This group includes the pottery that is usually called Stallings, Thoms Creek and Refuge. Hamps Landing ware, found in lower North Carolina and occasionally in Horry County (Steen and Legg 1992) is included here, though Joe Herbert cautions that the dates associated with it may not be reliable (2009: Personal Communication).
Allendale ( Stoltman 1974). This is a relatively thick bodied ware (avg. about 10mm) that is decorated with numerous random punctations that appear to have been made with a bundle of sticks (image at link). It is tempered with very coarse to granular sand. Stoltman suggests that it is contemporary with Wilmington based on stratigraphy, but David Anderson (1996) considered it a Refuge variant.
Awendaw (Waddell 1974; Trinkley 1980, Cable 1998) For some Awendaw is a distinct type, but others see it as a phase of the Thoms Creek ware group (see Anderson et al 1996). It is a fine to medium sand tempered ware notable for finger pinched and finger grooved surfaces (images). This trait is more common in the central coast. Awendaw is usually associated with shell ring sites (Trinkley 1980; Saunders and Russo 2002).
Hamps Landing (link Hargrove 1993; Herbert 2003); South 1976). Stanley South collected a large number of sherds he said were “hole tempered.” These were later found to be two different types far separated in age. Tom Hargrove recovered examples that were found stratigraphically between Thoms Creek and Hanover, suggesting an early date, and found enough differences to sort them from the later Oak Island type. Joe Herbert re-examined them in his work, and describes eight types including cord marked, net impressed, fabric impressed, paddle edge stamped, simple stamped, bundled stick punctate, smoothed over stamp, and scraped. He feels that this diversity in surface treatments may mean that the early carbon dates in the 2003-2377BC (Calib 5.1 median) range obtained for the ware may be erroneous. Both Herbert and Jason Moser (et al 2009) obtained TL dates that were more recent.
Refuge (see this link; Waring 1966 (in Williams ed. 1977); DePratter 1979, 1991; Lepionka 1983; Anderson et al 1982, 1996; Cable and Cantley 1998; Espenshade and Brockington 1989, Crook 2008. To me the Refuge series epitomizes many of the problems with pottery research in the Southeast. It was first found at a site in Jasper county and defined by Antonio Waring and Joseph Caldwell (1939). There it was tempered with variable amounts of sand and finished primarily with simple stamping. It differed from Deptford simple stamped in that the stamping was more sloppy and often overstamped. Waring emphasizes the overstamping and generally “disorderly” (Williams ed. 1977: 200) application of all surface treatments in what he called Refuge. Surface treatments also include plain, dentate stamped, punctated, incised and brushed variants. Sherds are said to be thicker and the paste coarser than Deptford.
At the Refuge site the ceramics defined as Refuge and Deptford overlap, stratigraphically, and most observers agree, morphologically, but the levels with the majority of both types are well separated vertically (Williams, ed 1977). Recently Ray Crook (2009) has published a report on two Refuge sites that produced dates between 1194 and 864BC. These are the Bilbo site, on the Georgia side, and the Delta site on the South Carolina side of the Savannah River.
In Georgia they do not use the Thoms Creek type name. This leads to what has been called “the state line effect” (Anderson 1996 et al). There Refuge is seen as the bridge type between Stallings and Deptford. In South Carolina this position is mostly ascribed to Thoms Creek wares. Incised and punctated wares in Georgia appear to be what we would call Thoms Creek in South Carolina, though simple stamping is not common on Thoms Creek.
As with the Oemler wares researchers have seized upon the one thing that appears to be unique about Refuge pottery- the dentate stamped decoration- and used that to identify wares as Refuge wherever they are found (Anderson et al 1982, 1996; Cable and Cantley 1998; Espenshade and Brockington 1989; Adams 2005, etc.). In fact however, dentate stamping is a minority decorative type at both the first and second Refuge sites, making up .8% at the second, and about 4% at the first. At the Delta site only eight of 1705 sherds were dentate stamped (Crook 2009: 37). Dentate stamping is rare in Georgia. It is not found on St. Catherines Island, for instance (Thomas and Larsen 2008: 117). So dentate stamping is not a strong characteristic of Refuge pottery at the type site or in the immediate area and thus it is not valid to call a sherd "Refuge" based on this single characteristic.
At Fort Johnson multivariate analysis of the ceramic collection found dentate stamping and bundled stick punctate on the same paste as finger pinched Awendaw phase Thoms Creek wares. This paste identification was made using microscopic analysis, and is believed to be accurate. Excavations at the Awendaw Phase Lighthouse Point shell ring, a mile or so away, produced at least one carbon date for Awendaw, 1083BC (Calib 5.1), that is congruent with Crook's dates. At Minim Island what was identified as Refuge was in the same context as Thoms Creek wares, and received dates of 804 and 1212BC (Calib 5.1 applied). So in a sense the Refuge type is, as Waring thought, a step between Stallings and Deptford, but might more effectively be thought of as a terminal Thoms Creek phase than as a distinct type. Alternatively a distinct type may exist at the mouth of the Savannah, and another to the north on the Santee.
