The National Geographic Society (NGS) field team arrived at Pettigrew State Park on September 18, 1992. Equipment was offloaded and stored in the park headquarters, and a campsite erected for project personnel in the park campground. At 8:00 a.m., September 19, a preliminary planning conference was convened between Park Superintendent Sidney Shearin, Claude Petmne and Donald Shomette, at which it was decided that a preliminary shakedown operation would be conducted in the vicinity of canoes Nos. 5, 6, and 7.

Following the meeting, mobilization and field testing of equipment commenced. NGS and park personnel began measuring, cutting, and hand spooling 30,000 feet of twine into 60 spools of 500 foot lengths. The twine was to be employed in the erection of lane guides in each of the three transects. Simultaneously, an overland reconnaissance of shoreline access points to the projected B and C transacts was conducted between Thirty Foot Canal and Western Canal. Following the reconnaissance, it was decided that all land approaches to the two transacts, except across a narrow open stretch between the foot of Mountain Canal and the lake were impractical. All vessel launches would have to be by water and launched from the park landing or hand carried and launched at the Mountain Canal access area.

A shakedown operation for the GPR and GPS systems for the purpose of calibrating and testing survey gear, and a hands-on evaluation of the environmental conditions west of the vicinity of Transect B was planned for the afternoon. Owing to a malfunctioning engine, the principal survey vessels had to be towed to the proposed field testing area by the park's own Boston Whaler. Nevertheless, two survey craft, a John boat and a rubber inflatable, were successfully deployed, and towed to the approximate location at which Nos. 5, 6, and 7 canoe sites were situated.

The John boat, in which the GPR unit and strip chart recorder were mounted, was connected to the inflatable, in which the smaller of the two GPR antennas, the 500 mghz unit, was mounted, by a six foot cable. The strip chart recorder assemblage and antenna were, of course, connected by their own cable. This was done to prevent "noise spikes" or electronic interference generated aboard the survey boat and from extemal sources from influencing or distorting the data received by the antenna. Later, a plastic "raincoat" was affixed over the antenna cable connection to prevent weather and water damage to the unit. Waterproof tarpaulins and "space" blankets were employed to protect the GPR and recorder during inclement weather. The vessels were propelled by manual effort, that is by being simultaneously pushed and pulled by project personnel walking in the shallow water. (Page 54)

The shakedown testing was conducted in waters ranging from two and a half to four foot depths. The bottom sediments alternated between soft, black, water-charged clayey organic muds to hard panned, finely laminated sand and organic-rich clay, and laced with occasional bald cypress knees jutting up from the bottom. Surface conditions were generally mild, with some wind-generated wave activity causing visibility in the water to drop to less than a foot. Surface air temperature was in the mid-80s. Initial deployment of the 500 mghz antenna proved unsatisfactory owing to poor data reception, and it was decided that the 300 mghz unit should be tested. Signature quality was noticeably improved. It was decided that the 300 mghz antenna, although bulky to deploy, and frequently acting like a sail on the Zodiak which caught the wind and caused the raft to sway from its course unless manually handled, would be used throughout the remainder of the survey.

During the shakedown, the evaluation of bottom data and an attempt to begin interpretation of representative GPR signatures was undertaken. Surficial sediment texture and subsediment density, was examined using probe rods, visual observation, and by touch in a preliminary effort to correlate signature imagery with the environment which was to be surveyed (see Appendix A, sample Signatures Nos. 1 thru 10). Without the survey team's previous knowledge of its precise location, canoe No. 5 was located in three feet of water approximately 90 feet from the shore using only the GPR. The canoe, which was found to be in several pieces buried beneath approximately six inches of soft black eolian sediment, was deemed suitable for use as a "phantom" site, from which preliminary model signatures could be taken. Survey runs were conducted over and adjacent to the site (see Appendix A, Signatures 4-10).

Although no soil samples were taken, it was determined that the bottom environment in which canoe No. 5 lies possesses a soft mantle of fine grained black sediment approximately a foot thick, underlain by a hard lens of very fine and silty sand. The canoe appears to be resting on the hard sand, which, from the signatures and probing, appear to be approximately four inches thick, and underlain by additional alternating lenses of soft and hard sands and clayey sediments (see- Signatures 1-3).

