting
at E10 and finishing at E50.
A composite image of GPR data collected
along the northern edge of the roadway between
E10 and E60 is presented in Figure 4 as a twodimensional profile plot (see Figure 1 for location). Above this image is a graph representing
the average down-the-trace variance. A single
Archaeol. Prospect. 8, 247–256 (2001)
Geophysical Investigation at Falling Creek
255
275000000
225000000
200000000
175000000
150000000
125000000
75000000
50000000
25000000
CULVERT
100000000
BEDROCK NEAR
SURFACE
Average Variance (µV2)
250000000
Estimated Depth (metres)
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
10
15
20
25
30
35
40
Position (metres)
45
50
55
60
The gray line in the variance plot represents the average variance within each trace between 0.5 and 3
metres estimated depth (trace interval interpolated to 0.2 metres)
The black line represents the average variance within a moving window of 2 metres.
Figure 4. Ground-penetrating radar profile and variance analysis.
area of interest was identified, as well as anomalies that are not believed to be of interest.
(i) The region between E25 and E40 is characterized by high variance, with strong reflections
and peaks in variance at E31 and E38. A
large bipolar magnetic anomaly is centred at
N18/E28, which falls within this anomalous
area. This magnetic anomaly is described
in (i) of the magnetic survey results. The
GPR data collected from the southern edge
of the roadway (not shown) shows a similar
response in this region.
(ii) High variance at the beginning of this plot
(E10) may be caused by a bedrock outcrop near the surface beneath the roadway.
Copyright  2001 John Wiley & Sons, Ltd.
(iii) The large inverted hyperbolic event located
at E44 is caused by the drainage culvert that
runs under the road from north to south.
Conclusions
The geophysical investigation was successful
in identifying evidence of intact archaeological
features possibly related to the Falling Creek
Ironworks.
Anomalies in the magnetic field gradient and
GPR data are consistent with the anticipated
signatures of a blast furnace located at N18/E27,
with ash and slag deposited downslope to the
north. The extent of this suspected ash and slag
deposit may indicate that a substantial amount of
Archaeol. Prospect. 8, 247–256 (2001)
G. Jones
256
ore had been processed before the destruction of
the ironworks. Other magnetic anomalies may be
caused by features associated with the ironworks.
The extent to which materials from the 1749–1782
Cary Forge operation contributed to these data is
unknown.
Linear and rectangular anomalies in the resistance data may be associated with the remains
of domestic, storage or shop structures. In two
areas, linear features in both the resistance and
magnetic data have a close spatial correlation.
The source of these features is very likely intact
linear archaeological features. Magnetic anomalies thought to be associated with slag deposits
and ironworking activities generally have little
correlation in the resistance and magnetic data.
A possible reason for this is that there may be
insufficient contrast between these features and
the generally coarse, poorly consolidated soils at
the site.
At the time of this writing, subsurface testing
of geophysical survey results had not been performed. The interpretation of geophysical survey
data is an iterative process based on communication between archaeologists and archaeological
geophysicists, and on the integration of geophysical and subsurface testing results. The interpretations presented here must be considered
Copyright  2001 John Wiley & Sons, Ltd.
preliminary and somewhat speculative. As subsurface testing results become available, more
elaborate interpretations may made with a higher
degree of confidence.
References
Barker P, Fletcher M, Bradley J. 1998. Reflections on
the past: progress in the application of GPR in
archaeology. Proceedings of the Seventh International
Conference on Ground-penetrating Radar, 27–30 May,
Lawrence, Kansas.
Clark AJ. 1990. Seeing Beneath the Soil: Prospecting
Methods in Archaeology. B.T. Batsford: London.
Higgins TF III, Downing C, Linebaugh D, Opperman A, Turner R III. 1995. Archaeological Investigations of Site 44CF7, Falling Creek Ironworks, and Vicinity
Chesterfield County, Virginia. Report Series No. 4, Virginia Department of Historic Resources Survey and
Planning, Richmond.
MacCord HA. 1964. Exploratory excavations at the
first ironworks in America (44CF7). Quarterly Bulletin of the Archeological Society of Virginia 19(1):
2–13.
Scollar I. 1999. Archaeological Prospecting and Remote
Sensing. Cambridge University Press: Cambridge.
Vernon RW, McDonnell G, Schmidt A. 1998. An integrated geophysical and analytical appraisal of early
iron-working: three case studies. Journal of the Historical Metallurgy Society 32(2).
Archaeol. Prospect. 8, 247–256 (2001)
Документ
Категория
Без категории
Просмотров
2
Размер файла
320 Кб
Теги
investigation, site, virginia, industries, fallin, creed, early, geophysics, ironworks
1/--страниц
Пожаловаться на содержимое документа