In terms of ceramics, the North Carolina coastal plain has traditionally been divided into three regions, with the central coastal region sharing traits of the both southern and northern regions (Table 2). Loftfield (1975) and South (1976) describe the Woodland ceramic sequence for the southern coast. Haag (1958) and Phelps (1983) outline the sequence for the northern coastal plain. Loftfield (1987a) later refers generally to Carteret, Onslow, and Pender counties as the central coastal area. The Cape Island site is in this central coastal region, an area of interesting overlap of the northern and southern sequences.

Table 2. Woodland Ceramic Series on North Carolina Coast.

This geographic separation has caused some confusion regarding the ceramic sequences. However, the geographic separation of north, central, and south regions is real in terms of ecological and geophysical attributes, as well as in terms of cultural affiliation, especially in the Late Woodland. Speaking ecologically and geophysically, a northern configuration extends from the Cape Hatteras vicinity northward, while a southern configuration begins somewhere in the area of the Cape Fear River. The barrier islands in the northern region (the Outer Banks) extend well offshore, enclosing large, shallow sounds and bays. In the southern region, the sounds become increasingly narrow (Adey and Burke 1976; Garrett 1983). Additionally, the northernmost and southernmost manifestations of many flora and fauna (both terrestrial and aquatic) are found between Cape Hatteras and Cape Fear (Eckman 1953). This location forms the central region of the North Carolina coast. The central North Carolina coast also appears to be the southernmost extent of the Algonquians during the Late Woodland. This natural and cultural boundary area offers much information about the prehistoric adaptations along the North Carolina coast.

Herbert and Mathis (1996:141) have recently argued that there is little basis for maintaining a distinction between the southern and central areas of the coast. However, we feel that any studies of the ceramic sequence for the North Carolina coast should keep these natural and cultural barriers in mind.

EARLY WOODLAND PERIOD

The onset of the Early Woodland period along the North Carolina coast is debatable. The appearance of Stallings Island fiber-tempered ceramics and its succeeding Thom's Creek wares around 25001000 b.c. generally form a satisfactory marker for the Early Woodland period. Certainly by 1000 b.c., coarse sand-tempered ceramics called Deep Creek or New River (discussed below) were associated with Early Woodland peoples.

The Stallings wares are typically decorated with punctations, incising, and finger pinching (Trinkley 1989) and are often associated with the Late Archaic Savannah River projectile points. Thom's Creek sherds are tempered with sand and include punctations, incising, finger pinching, and simple stamped surface treatments (Trinkley 1989).

Trinkley (1989) describes three site types for Thom's Creek peoples: large, amorphous shell middens, small, low-density sites, and the large shell rings. Although Thom's Creek ceramics are often recovered from North Carolina coastal sites (e.g., Hargrove 1993; Phelps 1983), little is known about the makers of Thom's Creek in this area. South (1976) did record a 9.1-m-diameter structure associated with Thom's Creek pottery at 31BW1 in Brunswick County.

Northern Coastal Plain. At about 1000 b.c., a course sand-tempered ware decorated with fabric impressions, net impressions, simple stamping, or cord marking appears along the coast. In the northern coastal region, this ware is called Deep Creek (Phelps 1983). Phelps (1983) defines three phases for the Early Woodland in the northern coastal plain. Deep Creek I appears at about 1000 b.c. The predominant ceramic type is Deep Creek Cord Marked, with Net Impressed, Fabric Impressed, and Simple Stamped types in minor amounts. Deep Creek II begins about 800 b.c. and is marked by an increase in net impressed, fabric impressed, and simple stamped surface treatments. In Deep Creek III, simple stamping decreases (and disappears by the Middle Woodland), while cord marked, fabric impressed and net impressed retain equal frequency.

Southern Coastal Plain. New River is the type name given to a similar ware in the central and southern regions of the coast (Loftfield 1975). Loftfield (1975) reported cord marked, fabric impressed, plain, thong marked (or simple stamped), and net impressed types. Phelps (1983) subsumes New River under the Deep Creek series. However, Herbert and Mathis (1996) argue that even though the paste characteristics and surface treatments appear comparable, differences in the frequency of surface treatments do exist. They elect to retain the use of New River until further inquiry clears up this ambiguity. New River is used in this report to refer to Early Woodland sand-tempered ceramics.