The signature range runs from minimally interpretive to pronounced and identifiable, and was felt to be directly related to the density of the soils over which the antenna was deployed, as well as the calibration. Initial efforts focused on deep penetration rather than resolution. Hence, surficial imagery proved far less qualitative than desired. Signatures 4, 6, 7, and 10 were produced by running over the four ends of the vessel's two discrete and exposed components. Signature 10 was taken over the shovel-shaped bow of the larger of the two components, which was hand cleared of sediments, and is the most clearly definable. Signatures 5 was produced by running alongside the south side of the larger of the two components, but not directly across it. Signatures 6 and 8 were taken by running directly across the midships of the larger of the two components of the vessel. Signature 9 was made (Page 55) by running diagonally across the central hull of the larger canoe. Examination of the site by divers failed to locate the identification tag which had been afixed to the hull after its discovery in 1986. Thus, the canoe site was marked with a vertical pole to permit rapid relocation by the UAU team for later reverification of its identity, and a re-evaluation of its condition.

The successful mobilization effort effectively proved that submerged and buried cultural resources could be located in Lake Phelps using GPR technology. Moreover, it produced a small signature library against which like targets documented in the surveys of each of the transects could be compared. However, the quality of imagery was considered still somewhat less than adequate for the identification of like sites in the Lake Phelps environment, and further experimentation with gain, time, and equipment was anticipated.

On September.20, 1992, a second planning meeting was convened at Pettigrew State Park headquarters. In attendance were Lawrence, Shearin, Petrone, and Shomette. At the meeting, it was decided that the sequence of investigation would begin at Transect B, rather than Transect A. The transect's easternmost end would be established 1,000 feet northwest of Thirty-Foot Canal, and fifty feet from the present shoreline, in lieu of the proposed transect area slightly further to the west. This was agreed upon in order to address immediate logistical concerns, and because absolutely no prior organized investigation of the area had been carried out owing to environmental considerations. Moreover, survey craft could be transported to the site without the necessity of power, either by oar or by being towed manually, which was of importance owing to the lack of an engine for the John boat (which would remain under repair for the duration of the survey), engine problems being experienced with both of the UAU vessels, and the necessity of the temporary deployment of the park boat elsewhere. It had unfortunately been discovered, too late for the survey, that direct shorefront access to Transect A would be difficult, if not impossible, in that the shoreline property was under private ownership and permission to cross the property had not yet been secured.

Shearin noted that the bottom land in the newly proposed Transect B area was typified by deep depositional silts which had only recently been built up. He attributed the buildup directly to the erection of the park landing, which projects several hundred feet into the lake to the immediate east of the new Transect B area. The dynamics of this buildup, and the erosional activity along the shoreline to the immediate east, is common to any area in which wind-driven wave activity is pronounced. The size of a wave is directly related to the strength and duration of the wind, and the distance of water over which the wind blows, known as fetch. Because fetch is limited on Lake Phelps, waves are, of course, small in size. Yet waves approaching the shoreline at an angle create (Page 56) longshore currents that can transport sediments along the beach or shoreline. Sediments can move along shore in either direction, but over the long term, a "net" transport direction is typically established in response to the largest storm waves [Ward, Rosen, et al, 12-13]. In Lake Phelps, the prevalent direction of incoming waves are from the south-southwest, from which direction the most prominent storm weather activity moving across the lake also appears [Shearin, p.c.]. Hence, buildup on the updrift side of the landing (the west), and some erosion on the downdrift side (the east), caused by longshore transport of the sediments, is noteworthy. It is all the more important in that the effect of sediment buildup may well have led to the masking of potential archaeological resources in this area.

