One of the most difficult aspects of the Tillett site study was interpretation of the stratification. In the beginning, each new square presented a different sequence of layers, and we had only the limited and disturbed 1972 tests and present surface clues as a guide to interpretation. This early confusion was related to the extreme amount of disturbance and site surface modification from various sources over a considerable span of time; once these processes were understood, the original stratification and stratigraphy proved to be remarkably simple.

At some time prior to historic modifications of the site (pre-A.D. 1700), the basic stratification sequence consisted of three zones, or layers, of soil, each distinctive in color, content and texture. There are:

Zone I. A shallow humic soil resulting from organic decay on the floor of the maritime forest covering the higher elevations of the site. As surface elevation declined toward the edges of the marsh, this zone was replaced by peat or muck (Hobonny soil).

Zone II. The primary deposit resulting from cultural processes, this is a midden stratum of mixed dark, organic soil, shell, and other cultural debris accumulated to an unknown height from ca. A.D. 400 to the time of site abandonment. The layer is mostly shell with very little organic soil in square -56R2. Visual separation of cultural stratigraphy is not possible and component separation must be accomplished by mechanical methods of excavation control. (Page 16)

Zone III. A yellow-tan sand of unknown depth which was the surface layer prior to human occupation. The top of this stratum is now at or near normal sealevel in the site locality. The top 10 to 15 centimeters of the sand has a light gray color resulting from organic content, probably representative of humic accumulation from the maritime forest covering the original upland ridge (the peninsula, or "finger", which extends southeast from the Wanchese hammock), and also partly the result of organic leaching from the midden which subsequently covered the zone.

This sequence is best seen, much modified, in the profile of the southern section of the site from Great Gut to grid station -40 (Figure 5a). In the profile drawing, Zones II (midden) and III (base sand) have each been divided into A and B units reflecting the organic staining at the top of Zone III, and the upper section of shell midden disturbed by recent grading and other activities (Zone IIA). The grading disturbance extended to a depth of 10 to 20 centimeters and was visually discernible by the fragmented shells and greater percentage of dark organic soil. This is readily seen in squares -56R2 (Figure 6b) and 4L4 (Figure 9a). The western profile of square -42L6 (Figures 5b, 9b) presents the least disturbed exposure of the classic site stratification. Also a major disturbance in the southern section were large areas of previous excavations filled with mixed soils. One of these areas in square -48 extended to approximately 40 centimeters below surface (Figures 5a, 8b); a black, peat-like layer containing undecayed grass (black organic fill in Figure 5a) and cigarette butts was probably the surface layer at the time of that excavation and is an example of reversed stratigraphy from refilling the square. The area disturbed by previous excavation extended into squares -42, -42L2 and their one meter extensions but was not as deep as that in -48. Except for Feature 7 (see Features section), which may be our 1972 test A, the other excavations in this section of the site were probably those carried out in 1938 by Green and MacNeill. Local informants told us that many people have also "pot-holed" the site over the years.

Near the center of the site, the disturbed midden was exposed on the surface and had a maximum depth of 20 centimeters in square 4L4 (Figure 9a). Eastward from grid center, the midden disappeared between squares -1R9 and 1R16, and stratification in the latter showed a 20-centimeter deposit of disturbed humus with no cultural content overlying the Hobonny soil of the marsh that had encroached upon the base sand. West from grid center, evidence of shells on the surface extended nearly to Great Gut, but subsurface midden was found to end in square -1L13 (Figure 3). At the north end, square 30R2 contained a top layer of dark humus overlying a layer of Hobonny peat, which was obviously redeposited from previous disturbing activities. Below the peat, a shell-filled feature (Feature 10) was observed intruding into the Zone III sand, indicating that midden once covered this part of the (Page 18) site. Square 20R21, much disturbed by the Historic component activities on the higher elevations of the site (Figure 3), had a 10-centimeter layer of forest floor humus (Zone I) over the Zone II shell midden with an average thickness of 30 centimeters (Figure 7b).

