Charles T. Young and Kimberly Finch

GPR and Archaeological Excavations at the West Point Foundry, New York

The West Point Foundry produced ordnance, sugar machinery, and the nation's first locomotive. Now only foundation walls are visible at the site. Conventional archaeology provided information on site history and evolution, but the buried water channels that supplied a giant waterwheel remained unlocated. Ground penetrating radar (GPR) was used to search for the channels because it was relatively rapid to deploy, and the data show depth directly. Surveys were carried out at nine areas, and selected anomaly sites were excavated, revealing water channels and other structures.

Introduction

For nearly 100 years, the West Point Foundry in Cold Spring, New York, supplied the United States and the world with iron products. Best known for the production of the Parrott gun and America's first locomotive, the Best Friend, the foundry supplied many other important products such as equipment for sugar mills and components of city infrastructure. Archaeological fieldwork commenced in 2002 and continued in 2003 assisted by the GPR reported here.¹ The main research goals associated with the GPR surveys were (1) to assess the applicability of GPR surveys on industrial sites; and (2) to better understand the location, construction details, and function of components of the foundry's waterpower system.

The West Point Foundry is located at 41.42N and 73.955W, next to the village of Cold Spring, New York, as indicated in Figure 1. The 89-acre site occupies a steep valley bounded by the village of Cold Spring three-quarters of a mile to the northwest, the Hudson River, and Foundry Cove marsh to the south and west. The site is bisected by Foundry Brook.

Figure 1. USGS Topographic map indicating study area. Cold Spring in Putnam County, 55 miles north of New York City and approximately 100 miles south of Albany.

The foundry relied on an intricate network of elevated surface and subsurface drains, races, flumes, waterwheels, turbines, dams, and ponds that powered operations and regulated water flow throughout the site. The locations of many of these features, as well as details of their construction and functions, are not well understood. Historic maps do not provide a clear progression of construction of the components of the water system, and historic documents, maps, and illustrations provide incomplete and contradictory information about these systems. Geophysical and archaeological evidence gathered by students from the authors' university provided insight into features identified on historic sources and discovered previously unknown buried features.

Methodology

Ground penetrating radar (GPR) is one of the three main geophysical tools commonly used to map buried cultural resources. Other methods are magnetic and resistivity mapping. The great value of GPR is that in favorable conditions it can provide a realistic, real-time cross-sectional image of the subsurface. The GPR equipment sends a brief electromagnetic pulse into the soil and records the return waveform. In archaeological work, radar antennas are commonly chosen to provide roughly 1 to 2 meters of penetration. The returned waveforms are displayed in real time on an operator's console and are stored digitally for later processing and presentation. Waveforms are plotted vertically, side by side, with time increasing down the screen, creating a two-dimensional cross section of distance along the traverse in the horizontal direction vs. travel time in the vertical direction. The time dimension down the page is approximately equivalent to depth. The resulting cross section resembles a cutaway view of the earth beneath the profile, but there are distortions because radar wave velocity is not constant in the earth and because of spreading and scattering of the radar wave. Small objects scatter the radar wave and appear as downward opening arcs. Experience and understanding of wave propagation are required to identify buried cultural resources in the cross sections. Application of GPR to archaeology was discussed thoroughly by Lawrence Conyers in 2004.² Specific application of GPR to industrial archaeology in particular is not discussed in the literature to any length, but it overlaps other archaeological and environmental clean-up applications. John Reynolds discussed principles of GPR for environmental and geological studies in 1995.³

All geophysical measurements respond to changes in subsurface physical properties. The most common geological background for GPR data is partially continuous subsurface layering. Buried cultural resources are usually visible as a disruption of this layering, and thus their appearance is anomalous. The region where the natural layering is replaced by some other feature on a radar cross section is termed an anomaly. There has been considerable reworking of subsurface layers for the radar data discussed here because of long human occupation of the site, resulting in many point scatterers (downward opening arcs) from bits of iron and many breaks in layering, some of which may be due to buried brickworks or filled-in trenches. The data presented here were collected every 10 cm along the survey line and are displayed as grayscale cross sections. Common radar data processing has enhanced the visibility of subsurface structures, including those selected for excavation. The downward opening arcs are suppressed by filtering the data in the horizontal direction with a running average. Other processing consisted of compensating for the decay of the signal at greater times with AGC (digital automatic gain control filtering) and of suppressing an early time low-frequency offset known as "wow," which is due to transient charge buildup in the soil near the antennas.

A NogginPlus radar unit and SmartCart, manufactured by Sensors and Software of Mississauga, Ontario, Canada, was used for the study described here. The cart contained a wheel odometer, battery, digital data logger, and video display. The latter permitted direct observation of the radar cross section, which built up as the cart passed over buried features. A 250 MHz antenna was used, which was the lowest frequency available for the NogginPlus and provided the greatest penetration.