Transect B was to be 500 feet in length and 200 feet in width, divided into 20 discrete survey lanes, each ten feet in width and 500 feet in length. Lanes were to be designated alphabetically, A through T, with A being the northernmost (or shoreward) lane, and T the southernmost (or seaward) lane. A total of sixty-three 2" X 2" stakes, each five feet in length, were to be driven into the bottom by hand at each end comer of every lane, and at the 250-foot midpoint. Each lane was separated from the next by twine lane guides run from stake to stake. The position of the transect corners, and each lane guide was surveyed in using a hand-held KVH Datascan (an integrated monocular electronic fluxgate compass and range finder), and measuring tape. The longitude and latitude of the four corners of the transect were recorded using the GPS, and by taking visual shore-bearings off of prominent landmarks. Survey and construction of the transect was carried out by three persons and was completed in 4.5 hours.

The GPR survey was conducted in a similar fashion to the shakedown work of the previous day, with some modifications. Despite initial plans that the survey could effectively be carried out by a field team of four persons, a survey team of seven to eight persons was found necessary to adequately conduct maximal controlled operations and ground truthing. Personnel were deployed as follows: one person to operate the GPR; two more to push and pull the two survey vessels; a fourth, designated the "marker," to walk beside the GPR antenna/raft assemblage to tie day-glo colored marker flags to the lane lines at each target determined by the GPR operator (or any other surface anomaly discovered); a fifth to assist the GPR manager, take notes, and distribute marker flags to the "market"; two more to conduct probing operations with five-foot-long stainless steel probes, and undertake visual underwater inspections of surface anomalies; and an eighth to conduct underwater excavations of high-probability targets using standard underwater airlift and water jet systems.

At the beginning of each lane survey run, the time and directional orientation was noted and marked down on the strip chart record. The precise moment of the end of the lane survey was also noted and marked. As each suspect feature or cluster of features of significance appeared on the strip (Page 57) chart, or as surface anomalies were encountered by personnel in the water, the GPR manager would call for a "mark." The marker then tied a colored and numbered flag at the point of the hit, on the landward side of the lane. Targets were numbered sequentially. Hence, a target designed D-3 was the third target noted in lane D, and so on. Not all targets, however, were noted or marked during the survey, but were determined to be worthy targets afterwards or during later reviews of the strip chart data for mapping purposes by this author. These were addressed with an alphabetical suffix followed by a numeral. A target thus noted might have a designation D3A-1. These were converted for the purpose of this report to numerical and sequential listings, and will appear as such herafter in this report, in the signature library, and on the target distribution map.

After being marked, a target was then investigated by probing, and/or hands on excavation. Small artifact concentrations, such as potsherds, which were deemed suitable for recovery, were then recovered and bagged in zip-lock bags. Larger items, such as canoes, logs, etc., were cleared and recorded at Lawrence's or Shomette's discretion, but not removed. Soil samples and organic material samples were also recovered during excavation of a site and bagged in zip-lock bags.

Following the survey, a classification system of distinct sub-bottom stratum formation signatures was developed for both Transects B and C. These appear in Appendix B. A total of 43 discrete targets, probable targets, or significant geological feature were located in Transect B, within 100 feet of the shore of shore. A total of four targets, designated 29 A-2, 42, 51 A-1, and 54 were deemed significant enough for excavation or clearance. Three targets, 29 A-1, 51 A-1, and 54, were buried in over three feet or more of sediments. Target 42 proved to be a partially buried cypress log, lying prone, and approximately 7.5 feet in length. Investigation of the area surrounding target 54 revealed additional disarticulated fragments of the log, and probable root structure lying beneath the sediments. Targets 29 A-2 and 54 were determined to be extremely concentrated, hard-panned lenses of sand.

A complete excavation of target 51 A-1, located in 35 inches of water, was undertaken on September 22-23 a stratigraphic profile was produced, and soil and organic samples recovered for C-14 dating. The target was temporarily removed from the site to determine its identity, and to examine underlying strata and organic materials. The upper surface of the target was encountered at a depth of 26 inches below the surface of the bottom, and proved to be timber, six feet in length, 13 inches in width at its center, and approximately 2.5 inches thick at its center, flat on one side and slightly rounded on the other. Both ends were rounded, giving the piece much the same appearance and shape as a surfboard. Temporary removal of the subject revealed that it was lying in a four-inch- thick bed of gray-black clay, one inch above a fiberous bed of organic root-like material and peat (samples of which were recovered), and adjacent to broken branch or root fragments of a cypress log. (Page 58)

Both the fiberous material and the branch or root fragments showed considerable evidence of having been burned. The surfboard-shaped wood bore no apparent evidence of having been worked, although its entire surface seemed abnormally smooth and its upper surface was perfectly flat. Following the examination of the piece and removal of a fragment for C- 14 dating, it was redeposited in the excavation hole.