Hargrove (1993) recovered ceramics that appear to be tempered with crushed marl or limestone. This ware, termed Hamp's Landing, is similar to the Late Woodland shell-tempered ceramics in that voids resulting from the leaching of the temper are present, but the voids are usually larger (14 mm) and amorphous, as opposed to the nearly stratified appearance of the voids seen in the shell-tempered wares.

The Hamp's Landing ware appears to be either late Early Woodland or Middle Woodland, based on stratigraphic data from the Hamp's Landing type site (Hargrove 1993). So far, net impressed, cord marked, simple stamped, and a small percentage of fabric impressed surface treatments have been recognized on Hamp's Landing wares.

Little is known about the adaptive strategies of the Early Woodland inhabitants along the coast. However, site location data suggest a dispersed, high-mobility system similar to that of the Late Archaic. No direct evidence has been found that maize or other crops had begun to play a role in the diet of Early Woodland inhabitants (Phelps 1983). The diverse marine resources available (Yesner 1980) precluded a need for the development of horticulture during the Early Woodland period. Thus these people probably maintained an Archaic lifestyle based on fishing, hunting, and gathering.

MIDDLE WOODLAND PERIOD

Sites indisputably coastal in adaptive nature do not appear on the North Carolina coast until the Middle Woodland (Loftfield and Littleton 1981). Minor amounts of Early Woodland pottery are found on shell midden sites, but large, single-component, Early Woodland, shell middens are not known.

Loftfield (1987a) has suggested that by the Middle Woodland period, a pattern of adaptation to the coastal area was evolving based upon exploitation of estuarine resources, with sites becoming rarer in inland areas. A slightly earlier tendency to move sites from knoll tops to bottomlands in the Late Archaic and Early Woodland periods was abandoned in favor of the coastal locations (Loftfield and Littleton 1981).

Northern Coastal Plain. In the northern coastal plain, Mount Pleasant ceramics appear at approximately a.d. 200 (Phelps 1983). Surface treatments on the Mount Pleasant wares include fabric impressed, cord marked, net impressed, and smoothed. Phelps (1983) indicates that some incising has been observed on these wares. Herbert and Mathis (1996) note that Mount Pleasant has many similarities with Deep Creek and suggest that it may be derive directly from Deep Creek. The Middle Woodland probably ends around a.d. 800.

Ceramics very similar to the shell-tempered Mockley wares (which define the Middle Woodland along the Delmarva Peninsula) have been recovered from the northern coastal plain, and possibly from the central coastal plain (Herbert and Mathis 1996). However, shell-tempered pottery is everywhere along the northern and central coast by the Late Woodland. A definitive identification of these "Mockley like" ceramics has not yet been attempted.

In the northern coastal region, flexed and semi-flexed burials have been recovered from the Jordan's Landing site (31BR7), the Baum site (31CK9), and the Tillet site (31DR35). A primary cremation was also recovered from the Baum site (Phelps 1983). The northernmost sand burial mound documented in the state is near Vanceboro in Craven County (Phelps 1983), near the southern edge of the northern coastal plain.

Central and Southern Coastal Plain. The Hanover series (South 1976) is found in abundance along the central and southern coastal regions. Surface decorations include cord marked, fabric impressed, and plain. The series is distinguished by large grog tempering that often produces a lumpy interior vessel surface. Two dates for Hanover wares from the southern coastal plain are both ca. a.d. 300400 (Eastman 1994; Wilde-Ramsing 1984), but elsewhere Hanover has been dated to 100200 b.c. (Blanton et al. 1986:12).

Phelps reports Hanover cord marked and fabric impressed sherds in small quantities associated with Mount Pleasant wares in both the Tidewater and Inner Coastal Plain of the northern coastal region.