In determining excavation unit placement, archaeologists used radar results discussed here and site maps from 1840 through 1979 in conjunction with visual inspection of the surface. In particular, a 1979 historic base map contained some information related to underground components of the water system.⁴

Results

Base maps, surface features, radar cross sections, and surface projections of anomalous regions in the radar cross sections and photos of excavation units are presented in Figures 2 to 12. The horizontal axis labeling of the radar cross sections is the trace number that is also distance in decimeters. The vertical axis labels points along the trace. The vertical extent of the cross section is about 60 points or about one meter, based on a subsurface radar wave velocity of .08 m/ns. Anomalous regions along the cross section have been marked with a white box. The horizontal extent of the anomalous region has been transferred to the plan map of the previous figure and appears as a heavy black or grey line. Nine areas were examined with radar (Figure 2); results from three areas are reported here.

Figure 2. Base map of archaeological features highlighting locations of GPR project areas.

Area 5

Three north-south radar transects were acquired here along the remnants of a brick wall northeast of the boiler house, as shown in Figure 3. According to the 1979 historic base map, a tailrace outlet should be in Foundry Brook by the boiler house. Visual inspection revealed welling water in Foundry Brook here. Radar transects enabled archaeologists to note consistency with the historic map and to try to establish the origin of the welling water. An anomalous area consisting of a disruption of horizontal reflections appears between 12 and 15 meters along each survey line (traces 120 to 15), as marked in Figures 3 and 4.

Figure 3. GPR Area 5, including radar lines and interpreted anomalies.

Figure 4. Radar cross section of Area 5, Line 0. The white rectangle on the cross section between traces 125 and 150 (12.5 to 15 meters) is anomalous because the nearly horizontal background reflections present elsewhere in the cross section are disrupted.

The anomalies aligned with a tailrace outlet, several pipes identified on the 1979 historic base map, and with the welling water. Archaeologists excavated a 1 by 2 meter area over the anomalous area on Line 1, revealing a layer of cobbles and coursed brick underlain by brick rubble, as shown in Figure 5.

Figure 5. Excavation unit 15A, looking north. Notice water pooling along the northern wall and the masonry in the north wall. Photo by Mike Deegan.

Upon removal of the cobbles, water flowed from the northwest corner of the unit, quickly flooding it. Although archaeologists were not able to determine the cause or origin of the flowing water, it could form part of the tailrace or pipes identified on historic maps.

Area 6

The GPR transects were acquired here along two 30-meter lines east of the boring mill. Scattered brick, large pieces of iron conglomerate, and smaller artifacts such as iron nails and window glass covered the entire survey area. A slightly depressed area containing a large concentration of bricks bisected the northern end of each survey line. Because of these surface obstructions, only two survey transects were possible.

Visual inspection of the boring mill revealed a tailrace inlet on the northeastern wall. The 1979 map identified this race as a 45-degree line extending from the boring mill to the southern end of Foundry Brook. If the location of the race is correct, related anomalies would appear in the first 15 meters of each survey transect. Survey results revealed little activity in the southern half of the survey area and several anomalous areas along the northern half of the transects. Figure 6 illustrates the location of survey transects and anomalies.

Figure 6. GPR Area 6, including radar lines and interpreted anomalies.

The radar cross section for Area 6, Line 0 is shown in Figure 7.

Figure 7. Radar cross section of a portion of Area 6, Line 0, outlining an anomaly between 13 and 15 meters.

GPR data filtering removed background noise and clarified the presence of a region of horizontal reflections at 12 to 14 meters on Line 0. A shallow brick-filled depression was present at the surface slightly north of these reflections. The survey team excavated the region of the horizontal reflections, hoping to reveal the edge of associated subsurface features. Excavation revealed a brick structure interpreted as a machine mount in the southwestern quadrant of 4D, shown in Figure 8.

Figure 8. Feature 4, a machine base made of brick in the southwestern quadrant of unit 4D. Photo by Mike Deegan.

Area 7

This survey consists of six parallel transects in a 15 by 20 meter area south of Battery Pond. GPR transects were placed in this area to determine consistency with the historic base map and to search for a subsurface drain connected to a pipe outlet at Foundry Brook. Several anomalies appeared in the radar cross sections, one of which appeared consecutively and consistently in lines 0 through 4. Figure 9 is a sketch map of Area 7, including the location of the anomalies.

Figure 9. GPR Area 7, including radar lines and interpreted anomalies. STP1 was a 1.5 by 1.4 meter unit south of Battery Pond. Its location was chosen because of the strong reflections at 3 to 4 meters on Line 3 (Figure 11) and on a feature labeled drain on the 1979 historic base map, as well as the presence of a drain hole with flowing water that aligned with an active drainpipe emptying into Foundry Brook.