The stratigraphy of the excavated area closely corresponds with core samples produced by Holley along the north shore rim of Lake Phelps, although the former appears more compacted. The uppermost layer ( A) consisted of 13 inches of black, water-charged organic muds, beneath which was 12 inches of dense, finely laminated quartz sand, sandy silt, and silty organic-rich clay with multiple lenses varying in color from very pale orange, to light gray, and black (B). Beneath this layer was four inches of densely packed black and gray clayey soil (C), in which the target was buried. And, finally, beneath this lay the peaty material and burned wood ( D). Holley notes that "large in situ stumps (up to 1m in diameter) and large cypress logs (up to 5m long)" are common along the north shore rim of Lake Phelps, but makes no mention of them being found further down. Nor had he produced C-14 dates from organic materials or organic sands ftm this lithologic unit [Holley, 27- 44]. Unfortunately, funds necessary for the C-14 dating of the various samples recovered from 51 A-1 were not forthcoming. As of this writing, no positive date for the various strata or from the wood samples recovered have been determined. (Page 59)

Following the completion of the Transect B Survey, examination of three previously discovered canoe sites, Canoe Nos. 18 and 19, located in the vicinity of the outlet of Moccasin Canal, and Canoe No. 7, several hundred yards west of the Mountain Canal access area, was undertaken by Lawrence, Shomette, and Buck. All three sites were examined to determine whether plastic identification tags affixed to the ends of the canoes by copper wire during the 1986 survey had survived.

Owing to a driving rain, the surface conditions of the lake were flat calm, and underwater visibility was excellent. All three canoes, could be plainly viewed from the boat. A hands-on examination of No. 18, which is dated at 750 +/- 80 B.P., revealed the vessel to extend from a small cypress grotto southward, with its southern end partially covered by white sand and a film of mud in approximately two feet of water. Its northern end lying in 18 inches of water, extended into the grotto and was encompassed by the knees and trunk of a bald cypress tree which had grown up around it. Canoe No. 19, dated at 1740 +/- 60 B.P., was found to be exposed but partially covered by sediments of sand and mud in approximately three feet of water. Canoe No. 7, dated at 4380 +/- 70 B.P. and the oldest canoe located in the lake, was found lying in a shoreline marsh, in a NW-SE orientation, on a sand and mud bottom, in two feet of water. Lawrence noted that since last being observed in 1986 the vessel appeared to be in excellent condition, but seems to have migrated slightly. He suggested the factors causing the apparent migration might be wavelet activity, or the gradual encroachment of the shoreline upon the vessel. No identification tags were found on any of the vessels.

Following the Transect B Survey and the reconnaissance of Canoes 7,18, and 19, and after consultation with Lawrence and the UAU staff, it was decided that the location of Transect C should be moved from the proposed survey area off Big Point owing to continuing logistical problems. Moreover, Lawrence suggested that the buried canoe population in a more accessible location, in the vicinity of the Mountain Canal access area, might well equal that of the known population of four vessels in that reach (Nos. 10, 11, 15, and 16). It was thus agreed that Transect C be erected here. However, owing to the cogent observations of Lawrence and Shearin that the vast bulk of canoes which had been located to date were ensconced within 150 feet of the shoreline, it was decided to adjust the transect from the (Page 65) proposed 500-foot-long by 200-foot-wide format to one which was 1,000 feet in length and 100 feet in width, divided into ten lanes, again running SE to NW parallel to and beginning fifty feet from the shore, rather than the propopsed 150 feet. Moreover, a known survey datum position recorded during the 1986 investigation, noted hereafter as "T", would facilitate the erection of the transect.

