It is obvious from his description of the Kitty Hawk Bay site that Haag incorporated two completely distinct and separate sites (31DR14 and 33) into his site number C3. His statement that ... "For a distance of at least one-half mile along the northward facing shore of Big Colington, there are hills about fifteen or more feet in height that are capped with oyster shell midden." (Haag 1958:40) describes almost exactly the distance between the Kitty Hawk Bay and White Court sites. There is, however, no continuous distribution of oyster shell or cultural material for the distance quoted above, either along the dunes or on the beach. Instead, the separate sites are located where the higher dune elevations are, or were, in close proximity to the immediate shoreline (Figure 3).

The Kitty Hawk Bay site, as defined in the 1972 survey extends for approximately 240 meters along the crest and shoreward flank of a dune ridge of Fripp fine sand on the north shore of Big Colington Island (Figure 4a). The dune rises rather steeply from the present shore to a general elevation of 7 meters (20 feet), and attains a maximum height of about 9.1 meters (30 feet) toward its western end. Construction of Kitty Hawk Bay Drive (Figure 2) has modified the southern edge of the dune, and a parking and beach access area (Figure 4b) was cut through the middle section of the ridge. One cottage had been built just east of the site prior to the 1972 work and another has been constructed (Figure 4a) since then; the latter cottage occupies the space where Squares A and F were excavated. Vegetation on the site is a mixed pine and hardwood forest with a relatively dense understory of dogwood, shrubs and other low-growing plants.

Surface inspection of the site area revealed oyster shell spilling down the sides of the parking and access area cut through the dune ridge, and shell midden debris along the shoulders of Kitty (Page 10) Hawk Bay Drive east of the cut. Oyster shell was particularly thick adjacent to the driveway of the cottage on the eastern side of the side. Shell was also observed on the surface of the ridge in a few places west of the access cut. Very little shell was observed along the narrow beach, but this is not unusual in eroded coastal sites where the dynamics of storm wave and normal tide cycles of erosion and deposition constantly change the appearance of beach lines. Depending upon the events of nature, a visit to such a site one day may present only a sandy beach, but the next day may reveal a vast amount of shell and cultural debris.

Eighteen ceramic sherds (Table 2) were the only cultural materials collected from the site surface; these were reclaimed frm the floor and sides of the access cut. The 1955 survey listed 95 sherds (Haag 1958: Table 1), presumably collected from the beach, but Haag probably included both this site and the White Court site (31DR33) in a single collection. Since both sites have similar components and temporal span, separation of the earlier collection by site is not possible.

Surface evidence of the shell midden was used as a guide to location of the initial excavation units. Because of the time factor and the complexity of surface topography, all squares were laid out in relation to a staked temporary reference point set on the top of the dune ridge east of Square B; from this reference, other points were staked for convenient transit access to various sections of the site for purposes of topographic mapping and vertical control of the excavations. Placement of the excavated units, squares A through H, is shown in Figure 3. All were excavated in arbitrary 10 centimeter levels to control stratification and cultural content, but only the culture-bearing stratum in each unit eventually received permanent level numbers. Upon completion of each unit, the levels were then correlated with the zones of stratification recorded on the profile sheets. Soil from the excavations was sifted through a standard one-quarter inch mesh screen except in some situations where selected samples of each level were wet-screened through a window screen mesh. The content of all pits was similarly wet-screened for maximum recovery of small items and charred organic material. All cultural material and food remains were saved except the masses of oyster shell; in the latter case, separate samples were collected from each level for radiocarbon analysis and to provide a size range of the shells. Representative samples of other shellfish species were also retained.

