The modern community of Wanchese occupies an area of higher ground with a maximum elevation of around 10 to 12 feet (3-4 meters) above sea level on the south end of Roanoke Island (Figure 1). The higher area, a hammock supporting a climax forest of oak, hickory and pine, is generally circular in form with a peninsula-like extension to the southeast (Map references: USGS Wanchese and Oregon Inlet quadrangles, 7.5' series, 1953). The southeast- ward extension of high ground parallels the western side of the former Mill Creek, now Wanchese harbor, decreasing in elevation toward the south. The point and western side of the peninsula extension is defined by the course of Great Gut, a small stream which drains the marshes and high ground on the southern side of the hammock and flows into Roanoke Sound. A few meters out into the southern marsh on the north side of Great Gut is the Tillett site, recorded as site 31DR35 in the Dare County file of the North Carolina Archaeological Survey.

The site is a few meters west of State Road 1141 and its southern end is defined by a U-bend of Great Gut (Figures 2b, 3). Great Gut is variously referred to as a "canal" and a "stream"; in reality it is the latter, and is, more precisely, a small creek. Its form and course suggest an origin carrying run-off from the higher ridge, but as sealevel rose and the marsh encroached, it now also drains the wetlands. Since Great Gut has also served as a traditional access waterway to the southern margin of Wanchese, it has certainly experienced some maintenance of its channel, thus the possible origin of the "canal" designation. The new canal shown in Figure 3 is one of a series of waterways cut through the southern marshes in the 1950's for purposes of mosquito control. It intersects Great Gut at the southern tip of the Tillett site, and is now the main channel from the south side to Roanoke Sound. Boat traffic has abandoned Great Gut from the site eastward to Roanoke Sound and the channel in that section is silted and shallow. In former times, when the Tillett site was actively occupied, Great Gut or a stream ancestral to it probably served as the main access to Roanoke Sound.

The Tillett site is an oval area of higher ground surrounded by marsh except at the southern end where it fronts on the stream (Figure 3). Maximum elevation is only 1.3 meters, however, and from our investigation it is obvious that if the shell midden debris had not been deposited here in the past, there would be very little high ground. Originally, the site area was a higher elevation connected to the southeastern extension of the Wanchese main hammock. The highest areas of the site now exist on the northwestern side where the Green residence once stood, and southern tip where a greater depth of shell midden still exists. Present surface topography of the site, such as the depression near the central area (Figure 3), reflects the numerous cultural modifications it has experienced in the past. These include (Page 9) borrowing of much of the shell midden prior to 1895, construction and activities such as the boat slip (Figure 3) associated with the Green residence, and the more recent bulldozing by Mr. Tillett. The latter operation alone removed at least 30 centimeters of midden from the higher elevation on the south end.

Figure 3. Topography and Excavation at the Tillett Site, 31DR35. (Page 10)

Modern vegetation on the site in 1979 is summarized in Table 2. The lone maritime forest on the northeast elevation was primarily red cedar, yaupon and wax myrtle with a few persimmon trees. Isolated red cedars stood in the central and southern areas of the site (Figure 2b), but most of the area was covered with a head-high growth of consumption weed interspersed with wax myrtle and other plants. The surrounding marsh contained needle rush and salt meadow hay in a normal successional sequence as well as other wetland species.

In addition to the plants listed in Table 2, which were collected and submitted for botanical identification, we observed but did not collect poison ivy, blackberry, morning glory and cat brier.