Owing to a survey error, the format of the transect was that of a parallelogram rather than a rectangle but, nevertheless, served equally as well for the mission at hand. During the morning mobilization period, precise measurements of water depth at three sites, Canoes Nos. 5, 6, and 7, all of which were rapidly silting over, were carried out. It was readily determined that the lake is rapidly approaching its historic norm, and is gradually inducing the reburial of the briefly exposed canoes.

Transect C was erected in approximately five hours and was completed on the afternoon of September 22. The following day the survey was carried out despite inclement weather and the intensive infestation of the survey area by airbom insects, the latter of which frequently fouled the GPR strip chart recorder, causing minor technical problems and some delays.

Two known canoes, No. 10, dated at 1530 +/- 60 B.P., and No. 15, dated at 1530 +/- 60 B.P., were known to be in the transect area, both of which were buried. Neither canoe was immediately discemable on the strip chart record, although the sites were run over at different points a total of six times. Canoe No. 10 produce no recognizable signature, though passed over three times (see Signatures 79, 82, and 90 in Appendix A). Canoe No. 16, run over thrice, also failed to produce a recognizabe signature on two runs (see Signature 86 and 94 in Appendix A), and a signature which was initially overlooked and interpreted as a target only during a re-evaluation of the record after the survey (Signature 91).

A number of targets were, however, identified. These included root structures (Signatures 55, 58, and 59), exposed cypress knees (Signatures 60 and 61) and buried logs (Signatures 64, 65, and 66). Unconfirmed but probable identifications included stumps and roots (Signatures 1 and 68), buried logs (Signatures 62 and 83), sediment depressions or dimples (Signatures 55 and 56), overlapping soil horizons (Signature 85), and external disturbance of sediments (Signature 84). Two targets (Signatures 87 and 88) were believed to be surface blisters (buildup of sediments either from surficial water activity or sediment bulges resulting from uneven underlying strata platforms). One target proved to be a machine error (Signature 96), and eleven targets were unidentified (Signatures 63, 69, 70, 71, 74, 75, 81, 89, 92, 93A-1, and 95A-2). (Page 66)

Five targets were found to be of similar typology to three signature forms produced by Canoes No. 10 and 16. Signature 67 is similar to Signature 148 of Canoe No. 10; Signatures 72 and 73 are similar to Signature 91 of Canoe No. 16; and Signatures 77 and 80 are similar to Signature 160 of Canoe No. 16. As these comparisons were not made during the survey period, none of these sites were tested beyond probing, which proved inconclusive.

Following the survey, Canoes Nos. 10 and 16 were relocated by the UAU team and marked with poles. Preliminary evaluation of the strip chart data suggested that potentially significant targets lay in the survey area, and that less penetration and greater resolution data was more desirabe from the GPR. Owing to the continuing inaccessibility of Transect A, the potential richness of Transect C and the acceptable but still unrefined performance of the GPR, it was decided that Transect C, should be resurveyed.

The survey mode employed during the previous surveys was not altered for the resurvey of Transect C. Recalibration of the GPR, however, was considered necessary. Hence, prior to beginning the survey, a testing period was carried out on Canoe No. 11 and Canoe No. 15, to the immediate east of the tran sect area. Canoe 11 is dated at 1790 +/- 70 B.P., and Canoe 15 is dated at 1530 +/- 60 B.P. Both vessels were partially buried beneath the bottom, with only the tips of their gunnels exposed. Two feet eight inches of water lay over the targets.

A total of 21 runs were made over and about the two canoes (seven on Canoe No. 11, and fourteen on Canoe No. 15). A variety of GPR calibrations, ranging from 300 X 1 to 500 X 2 nanoseconds, all at 32 scans per second, were tested to determine the most appropriate resolution for producing an identifiable signature. These are presented as Signatures 172 through 192 in Appendix A. Test runs were conducted at varying distances along either side of the targets, usually at a distance of four to five feet, diagonally across the hulls, straight down the hulls, and across the hulls. From the data gleaned, it was thus determined that the most appropriate GPR setting for the resurvey of Transect C should be carried out on a System 4, 300 mghz, at 350 X 2 nanoseconds, at 32 scans per second. The survey of the entire transect, owing to poor weather conditions brought on by the onset of Hurricane Danielle, could not be completed. However, Lanes F through K were completed, and signatures of natural, geological, or man-made targets were recorded (see Signatms 97-148 in Appendix A). An additional 23 signatures were recorded during further (Page 67) testing of Canoe Nos. 10 and 16, and of areas in the immediate vicinity, and, with the exception of the canoe sites, two pottery assemblages which were discovered, and a stump, without geographic provenience (see Signatures 149-171 in Appendix A). Simultaneous to the survey, a test excavation was carried out near the eastern end of Lane J to examine the strata for comparitive analysis to the excavated site in Transect B.