Squares A and B were placed on either side of the access cut to sample the midden zone observable in the faces of the bank. Square A (2 x 2 meters) revealed a top zone of culturally sterile gray humic sand ranging from 2 centimeters in depth on the north (Page 12) side to 15 centimeters on the south. Below this an approximately 20 centimeter deep deposit of shell and black organic soil was the primary zone of cultural occupation. A 10 to 15 centimeter zone of gray mottled sand below this was the land surface immediately prior to occupation and contained some cultural material in the upper two centimeters. A zone of culturally sterile yellow sand underlay the mottled stratum and extended for an undetermined depth below 70 centimeters where the excavation halted. Square B (2 x 2 meters) on the eastern side of the cut (Figure 5a) had a top zone of gray humic sand, the modern land surface, of approximately 10 centimeters thickness. Below this was a deep zone of white, leached sand, typical of Fripp dune soil, extending to a depth of 1.1 meters on the north profile and 1.9 meters on the south. This sterile deposit represents dune building processes following the occupation of the site, and was underlain by the dark organic soil and shell of the midden zone, ranging in thickness from 10 to 20 centimeters. A typical zone of mottled organic soil underlay the midden, and capped the iron-stained sterile sand below. The excavation extended to a maximum depth of 2.1 meters below surface, necessitating the excavation of a 1 x 1 meter square through the sterile sand to a point just above the midden on the north side of Square B to provide access. The surface of Square B slopes downward toward the shore from the crest of the modern dune some 10 meters south of the unit; the midden zone in the square sloped upward toward the shoreline, indicating the crest of the dune at the time of occupation was further north. A one-meter square, D was opened north of B to confirm this configuration; in its south profile the top of the midden zone lay 30 centimeters below surface and sloped upward to a depth of 15 centimeters on the north. The midden had an average depth of 20 centimeters in this square. The overlying modern dune has formed south of the older dune on which the shell midden was deposited, and the crest of the midden lies nearly at the 10 feet contour over the beach. Obviously, erosion has removed a large amount of the midden in this section of the site.

Stratigraphy in Squares F, G and H (all 2 x 2 meters) was similar to that in A, differing only in the depth of modern dune accumulation over the midden, and in the thickness of the midden itself. These squares also produced evidence of midden lenses, the shell and dark organic soil thinning out and abruptly ending, suggestive of not a comprehensive midden coverage over the entire site but of discrete midden deposits with intervening non-deposition areas. The north profile of Square G is shown in Figure 5b, with the dark stain of a decayed pine tree tap root in the northeast corner. This and other observed tree root stains indicate that the dune was stabilized with vegetation similar to the modern forest at the time of initial occupation.

Square C was a 1 x 1 meter excavation placed beside the driveway of the cottage on the eastern side of the site (shown (Page 13) in the left side of Figure 4a) where a dense midden deposit was observed. The deposit proved to be secondary, the result of road and driveway construction, as were the other deposits along Kitty Hawk Bay Drive.

(a) (b) Figure 4. The Kitty Hawk Bay site, 31DR14. Aerial view (south) of the site showing cottages and dune ridge vegetation (a). The beach access cut through the dune ridge in approximate center of the site (b). (Page 14)

(a) (b) Figure 5. Stratigraphy at Kitty Hawk Bay. West profile of Square B, showing accumulation of modern sand dune above the midden (a). North profile of Square G showing lens-shaped midden deposit and a tree root intrusion in the northeast corner (b). (Page 15)

All of this secondary deposition probably resulted from excavation of the access cut through the dune, the material from which was removed and spread along the roadbed during construction. There was no evidence that midden deposits occurred on the southern flank of the dune ridge. Square E, 2 x 2 meters, opened on the dune slope between Square B and the cottage, produced no evidence of cultural material or occupation zones.

The stratification sequence observed in the excavations consists of five zones:

Zone I. A gray humic sand stratum, ranging between 2 and 15 centimeters in depth, is the modern surface layer of decaying organic material mixed with the leached sand of the modern dune surface. This zone was absent in a few places on the western side of the site, probably due to modern wind erosion.

Zone II. This stratum of leached white sand is the primary wind-deposited soil of the modern dune, classified as Fripp fine sand (U. S. Department of Agriculture 1977: 44-45). Like Zone 1, this culturally sterile stratum was absent or very shallow in some areas; it was generally deeper along the highest elevations of the ridge, and attained a maximum depth of 1.9 meters in Square B. The depositional pattern indicates slow southern or south westward movement of the dune, retarded to some extent where forest cover exists.

Zone III. The oyster shell and dark gray-to-black organic midden soil of this stratum is the primary occupation layer of the site. It ranges in thickness from 10 to 25 centimeters, and had been deposited in lenses on the older land surface, rather than as a comprehensive accumulation. One radiocarbon date was obtained from oyster shell in the third level of this zone in Square B, where it was most deeply buried and protected from modern intrusion and contamination. The date, A.D. 1045 (B.P. 905 + 65; UGa-1090), applies to the earlier part of the Colington phase component. A second radiocarbon sample, obtained from oyster shell in the upper level of the midden in Square A, was dated to A.D. 1320 (B.P. 630 + 80; UGa-3847).

Zone IV. A gray, humic or organic stained sand, averaging 10 centimeters in thickness, underlay the (Page 16) midden and probably represents the land surface capping the older dune ridge at the time of the initial Colington phase occupation. A few artifacts occurred in the upper part of this zone. This older dune surface had a crest some distance north of the present dune, and has apparently undergone extensive erosion.