Except for the marsh, the other plant assemblages on the site reflect the amount of disturbance experienced there over the years. In 1972, there was a more extensive maritime forest cover on the southern and northwest sides of the site, but this was apparently destroyed by the 1974 bulldozing. (Page 11) A soil survey and mapping of Roanoke Island is still in progress, but advance sheets and descriptions provided by the Soil Conservation Service classify the soil on the Tillett site the same as the surrounding marsh (Roanoke Island Advance Sheet 1051). The soil type is Hobonny (11), described as a deep, organic marsh soil, frequently flooded, and qualifying as "wetland" for purposes of development or disturbance. As is often the case, small but distinctively different areas not easily differentiated in aerial photographs are included within the larger mapping units which encompass them. The base soil on the Tillett site proper is a sand, now buried beneath the cultural deposits, and the Hobonny soil of the marsh encroaches upon the edges of the more ancient sand base. The actual surface soils on the site are either dark organic midden soil with shells or dark humic soil where midden is not present. In some areas of disturbance, remnant midden and/or features were found below the more recently accumulated dark humus. In all cases, the surface and subsoil zones were wet from the high water table, enhanced by abundant rainfall at the beginning of the project. The wet soils and high water table made this site one of the most difficult to excavate in all of the author's experience.

As previously mentioned, the 1972 work at the Tillett site was done in five working days with a crew varying from three to four students primarily engaged in a training exercise. This included setting in a reference point for horizontal and vertical control, excavation of one test square (A) to a depth of 1.1 meters, excavation of one 10-centimeter level in another square (B), mapping these units, collecting the site surface where possible, and back-filling the squares. Both squares were located on or near the slight elevation at the southern end of the site (Figure 3), as was the reference stake. The 1974 grading activity removed the stake so that it was impossible to plot the squares in relation to the 1979 work. Feature 7 (see Feature section), exposed in 1979, may be the 1972 square A, but not enough of the outline was uncovered to be sure. The information learned from square A was important in understanding the stratigraphy and disturbances in the site. The square was excavated in 10-centimeter arbitrary levels, the top eight of which were mixed soil from a refilled excavation of unknown origin; below this, levels 9 and 10 contained intact shell midden and the humic-stained top layer of the base sand. Level 11 was culturally sterile sand and lay 10 centimeters below the ground water level. The excavated soil was screened through standard 1/4" mesh sifters. Upon preliminary analysis of the recovered material it was found that level 9 contained only earlier Mount Pleasant ceramics, and the mixed levels above (1-8) had both Mount Pleasant and the shell-tempered Colington ceramics. Both lithic material and well-preserved faunal remains were present, giving some promise for less disturbed areas of the site. The.1972 experience was considered in planning for 1979 by renting a pump, providing extra time for clearing vegetation and requiring all crew members to have a pair of knee-high rubber boots. (Page 12)

All cultural data from the 1972 tests and surface have been incorporated with that from 1979 in the analysis.

The 1979 field portion of the project began on May 21 and was completed on June 30. The first and last days were consumed moving equipment and supplies from Greenville to Manteo, and reverse; in between there were 30 working days, including one Saturday (June 23) of the thirty days, 7.5 were expended in laboratory work because of rain, leaving 22.5 actual days of on-site labor. Laboratory facilities for basic tasks such as washing, cataloging, preliminary analysis, drafting and carpentry were made available to us at the Marine Resources Center, Roanoke Island, and the grant provided funds for the rental of two cottages as crew quarters.

On-site work began with clearing paths through the head-high consumption weeds on the site to determine its surface extent and obtain a collection of specimens. This proved to be practically useless for specimen collection, but did permit a visual estimate of the distribution of shell midden debris. After cutting an entry path and clearing a central area for tool storage, radial and peripheral lines were cleared to expose a sample of the surface. These paths can be seen in Figure 2b. From the exposure afforded by the clearing, the extent of midden debris on the surface was roughly plotted and directions for the grid baselines selected. Even with the clearing, surface visibility was improved to approximately ten percent, about seven percent greater than it had been in 1972. The combined total collection of artifacts, predominantly ceramic sherds, from 1972 and 1979 numbers 172 specimens, not including some fragmentary animal bones. Part of the 1979 collection was obtained from the eroding shell midden area along Great Gut (Figures 3, 4a) so the actual upland surface total is very small for a site of this size.