During the formal survey operation, Canoe No. 10 produced two signatures (Signature 102 and 118), while Canoe No. 16 produced two identifiable signatures (Signatures 125 and 133), and one run without identifiable signature (Signature 112). Three stumps were identified (Signatures 103, 104, 105). One target was positively identified as laminated or compacted sand (Signature 139), and another, as a probable buried log or tree branches (Signatures 126 and 130). Five signatures are believed to represent overlapping layers of sub-soils (Signatures, 100, 127, 131, 147, 148). Signatures 113, 122, 142, and 143 are believed to indicate slightly buried surface features. Unidentfied targets were indicated in Signatures 97, 98, 99, 101, 106, 107, 109, 110, 111A-1, 111A-2, 115A-1, 120, 124, 128, 129, 134, 135, 136, 141, and 145. Unidentfied hard targets which shut down the GPR appear as Signatures 121A-1 and 140. Three targets, Signatures 118, 119A-1, and 119A-2, are problematically addressed to surface disturbance of soils caused by obstructions of eithe a canoe or other features. Signature 137, which caused a GPR shutdown, is the negative area of the excavation pit in lane J. Signature 143 is a problematic deflection of the natural signature of Canoe No. 16, or possibly soil displacement or disturbance caused by the canoe presence.

Six targets produced signatures typologically similar to known canoe signatures. Signatures 115A-2 and 117 are similar to Signature 160 of Canoe No. 16; Signature 115A-2 is similar to Signature 163 of Canoe No. 16; Signature 123 is similar to Signature 186 of Canoe No. 15; Signature 138 is similar to Signatures 159 and 160 of Canoe No. 16; and Signature 147 is similar to Signature 161 of Canoe No. 16.

Five targets, Signatures 100, 108, 114, 121A-2, and 132, bear typological similarities to Signatures 167 and 169 of pottery sherds.

During the post-survey testing, six runs were made over Canoe No. 10 (Signatures 149-154), of which one resulted in shut down of the GPR unit twice, and a second was made alongside the hull. Five runs were made over Canoe No. 16 (Signatures 158-161, and 163). Signatures 164 and 165 proved to be logs, while Signature 170 proved to be a stump with root structure. Signature 167 proved to be prehistoric pottery, five sherds of which were recovered. Signature 169 was of seven pieces of prehistoric pottery discovered in association (Page 68) with burnt wood and then recovered. Signature 168 is typologically similar to 167 and 168. Theremainders of the Signature file, 155, 156, 157, 162, 166, and 171, are withoutprovenience or identification.

Owing to the onset of foul weather, and forecasts for three more days of the same, it was decided to cancel the remainder of the survey. On September 24 Transect C was dismantled by the UAU team and its raw materials deposited at Pettigrew State Park.

Prior to their own departure, a cursory investigation was conducted by the UAU team at a site scheduled to be developed as a park boat landing on the west side of Lake Phelps. An undocumented log canoe, No. 31, was discovered, upside down and resting on wind-driven flotsom in the nearshore area by UAU archaeologist Leslie Bright. The canoe was unique in that it possessed a small man-made hole in the lip of the gunnel at one end, possibly to facilitate the securing of an anchor line. No other canoes found to date possesses such a feature, although at least two probable stone anchors have been recovered to date from the lake's waters. Bright suggests that the vessel may well be from the Contact period or later. (Bright, p.c.; Phelps, p.c.).