Zone V. The culturally sterile yellow (iron-stained) sand of this zone is the older dune surface prior to occupation, extending downward to an undetermined depth beneath the site.

Cultural stratigraphy of the Kitty Hawk Bay site is quite simplistic; except for a scatter of modern debris on the surface, the only culture-bearing stratum is the Zone III midden and the very top of the older land surface (Zone IV) below it. While a few specimens of Mt. Pleasant series ceramics, indicating an earlier use occupation and deposition of the shell midden lenses occurred during the Colington phase. Cultural material representative of this phase was found in all levels of Zones III and the top of IV.

From approximately 4.25 cubic meters of excavated midden zone, a total of 287 artifacts were reclaimed. This quantity is relatively low for the amount of midden accumulation and suggests, along with the depositional pattern of the midden, brief and intermittent use and specialized site function.

Classification and distribution of the artifacts is summarized in Table 2. Although Zone III and the top of Zone IV were excavated by levels, subsequent analysis of the artifacts showed no temporal or typological differentiation by level, and all are listed under Zone III in the table with the exception of Square C, a secondary deposition, and the site surface.

Ceramic sherds account for the majority (234) of the artifacts from the site, and are classifiable into the series and types below.

Colington series: Ceramics of this series were recovered from all levels of the occupation zone and are the majority ware with a total of 206 specimens. Three types are represented in the collection (Table 2); Fabric impressed (140), simple stamped (5), and plain (8). The remaining 53 sherds of this series were too small or had surfaces too eroded or poorly impressed to classify into types. Colington series ware is tempered with crushed shell, the particles ranging (Page 17) from minute to large (5 mm.) laminar flakes, and occasional lumps from the valve ends of shells (Figure 6a-c). Some Colington sherds also have liberal amounts of sand, occasional small pebbles, and lumps of clay included with the shell, probably as accidental inclusions. Surface color ranges from dark gray through tan and brown to light orange, the variations probably resulting from original firing and fire-clouding modified by subsequent use and extent of weathering or oxidation. Vessel wall thickness averages 6.5 mm. for the 31DR14 specimens, with a range from 5.0 to 9.5 mm. Surface finishing techniques of fabric impression (Figure 6a-d), and plain (smoothed surface) (Figure 6e-f) are present in the collection. The sample offered little information on vessel shape, but both straight (Figure 6c-d, f) and everted, or flared, rims (Figure 6a-b) are present. Both flat and rounded lips occur; frequently these were stamped with either the edge of fabric wrapped or simple stamp paddles. The interior rims of fabric impressed sherds often have edge impressions of the paddle applied for a short distance down from the lip. Most such impressions are perpendicular to the lip, but a few had been applied in a chevron design (Figure 6d).

Cashie series: Seven sherds of this series were reclaimed from squares G and H; six are fabric impressed and one is simple stamped (Table 2), representative of two of the four types in the series. The Cashie ware is typical of the inner Coastal Plain during the Late Woodland, and is presumed to be trade ware in many of the Colington sites. The specimens from 31DR14 have the characteristic small pebble tempering, contorted paste, and smoothed over, lumpy interior surface.

Mount Pleasant series: This Middle Woodland period series is also a minority ware at the site, where it was found mixed with the Colington phase deposits. Represented by only 19 sherds (Table 2) identified by their tempering material of sand and pebbles, the Mount Pleasant ware indicates some use of the site prior to the Colington occupation. Cord marked (12) and fabric impressed (5) surfaces are represented along with 2 residual specimens.

Probably contemporaneous, in part, with the Mount Pleasant series, and here included with them, are a few clay tempered sherds which may be more at home within the Hanover Series (South 1976: 16-18). A few of these sherds have only clay lumps, differing from the "sherd temper" of Hanover ware, as the tempering medium, but in two cases there are liberal amounts of coarse sand along with the clay. Cord marking, fabric impression and incision are present as surface finishing techniques on these specimens. Two fabric impressed sherds are shown in Figure 6j (clay temper only) and 6i (clay and sand), a clay tempered specimen with a crenelated lip and incised parallel lines below the rim is illustrated in Figure 6g, and a clay tempered cord marked sherd in Figure 6h.