Stallings ( Sassaman 1993; Sassaman et al 2006; Stoltman 1972; Elliott et al 1994; Anderson et al 1996) Stallings ware is fiber tempered, leaving voids in the body and tracks on the surfaces. Vessels tend to be bowls. These are usually open, but some carinated examples are seen. This pottery was in use on the coast around the mouth of the Savannah by about 4500BP, and is found in association with shell rings in some cases (Saunders et al 2004).
Ken Sassaman has devoted considerable effort to studying Stallings wares and his book and other work should be consulted for details. He identifies three phases. Stallings I (4500-3800BP), or Early Stallings sees the introduction of plain wares, many with thickened rims, while the later “Classic Stallings”, or Stallings II phase (3800-3400BP) features hand decorated wares. Stallings III (3400-3000BP) sees the abandonment of rim thickening, and more diverse assemblages, with some dominated by plain wares and others by decorated wares. The Stallings Island site was abandoned by about 3500BP (Sassaman et al 2006).
Sassaman avoided calibrating his dates because the correction methods were changing on a regular basis (Personal Communication 2009). He kindly shared his data and allowed me to add his dates to the project database. Calibrated dates range from 4616-3451BP for the most part, with outliers on the early end at as early as 5058BP and as late as 3101 (Calib 5.1, 1 sigma median). These average out to about 3978BP. The "average" Thoms Creek median is a little later, at 3898BP. In the final database there are several examples of dates that are apparently mistakenly attributed to Thoms Creek wares that are as late as the 14th century AD.
Fiber tempered wares are most common south of the Santee, though examples are seen well into North Carolina (see distribution maps). In excavations at 38DA75 on the Pee Dee plain Stallings and plain and decoarted Thoms Creek wares are generally found well below later cord and fabric marked wares. These are often in context with asymmetrically resharpened Savannah River (variant) type points and small "Woodland Stemmed" points.
Saunders et al (2002) see an overlap between Stallings and Thoms Creek at the Fig Island shell ring both temporally and typologically. Local clay sources are thought to affect the amount of sand in the paste, as a replication experiment with Fig Island clay shows, is entirely possible. Thoms Creek has long been known to overlap with Stallings both in age and distribution, however (distribution maps). In a 2001 paper this is tied to the idea of situational learning and the development of communities of practice (Sassaman and Rudolfi 2001). The people making Stallings and Thoms Creek may have been distinct social groups. The common intermingling of the two types suggests, at the least, a close relationship between the groups, with potters moving from group to group through marriage or other means and thus creating this community of practice that specialized in plain and hand decorated wares.
Thoms Creek (Griffin 1945; Anderson et al 1996; Southerlin et al 2002; Cable and Cantley 1998; Espenshade and Brockington 1989); Saunders ed. 2002 ; Trinkley 1980; and many others). The Thoms Creek series was named for a site near Columbia (38LX2). James B. Griffin examined a small sample from a site where, ironically, Thoms Creek is a minority type. It is best known from its coastal appearances, but is in fact found throughout the Coastal Plain (see distribution maps). It decreases to the north, with a few sites in North Carolina producing it as far north as the Neuse River. At the Johannes Kolb site on the Pee Dee it is not extremely common, but it is found regularly, and almost always in context with Stallings plain wares and Savannah River type points that are asymmetrically resharpened. It is usually found at or below the base of the topsoil.
Temper varies. In general it is medium to coarse. On the coast it is mainly fine sand, but inland coarse to very coarse sand is sometimes seen. Temper density varies as well, with some sherds appearing temperless, and others coarse and sandy to the touch.
At the Fig Island shell ring complex Rebecca Saunders notes that in some cases sand that she thought was naturally occurring made up over 30% of the paste. This was the case with both Thoms Creek and fiber tempered ware - so much so that they went through a reanalysis to sort the two. A clay sample was taken at Fig Island by Sean Taylor, and vessels were made with unprocessed clay. Microphotos show that fine to medium sand does indeed make up a large percentage of the paste. Not only were there natural sand inclusions, but there was also organic material- probably marsh grass, and red inclusions that may be hematite, but which could be confused with grog. Examples from the type site, Minim Island and the Kolb site were also photographed for comparison.
The distinctive decorations are probably a better diagnostic than a given paste recipe, though context must always be considered, as later wares used some of the same decorations. Surfaces are often plain and well smoothed, with decorations including separate and drag-and-jab punctate, incising, shell punctate, finger pinching and grooving (fingertip sized), and simple stamping. Interiors are usually well smoothed to burnished. Vessels appear to be bowls primarily.
Anderson assigns a date range of 3000-4000BP for the ware group. Calibrated dates vary widely, and many are probably invalid. In our database dates purportedly for Thoms Creek range from 2533 to 5952BP. Four dates ranging between 832 and 1323AD are obviously incorrectly associated, and are not considered here. The "average" median, excluding the two youngest and two oldest is 3898BP. At Fig Island carbon dates ranged from about 3700 to 3900BP.