On September 25 the National Geographic team completed demobilization. (Page 69)

The known submerged archaeological resources resting beneath the waters and silts of Lake Phelps, North Carolina, without question, comprise one of the most important prehistoric assemblages of its kind in the Western Hemisphere. The range, quantity, and type of those resources which have yet to be discovered and studied, based upon findings already made, must surely also be of enomious significance.

From the evidence assembled between 1984 and 1992, it is apparent that further investigation will undoubtedly yield additional data. The September 1992 Lake Phelps Survey was of singular importance, however, in that the systematic employment of sub-surface radar for the exploration and discovery of prehistoric lithic and organic materials buried beneath the sediments of a fresh water environment is feasible. The development of the signature library acquired during the survey, the first of its kind, and presented within this report, is suitable for use as a baseline resource for any future GPR work done in Lake Phelps. Its utility for the evaluation of resources resting within similar environments will undoubtedly prove of considerable value for future archaeological investigation.

For the first time, the ability to successfully locate and identify buried prehistoric resources beneath the waters has been provided. Certainly, refinement and fine tuning of the technology will increase the capability of such research immeasurably. It is hoped by this author and his colleague, Claude Petrone, that future examination of the resource base of Lake Phelps will be possible.

From the findings of the 1992 survey, a number of recommendations for the practical improvement of GPR survey in the Lake Phelps environment, as well as survey goals for for future investigations should be presented herein.

1. The research on Lake Phelps and its superb resource base must continue, with a long range program capable of locating, analyzing, and understanding all of its prehistoric treasures. In the immediate future GPR survey should focus on: (a) the 31WH13 research transect; (b) the Old Canal (Colas Canal) region; and (c) select transacts along the south shore of the Lake, centering on the sector in which pottery sherds were recovered by Holley. The 31WH13 transect may prove to be an important area of research in that prevailing wind and current activity on the lake would possibly have caused the migration of abandoned log canoes toward that sector (as suggested by the serendipidous find of Canoe 31 during the last day of the survey). The Old Canal sector may yield not only prehistoric materials, as suggested by the findings on land at the 1982 Somerset excavation, but historic resources (Page 70) related to the first European and African-American habitation of the region. Exploration along the southern edge of the lake, an area entirely ignored thus far, would either confirm or deny current hypothesis regarding prehistoric settlement and usage patterns on the lake.

2. A long-term baseline study should be considered to systematically chart and record the dynamics of Lake Phelps as it pertains to shoreline migration and sedimentation. This will facilitate a more adequate understanding of the processes by which the prehistoric resource base in the lake was exposed and now seems to be in the process of being recovered.

3. Core samples should be collected at appropriate sites or target areas and analyzed in concert with the GPR record to provide a more comprehensive understanding of site characteristics, and the determnining factors of why sites are distributed as they are.

4. Future GPR survey work should incorporate the System 8 because of the built-in filtering capacity, a necessity for accentuation of weak feature signatures on the prehistoric beach.

5. Either a laser survey system or a traditional Mini-Rangar system should be employed for precise local positioning. Reliance on GPS is unacceptable.

6. More vessels should be on hand to improve the speed in laying out survey grids and for logistical needs. The principal survey vessel should be a large, shallow draft craft, with copious deck space to permit the employment of the GPR monitor and a 38VDU (visual display unit), and to permit the stacking of the various component modules. The vessel should also have adequate shelter to permit work in all weather conditions, and an enclosure to prevent the invasive by and destructive (to equipment) activity of insects.

7. Signatures of potential canoe and pottery sites located in the Transect C sector during the 1992 survey, but not tested, should be relocated, re-examined, and thoroughly tested.

8. The samples from 5lA-1 should be submitted for C-14 dating to determine the age of the lake at that particular site. Such data will also permit parameters of determination of the potential depth of canoe burials, at least within the previous 4,000 years, within the shoreline range encompassed by Transect C.

9. Research should be conducted to determine if it is feasible to synthesize collected sub-surface radar data into a layered stratigraphic map suitable for digitization and enhancement, dissection, rotation, etc. Such data would prove useful in developing and testing a predictive model on resources resting on and beneath the lake bottom. (Page 71)