(a) (b) Figure 6. Water level and stratigraphy in square -56R2, (a) Water level and erosion in the square with high wind tide in the canal; photo taken at noon after pumping all morning; view south. (b) The south profile of -56R2 immediately after pumping; Feature 5 exposed, Great Gut canal and wet-screening operation in the background. (Page 19)

An excellent geological study of the marsh and bay evolution in Broad Creek (Benton 1980) provided data directly applicable to the Tillett site. Broad Creek is located northeast of Wanchese and, like Great Gut, empties into Roanoke Sound (Figure 1). Benton-concludes that Broad Creek was an upland drainage during the late Pleistocene and at the beginning of the Holocene, and has since become a drowned, lateral stream embayment, a process which continues as sealevel rises. The upland surface soil through which the original stream cut its channel, his Unit 1 (Benton 1980:82), is generally the same as the Zone III sand at the base of the Tillett site. In the Broad Creek basin, this older surface sand has been progressively buried by deposits of peat as sealevel rose; at first the peats were of freshwater origin but as brackish water intruded, classic coastal brackish marshes developed. This shift from freshwater to brackish marsh occurred around 1500 years ago (ca. A.D. 400-500) in the Broad Creek basin and the southern side of Roanoke Island (Benton 1980:128). A radiocarbon date of A.D. 460 + 85 from Feature 9, a midden-filled depression in the base sand representing the initial accumulation of cultural debris on the site, places the beginning occupation coincident with brackish marsh development on the south side. There is little doubt that the selection of this site location is related to the procurement of shellfish, fish and other food resources made available by salt water encroachment in the sounds around the south end of the island.

From ca. A.D. 400-500 onward to the final prehistoric occupation, the accumulation of the Zone II midden was primarily a cultural process. Season after season, the site occupants obtained the food resources and discarded the remains (shells, bone). Based on the evidence of beginning midden accumulation on Colington Island (phelps 1981), the midden did not accumulate in a comprehensive, homogeneous pattern, but in smaller depositional units correlated with random selection of living and activity areas on the site through time. Although disturbance prohibited adequate study of these depositional patterns, there was some suggestion of a difference in square 4L4 where clam shells (Venus sp.) were in equal quantity with oysters (ostrea sp.), a different situation from that observed in the southern part of the site. Since the evidence of late occupations had been removed or destroyed, there is no way of accurately estimating the final date of occupation, but given stable environment and resources, the expectation of site use and resource exploitation at this locale probably continued to the end of the Colington phase around A.D. 1650 (Page 21) The final deposition in the original stratification sequence (Zone I), accumulated naturally from decaying vegetation on the maritime forest floor after the site was abandoned. Few areas of the site preserve the original humus, but the process and deposit of a similar zone is seen in square 20R21, where the upper humic zone probably accumulated after the disappearance of the Green residence in the early 20th century.

When the Tillett site was first occupied around A.D. 400, sealevel was approximately 1.3 meters lower than today, and it is obvious from midden eroding at the south end of the site that the channel of Great Gut was farther south than at present. The estimated rise of 1.3 meters is based on Benton's (1980:111-115) Broad Creek data, using an average rate of 11 centimeters per century. The average water level in Great Gut, determined by daily readings, was .8 meter, a level correlating with the top of the Zone III sand (Figure 5a). Projecting backward in time, the land surface at the southern end of the site would have been approximately 2.1 meters above water level, and at the presumed termination of site use (ca. A.D. 1650), .7 meter above water level. Today, the land surface is about .4 meter above Great Gut but wind tides create a fluctuation in the water level between about .6 and 1.0 meter. When the water level is at maximum, it literally flows from Great Gut through the shell midden of Zone 11, as illustrated in square -56R2 (Figure 6a), nearest the stream.

Cultural stratigraphy and separation of components at the Tillett site was determined primarily through the seriation of ceramic types, as the best available indicators of change. The seriation was facilitated by excavation of the site in arbitrary ten-centimeter levels which were later correlated by the zones of stratification.

Two prehistoric components were identified. The first occurred between ca. A.D. 400-800 and belonged to the Mount Pleasant phase of the Middle Woodland period. Pure deposits containing this component in undisturbed context were Zone IIIA, the humic stained surface of the base sand, in square -48 and Test Square A, and the lowest level of the Zone II shell midden in squares 4L4, -42L2 and -42L6. From the position of these deposits, it was obvious that the Mount Pleasant occupation began on top of the sand and was responsible for initial accumulation of the shell midden.

The second component, Colington, continued the Zone II deposition of the shell midden from ca. A.D. 800 to perhaps A.D. 1650, although the latter date is an educated guess. Activities associated with the Colington component certainly disturbed and mechanically mixed the upper layers of the Mount Pleasant midden with its own, but Colington ceramics were predominant in the upper 20 centimeters of the midden. (Page 22) The Historic component, or Green residence, materials were associated only with the site surface and post holes intrusive into the earlier levels. (Page 23)