Only two modified lithic artifacts were found in the excavations. One of these is a large, flat slab of calcareous sandstone (Figure 7d) with a maximum length of 18.6 cm., width of 17 cm. and thickness of 2.1 cm. Numerous long, concave grooves, quite smooth along their inner surfaces from repeated use in sharpening bone tools, occur on one face of the specimen. Calcareous sandstone artifacts of this type have been found in a number of sites on the Outer Banks and on the adjacent mainland, where they appear to be the standard abraders during the Colington phase. This sandstone is known from beach outcrops along the southern North Carolina coast (Moorefield 1978), and recent studies by the East Carolina University Department of Geology indicate that the same formation may underlie the recent sands in the Cape Hatteras locality (S. R. Riggs; personal communication). The other artifact is a fragment of a highly polished, green steatite gorget of the expanded center type (Figure 7e), measuring 2.7 cm. from the central break to the rounded end. Both the abrader and the gorget fragment were found in level 3 of the midden in Square B (Table 2).

The remaining lithic specimens are 48 unretouched spalls, sparsely scattered throughout the excavations (Table 2). Most of the flakes are of a size indicative of initial working of small pebble cores, but a few smaller ones (1-5 mm.) may be the result of final tool production or retouch. Only 3 spalls are green argillite (slate), a material much more common on the mainland and (Page 19) farther inland in the Coastal Plain; the other 45 are jasper, ranging in color from medium gray to red, tan and yellow. This is the typical lithic raw material on all the Outer Banks and immediately adjacent mainland sites, and is obtained in pebble and small cobble sizes from older river channels which lie underneath and generally perpendicular to the present islands. Beach erosion exposes the old channel beds and this lithic resource.

Figure 6. Ceramics from 31DR14. Colington fabric impressed rim sherds (a-c), and a rim interior with with fabric paddle impression in chevron design (d); Colington plain rims (e-f). Mount Pleasant series specimens with incised (g), cord marked (h) and fabric impressed (i-j) surfaces. The sherds in g-h, j are clay tempered, and i is typical Mount Pleasant sand and pebble temper with some inclusive clay lumps. (Specimens are half actual size) (Page 20)

Figure 7. Artifacts from 31DR14. Deer cranial fragment with antlers removed (a), bone awls (b-c), sandstone abrader (d) and a fragment of a polished steatite gorget (e). The abrader is shown one-half size; all others are actual size. (Page 21)

The distal fragments of two bone awls, both made from mammalian long bones (Figure 7b-c), and the tip fragment of an antler flaker were the only bone tools reclaimed. The awls have highly polished edges and tips, and appear to have been broken prior to discard. The flaker shows considerable wear and pitting at its point end.

The deer cranium in Figure 7a may have lost its antlers through annual shedding, or they may have been cut; the cancellous cores of the antler bases are too eroded to be certain on this point

The most obvious and numerous food remains are oyster shells, most of which appear to be Crassostrea virginica, which make up the bulk of the cultural deposit at the site. Scattered among the oysters in the midden deposits were the shells of a few other species, including quahog (Venus mercenaria), northern moon shell (Polinices heros), and tulip shell (Fasciolar species?), all of which are numerically insignificant and probably were not major food resources.

Fish were the second most popular food at the site, represented by a total minimum weight of 121 kilos of live specimens as determined by analysis of the bones (Appendix B). large salt water fish include red drum, black drum, sheepshead, jack crevalle, kingfish, striped burrfish, sharks and/or rays, and smaller species such as spot, croaker, and flounder were also taken. A few specimens of white catfish and long nose gar, more at home in fresh-to- brackish water, are indicative of the fluctuation of salinity in this part of the estuarine system. Osteichthyes and elasmobranchii constituted 96.14 percent of the useable meat weight of all faunal remains other than shell fish at 31DR14 (Table 9).

White-tailed deer, raccoon, and rabbit were identified among the mammalian fauna (Appendix A). Pig bones (Sus scrofa) of modern origin were surface finds, and domestic dogs, represented in the collection only by teeth, probably accompanied the site occupants and were not a food source. Exploitation of the mammalian fauna was minimal, as evidenced by the presence of only two individual deer, 2 raccoons, and one rabbit. (Page 22)

Avian remains included duck (unidentified as to species) and clapper rail (Appendix A) which accounted for 0.1 percent of the total usable meat from non-shellfish faunal remains at the site.

Reptilian fauna included musk turtle and snapping turtle from a fresh water environment, the brackish water diamond-back terrapin, the terrestial box turtle, and alligator.

The faunal assemblage from the site indicates a major adaptation to the aquatic and secondary utilization of the terrestial environment for nearly maximum exploitation of faunal resources in this locale.