Cape Fear wares (South 1976) are also considered a Middle Woodland ceramic on the central and southern coastal plain. Tempered with coarse sand, the Cape Fear ceramics exhibit fabric impressed or cord marked surface treatments. Some confusion lies in distinguishing Cape Fear from the Early Woodland New River and Deep Creek wares. South (1976) initially described Cape Fear as ranging in sherd thickness from 0.4 cm to 1.2 cm. Herbert and Mathis (1996) have suggested revising the maximum thickness down to about 0.8 cm. Investigating further distinctions (e.g., cord marking techniques) in future research may also help to more accurately define this series (Herbert and Mathis 1996).

Loftfield (1975) identified a fabric impressed and cord marked Adams Creek ware. Adams Creek is tempered with very fine sand. Some "water smoothed" gravel was included in some sherds, but rarely exceeded 1 mm in size (Loftfield 1975:164). When first recognized, it appeared that Adams Creek occurred at the later end of the Late Woodland. Following investigations at Permuda Island, Loftfield and Watson (1985) indicated that Adams Creek is probably a Middle Woodland ware that parallels the Mount Pleasant series to the north.

A prominent feature of the Middle Woodland is the low, sand burial mounds (or burials in natural sand ridges/knolls) in the southern coastal region. The cremations and platform pipes and their apparent location away from habitation sites argue for a "southern" influence (Phelps 1983). The McLean sand burial mound near Fayetteville (MacCord 1966), the McFayden Mound (31BW67) in Brunswick County (South 1966), the Buie and Parham mounds in Robeson County (Wetmore 1969, 1978), and the Cold Morning site in New Hanover County (Ward 1980) are all examples of this cultural practice in the southern coastal region. The McClean and Cold Morning mounds are considered Late Woodland by many, and may reflect a continuation of the practice.

Loftfield and Jones (1989) excavated a secondarily deposited cremation at 31PD144 in Pender County, in the central coastal plain. A radiocarbon date of 2135±49 b.p. (1 sigma, corrected), ca. 180 b.c., places it in the Middle Woodland period. The 31PD144 excavation was an emergency salvage; the burial had been badly disturbed during road grading. The disturbance precluded determination of this burial as a sand burial mound.

Middle Woodland components are found on most of the truly coastal sites in North Carolina. It appears that an adaptation to estuarine resources is fully in place by the Middle Woodland. It remains unknown whether the prehistoric inhabitants continued a seasonal movement or were permanently settled adjacent to the estuaries, but Middle Woodland sites of some type are plentiful in the interior. Middle Woodland sites are found inland, suggesting some movement.

No unequivocal evidence for horticulture or houses has been located. Mathis has located round to oval structures at Broad Reach, which he suspects may be Middle Woodland based on several features immediately adjacent to the structures and pottery associated with the structures (Mark Mathis, personal communication 1997).

Evidence from the McFayden Mound (Wilson 1982, 1986) in the southern coastal plain, combined with what now appears to be an absence of shell-tempered pottery in the south, suggests that the Algonquian southern expansion ended somewhere in the central coastal region and that Siouan traditions survived up to the contact period in the southern coastal plain.

LATE WOODLAND PERIOD

By the Late Woodland period, a completely estuarine adaptation appears to be in place for the entire North Carolina coast. The distribution of Late Woodland peoples along the North Carolina coast has generally been interpreted from linguistic studies and ethnohistoric accounts (e.g., Hariot 1991; Speck 1924; Mook 1944; Paschal 1953; Lawson 1967). Phelps (1983) outlines three phases for the Late Woodland. The Colington phase, associated with Algonquian speakers along the northern coast, the Cashie phase, associated with Iroquoian speakers of the northern inner coastal plain, and the Oak Island phase, associated with the Siouan language family, in the southern coastal region. Recent archaeological evidence indicates that the Algonquian speakers occupied areas at least as far south as Onslow County in the central coastal region (Bogdan and Weaver 1989; Jones 1990; Loftfield 1990; Loftfield and Jones 1995; Mathis 1995). Loftfield (1990) and Mathis (1995) have suggested that the range of the Algonquian groups extended as far south as the Cape Fear River.

Northern Coastal Plain The Colington phase ceramic series is a shell-tempered ware found in the northern coastal region (Phelps 1983). Surface treatments include fabric impressed, simple stamped, plain, and incised. Rims are often treated with incised linear and geometric patterns. Occasionally, punctated rims are recovered. Vessel shapes include conoidal pots, simple bowls, and small beakers with everted rims.