The GPR survey and visual inspection of this area revealed features related to water draining from Battery Pond into Foundry Brook. A considerable quantity of water was seen leaking from Battery Pond and flowing into a hole south of the pond, presumably into a subsurface drain. Visual inspection of the area also revealed water flowing from an iron drainpipe into Foundry Brook slightly downstream from the surface drain hole. The surface above the presumed location of the iron pipe is somewhat depressed. The radar cross section is shown in Figure 10. Primary anomalous regions are (1) a region of horizontal reflections from about 3 to 5 meters on the line (traces 30 to 40), flanked by downward opening arcs; and (2) downward opening arcs and other disruptions of reflections from about 8.5 to 10 meters along the line (traces 85 to 100). Anomalies on the GPR cross sections aligned with the presumed buried stone drain and surface depression. Based on the surface and radar features, archaeologists placed a shovel test pit (STP1) on the anomalous region at 3 to 5 meters on Line 3. Excavation of the test pit revealed a stone drain channeling water into Foundry Brook from a drain hole south of Battery Pond as shown in Figures 11 and 12.

Figure 10. Radar cross section of GPR Area 7, Line 3. The white rectangle on the cross section indicates a region of depressed reflections that might represent the top of the stone-capped drain.

Figure 11. Looking down into excavation STP1 at Area 7, Line 3, showing a subsurface stone drain leading to Foundry Brook. Photo by Kimberly Finch.

Figure 12. Conceptual cross section of STP1.

A break between stones revealed water flowing through this drain from west to east. A water dye test conducted during excavation established that water flowing into the drain hole to the west flowed through this drain and exited into Foundry Brook at the iron pipe visible southeast of the shovel test pit.

Conclusions

Results of the 2003 West Point Foundry GPR surveys demonstrate the effectiveness of geophysical exploration on an industrial site. In particular, the excavation results establish a relationship between the appearance of an anomalous region on a GPR cross section and the actual object in the subsurface. Despite several external factors that hindered GPR projects, including frequent rainfall and significant concentrations of surface and near-surface features, the survey identified the location of several subsurface features and helped identify areas for excavation.

The rough terrain created a major difficulty in deploying the cart-mounted radar antennas. The wheels did not make continuous contact with the ground surface, so the odometer did not trigger the radar reliably. The problem was solved by triggering the traces manually according to position on a measuring tape laid along the transect. Future GPR surveys should employ separate antennas or a cart or sled more suited to the rough ground at the West Point Foundry site. For example, separate antennas placed manually along the line would provide better maneuverability and stability, even though that procedure would be slower.

The 250 MHz antenna used at the West Point Foundry provided subsurface visibility to approximately 1 meter across most areas of the site. As discovered during excavation, most features associated with the water system lay between 0.5 and 2 meters below surface. Thousands of small objects, such as nails, iron fragments, and slate, also lay close to the ground surface, creating point reflections that obscured deeper features. Although computer filtering could remove some of these reflections, a lower frequency antenna could provide images at greater depths. GPR surveys at the West Point Foundry demonstrated the overall effectiveness of radar at an industrial site. Despite surface and subsurface materials, as well as a very saturated matrix, radar delineated several anomalies across the site. Using radar surveys, archaeologists were able to examine large areas of the site and objectively place excavation units, thereby preserving subsurface contexts for future investigation.

Acknowledgements

Author Charles Young supervised the radar fieldwork and processed the radar data. Author Kimberly Finch interpreted the radar data and supervised the archaeological excavations. Additional assistance in the field was provided by students enrolled in the 2003 Archaeology Field School at Michigan Technological University. Financial support was provided by the Scenic Hudson Land Trust, Inc.

Notes

^1. Alicia Valentino, "Visualizing the Past at the West Point Foundry, Cold Spring, New York" (master's thesis, Michigan Technological University, Houghton, Mich., 2003); Kimberly Finch, "Waterpower: A Geophysical and Archaeological Investigation of the Waterpower System at the West Point Foundry, Cold Spring, New York" (master's thesis, Michigan Technological University, Houghton, Mich., 2004).

^2. Lawrence B. Conyers, Ground-Penetrating Radar for Archaeology (Walnut Creek, Calif.: AltaMira Press, 2004).

^3. John M. Reynolds, An Introduction to Applied and Environmental Geophysics (New York: John Wiley & Sons, 1997).

^4. Edward S. Rutsch, JoAnn Cotz, and Brian H. Morrell with Herbert J. Githens and Leonard A. Eisenberg, The West Point Foundry Site Cold Spring, Putnam County, New York (Newton, N.J.: Cultural Resource Management Services, 1979).

Content in the History Cooperative database is intended for personal, noncommercial use only. You may not reproduce, publish, distribute, transmit, participate in the transfer or sale of, modify, create derivative works from, display, or in any way exploit the History Cooperative database in whole or in part without the written permission of the copyright holder.