A grid orientation was selected that best suited the generally oval site surface area with the north-south baseline along the long axis of the oval. Grid-north was set at N 45o E to magnetic north and the baselines were cleared and staked (Figure 4b). The only closely adjacent point of known elevation was the bridge carrying S.R. 1141 over Great Gut, and its deck elevation was obtained from the Department of Transportation Bridge Maintenance Division in Elizabeth City. Our reference point 1 was a stake driven even with bridge deck elevation at the northwest end of the bridge (Figure 3). From this, a second reference point (2) was established near grid center to facilitate readings on the site and topographic mapping of the site was accomplished. Elevations were taken as depth controls in excavated units and twice daily readings of the water level in Great Gut were recorded to provide site water level correlations and the effect of wind tides in the stream. (Page 13)

Figure 4. Views of the site area. (a) Looking west along Great Gut with shell midden exposed by low water. (b) View south along the north-south baseline from grid center after clearing. (Page 14)

(a) (b)

After the north and east grid lines were established, a series of nine transect tests--30 centimeter diameter shovel holes--were excavated to supplement the surface observations of site size in areas where no midden could be seen or where depth of deposits was questionable (Figure 3). The transect tests were recorded by their grid coordinates. Surface extent of the site was approximately 120 meters north-south (grid) and 70 meters east-west, somewhat smaller than the 600 by 200 feet reported by Talcott Williams in 1896. Over much of that area, however, there were little or no subsurface deposits remaining, probably because of the recent (1974) grading activity.

Excavations included four 1 x 1 meter squares (-1L13, -1L16, -1R9, -1R16), six 2 x 2 meter squares (30R2, 20R21, 4L4, -42L6, -48, -56R2), and the -42 trench area (two 2 x 2 meter squares -42 and -42L2 plus three 1 x 1 meter extensions -41.5R1, -43L1 and -43L2 opened to completely expose Burials 2, 3 and 4) (Figure 3). All of these units were excavated by 10-centimeter arbitrary levels except the -42 trench area which was excavated in zones (observable stratification layers), which by that time had been recognized. With the exception of square 30R2, which was completely disturbed and excavated to a depth of 10 centimeters, all squares were opened to the sterile sand subsoil level. The placement of squares was selected on the basis of two factors: (1) to test the stratigraphy of the site; and (2) to open as much area as possible in the most productive parts of the site.

Dry sifting through standard 1/4" mesh hardware cloth screens was attempted at first, but the organic soil was too wet and sticky to sift. Only later in the project when the upper levels of soil had dried somewhat, and then only in squares on higher elevations could this method be used. By the second week, 4 x 5 feet frames on legs with window screen over 1/4" hardware cloth had been constructed, and all excavated soil from squares, features and burials was processed through these. The excavated material was washed down in the screens using the pump and hose, the soil was trapped in a catch-basin below the screen, and the water ran into Great Gut (Figure 6b). This method proved to be more efficient both in processing time and in maximum recovery of smaller specimens.

The most detrimental condition during the project was the extremely wet condition of the soil and the high water level in the site. Both rain and tide level contributed to this. During rainy spells, it was beyond the capacity of the pump to bail enough water from the excavations to permit controlled recovery and squares were often abandoned until they dried sufficiently to continue the work. Squares closer to Great Gut were continually influenced by the water level in that stream and constant pumping in their lower levels as well as features and burials was necessary (Figure 6a). These conditions caused frequent slumping of profiles and made control of excavations difficult. Every archaeologist should have the opportunity of trying to trowel smooth the bottom of a level that is the consistency of pudding. (Page 15)

Standard field recording procedures and forms were used in the project, and photography included both color and black-and-white. Washing and cataloging of specimens began in the field laboratory but was completed at the Archaeology Laboratory following the field season. The specimens, data and photographs from the project have been permanently curated at East Carolina University under accession numbers 229 (1972) and 1181 (1979). Following cataloging, materials were separated for submission to the specialists whose reports are appended here, or who are mentioned in the text. Special analyses included modern botanical, radiocarbon, paleobotanical, faunal and fish remains, and human skeletal remains. (See Appendices A, B, C, D, E). The human material was preserved and reconstructed at East Carolina University prior to submission to the physical anthropologist. Analysis, comparative studies, and photography of all cultural material was performed at East Carolina University. (Page 16)