The Cashie wares, believed to be contemporaneous with Colington, are tempered with small pebbles that often protrude through both the inner and outer walls. Sand tempering is often used in thin-walled vessels (Phelps 1983:43). Surface treatments include fabric impressed, simple stamped, incised, and plain types. Vessel rims are decorated with punctations and incisions and infrequently with finger pinching. Vessel shapes include conoidal pots and simple bowls. Beakers, ladles, and dippers with long handles have also been reported. Cashie ceramics are found on sites in the inner coastal plain.

Central Coastal Plain. Loftfield (1975) reports a shell-tempered ware in the central coastal region of North Carolina. Referred to as White Oak, this ware exhibits fabric impressed, plain, burnished, cord marked, simple stamped, and net impressed surface finishes. The overwhelming majority of the White Oak ceramics (approximately 90 percent) are fabric impressed, followed by plain and burnished surfaces. Cord marked, simple stamped, and net impressed surfaces are rarely recovered.

Southern Coastal Plain. The Oak Island series was defined by South (1976) as a shell-tempered tradition. Phelps (1983) proposed that this ware represents the Siouan speakers of the southern coastal region. Herbert and Mathis (1996) report that recent reanalysis of Oak Island wares suggests that some may actually be the limestone-tempered, Middle Woodland Hamp's Landing ceramics. If this is the case, shell-tempered ceramics could be less prevalent than previously posited on the southern part of the coast. This in turn may support the idea that the southward expansion of the Algonquians ended somewhere in the central coastal region.

An interesting ceramic has been recovered from several Late Woodland sites in the central coastal region. This is a fabric impressed ware, tempered with both shell and grog. A partial vessel of this ware was recovered from a refuse pit at the Broad Reach site (Herbert and Mathis 1996). A radiocarbon date places the pit at a.d. 1415 (calibrated). Several of these sherds were recently recovered from the Hammocks Beach site (31ON82). No date range has been established for these sherds (Mark Mathis, personal communication 1996). A formal description has not been brought forward for this ware.

By the Late Woodland period, the prehistoric inhabitants appear to have adopted a total adaptation to the estuarine environment. All or part of structure patterns, suggesting permanent habitations, have been recovered at several Late Woodland sites (Loftfield and Jones 1995): 31ON33 (Loftfield 1979), 31ON196 (Loftfield 1985), 31CR218 (Mathis 1993), 31CR53 (Reid and Simpson 1994), and 31HY43 (Gardner 1990.

These structures include the "hallmark" longhouses of the Algonquian culture, as well as more rectangular patterns. Mathis uncovered at least one round to oval structure at Broad Reach. He is "tentatively assuming" that this is a Middle Woodland manifestation. Pits immediately adjacent to the structure have Middle Woodland dates, and Hanover is the predominant ceramic in that part of the site. Loftfield (personal communication 1997) had some possible circular patterns at Permuda Island, but additional work discerned no definite structures.

Evidence of horticulture has also been recovered from several Late Woodland sites along the coast: 31ON33 (Loftfield 1979), 31ON196 (Loftfield and Watson 1985), 31CR218 (Mathis 1995), 31CR53, and 31HY43 (Gardner 1990).

Burial of the deceased in ossuaries is one the most extraordinary characteristics of the Late Woodland peoples along the North Carolina coast. Some 10 or more ossuaries have been excavated in the north and central coastal regions. Phelps (1983) reports that five ossuaries from sites in Chowan County, Currituck County, and Carteret County and on Hatteras Island contained 3858 individuals. The Flynt site (31ON305) ossuary contained the remains of at least 150 individuals (Bogdan and Weaver 1989). Mathis (1994b) reported two ossuaries at the Broad Reach site (31CR218); Loftfield (1985) uncovered one ossuary at the Permuda Island site (31ON196).

Until the excavations at Broad Reach, it appeared that cultural affiliation could be readily determined based on the ossuary type at a given site. Ossuaries associated with the Iroquoian speakers on the inner coastal plain in the north generally contain two to five individuals interred as secondary bundle burials (Phelps 1983), assumed to represent family units. The Iroquoian ossuaries usually contain marginella beads as grave goods and are inside villages.