Evidence from the excavations at Kitty Hawk Bay confirms two components at the site. The earlier component, identified from the few Mount Pleasant series ceramic specimens scattered among the remains of the later occupation, belongs to the Middle Woodland period (300 BC-800 AD). During that occupation, the main site area was probably further north and adjacent to the shoreline of that time. Shoreline erosion and southward movement of the dune ridge since then have resulted in loss of the earlier main occupation zone, and the few specimens in the present site represent scatter on the older land surface at the southern edge of the original site. The existing shell midden and site boundaries are the result of the Colington phase occupation during the Late Woodland period (800- 1650 AD). Approximately half of the site area of that component has also been lost to erosion, and modern dune building processes have resulted in covering most of the remaining site. Radiocarbon dates of A.D. 1045 + 65, obtained from shell on the lowest level of the midden and A.D. 1320 + 80, from the top of the midden, bracket the Colington phase occupation. The shell tempered Colington ceramic series, found throughout the midden deposits, is the identifying hall-mark of this component.

The Colington phase intrasite pattern at Kitty Hawk Bay is one of lens-shaped deposits of shell midden scattered over the older dune surface, separated by distances varying from a meter to an estimated 10 or more meters. Each midden deposit is a small refuse dump, containing food remains (shell and bones), charred wood, and a few artifacts. Very few artifacts were found outside the shell areas, and no evidence of structures was observed. This pattern suggests recurring temporary use of the site by small groups of people over a relatively long period of time, obviously going there for exploitation of the food resources in close proximity. Two dog teeth among the faunal remains (Appendix A) indicate the presence of family pets or hunting dogs with the family group. (Page 23)

The artifacts represent the minimum equipment, possibly for a family group, of a few cooking and storage vessels brought with them, bone tools for sewing or leather repair work and abraders on which the awls were sharpened, an antler flaker for producing the necessary lithic tools, and a few spalls resulting from the production process. Even though the 1972 excavations are only a sampling of the site area, the total site assemblage probably would not include many other classes and functional types of artifacts, or a higher density of artifacts per level or square.

The temporary camps at the site were certainly located there to collect oysters from nearby beds, probably directly off-shore from the site in Kitty Hawk Bay. This primary subsistence activity was supplemented by fishing as a secondary exploitation of the marine environment. The fish species represented in the faunal assemblage would have required a number of techniques including weirs, nets, and hook and line. Hunting of land mammals and collecting or hunting reptilian fauna and birds appears to have been incidental to shell fish collecting and fishing. All of these resources were available within the immediate environs of the site; oyster beds on shoals in the bay, fish from the bay and sound, deer, rabbit, raccoon and box turtle from the forested dunes and other upland areas, turtles and alligator from the fresh water pond area near the site, diamond-back terrapin from the salt marshes, and ducks and clapper rail from the ponds and off-shore environment. Whether or not all these food resources were immediately consumed or some preserved for later use is difficult to determine. Oysters do not preserve well and presumably were consumed immediately; the degree of fragmentation and scatter of the fish, mammal and other bones also argues for immediate preparation and consumption of these foods.

Maximum abundance of the various species in the faunal assemblage at 31DR14 implies primary occupation between April and September (Figure 16), but prevalence of the larger salt water fish in the fall (Appendix B) seems to suggest late summer-early autumn as the period of maximum site use. Certainly, water salinity in the vicinity of Colington Island was higher during the period of occupation of 31DR14 than it is today, and the size and number of larger salt water fish among the remains argues strongly for an inlet through the Outer Banks in relatively close proximity to Colington Island.

The site catchment area, that zone around the site from which resources were obtained and for which the site location was originally selected, can be defined as the area within a half-mile radius of the site with respect to the observed subsistence pattern. If lithic resources are included, presuming that these were not brought with the occupants, a three-mile trip to the eroded ocean beaches to the east is required. (Page 24)

This small site, first observed and collected in the 1972 project, is located 125 meters southwest of 31DR14 and immediately west of the intersection of Kitty Hawk Bay Drive and Colleton Lane (Figures 2-3), from which its name is derived. At this point, Kitty Hawk Bay Drive had been cut through a north-south trending dune ridge (Figure 8a), exposing an old humus surface at an elevation of approximately 30 feet (9 meters). South of the road, a few Colington sherds were collected where they had eroded from the humus zone, and a small, disturbed oyster shell midden was observed on the north side where it had spilled down the cut bank.

Five Colington fabric impressed and 2 Colington plain sherds (Table 1) constitute the entire collection from the site. The total site area is unknown but is suspected to extend to the dune crest overlooking the Bay, an area which already had cottages constructed on it and for which we had no permission to enter. The observed shell midden and the ceramics suggest general contemporaneity with the Colington phase components of 31DR14 and 33, as well as a similar intrasite pattern. The proximity of the site to 31DR14 also suggests that it may have been an outlying or peripheral occupation area of that site. (Page 25)