The Algonquian ossuaries in the north and central parts of the outer coastal plain, and associated with shell-tempered ceramics, generally represent community burials. They contain large numbers of individuals, many of whom are represented by isolated skeletal elements. McCall (1987) noted Felis phalanges in two Algonquian ossuaries in the north region. Generally, however, no unequivocal grave goods are present. Algonquian ossuaries have been assumed to be a short distance from the villages.

In the southern part of the coast, ossuaries associated Siouan peoples are placed in mounds or atop high sand ridges. This practice began in the Middle Woodland period (see above) and apparently continued into the Late Woodland. The individuals in these mounds represent secondary burials placed in bundles. Cremations, or least charred skeletal elements, are often present. In some cases, grave goods are present. These burials appear to be placed at great distances from village sites. However, this contention needs to be tested with intensive survey designed to locate village sites.

The excavations at the Broad Reach site (31CR218) have raised interesting questions regarding burial patterns along the North Carolina coast. Mathis (1994b; personal communication 1996) reports two ossuaries, a primary interment containing two individuals, two primary flexed burials, six secondary bundle burials, and four burials containing only fragmentary elements from the site.

One of the ossuaries contained nine individuals, who were all complete and semiarticulated. There were one or two additional femora, suggesting individuals represented by isolated skeletal elements only. The skeletal material was arranged in discrete bundles, and several grave goods were observed. The grave goods include two shell-tempered pots, a small ground stone cup, a turtle carapace, and clusters of possible marginella shell bead bracelets. One of the bundles appeared to have been laid on mat constructed of over 800 marginella beads. Additionally, what has been tentatively identified as a bundled juvenile dog was present.

This ossuary displayed characteristics of all of the documented ossuary patterns on the North Carolina coast, which to that point had fallen neatly into culturally identifiable categories. The discrete bundles suggest Iroquoian or Siouan affiliation. The grave goods are typical of the Iroquoian pattern. The shell-tempered vessel, as well as a 14C date of a.d. 10321247 (a.d. 1168 calibrated) indicates a Late Woodland interment. By that time, the Algonquians appear to have established a hegemony along the coast, at least in the north and central regions. Interestingly, the ossuary at the Flynt site had all the traits of an unquestioned Algonquian affiliation: at least 150 individuals, no discrete bundles, and no grave goods. The Flynt site is less than 25 miles south of Broad Reach and yielded a 14C date of a.d. 12901402.

The second ossuary from Broad Reach contained no apparent articulated elements and portions of only four to six individuals. At least seven copper beads had been scattered among the skeletal remains in this ossuary. Additionally, a cremation was present. A cremation was included in an ossuary on Knotts Island in the north coastal region (Mathis n.d.). Knotts Island and Broad Reach are the only ossuaries with cremations documented on the North Carolina coast. Again, the second Broad Reach ossuary displays traits generally assumed to represent Iroquoian, Siouan, and Algonquian burial practices. Of the remaining burials at Broad Reach, all but one were Late Woodland. The exception, Burial 8, was a Middle Woodland burial with two adult females. Two complete Hanover pots were included in this burial. At this point, there appear to be no "typical" ossuaries on the North Carolina coast. In light of the myriad burial types unearthed at Broad Reach, there appears to be no typical burial pattern of any kind on the coast during the Late Woodland.

The "typical" Algonquian longhouse also seems to be in question. Mathis (1993) reports a longhouse-shaped structure at Broad Reach. However, this structure measured more than 17 m in length, and Mathis (1993) indicates that the entire length of this structure was never uncovered. Additionally, although Mathis (1993; personal communication 1997) assumes the round to oval structure from Broad Reach is Middle Woodland, no radiocarbon date is available.

There appears to be much interaction, with resultant cultural exchange, between groups along the central North Carolina coast in the Late Woodland. Alternatively, we could be seeing the result of different groups competing for, and intermittently gaining control of, this area during the Late Woodland. Future research in this temporal and geographic area should attack this question.