Textile Marked Wares

There is considerable overlap between this group and the paddle stamped wares in distribution, age, and paste characteristics, but the obvious surface treatment differences make this a logical criterion for sorting. The ware groups discussed below include types that have non-textile surface treatments, but the majority of the sherds assigned to these groups are textile marked.

The textile marking includes net, fabric, and cord. Major temper groups are grog and lithic. Grog tempered wares are discussed separately. Pastes range from fine and silty to coarse and sandy, with pebbles and crushed rock. The combinations of finish and paste variables is believed to be the key to validly sorting these wares, but the combinations can be bewildering, and because few sites are pure one component occupations, and all ceramics are usually found in the heavily bioturbated top 30-40cm of soil, they co-occur with only trends and tendencies standing out. This is especially true when the topsoil is not carefully excavated in levels and removed as a natural zone.

At the Kolb site Sean Taylor analyzed the sherds from a single 2x2m unit and identified eleven paste groups with 17 different variations of textile impressions- fine cord, heavy cord, cross marked, parallel marked, etc.. There were also three distinct Check Stamped paste groups, three Thoms Creek ware paste groups, and a fiber tempered group.

Textile Group 4, Check Stamp Group 3, and Thoms Creek Group 1 and 2, all share a paste that has a high density of fine to medium sand temper, and they are only found in the lowest level (of four) in the unit he analyzed, but Textile Group 11 - fine sand and grog tempered - is found in three levels - the top two and the lowest. In terms of surface treatments all levels were dominated by cord marked, Fabric impressed wares were present in every level too. We almost always find Thoms Creek and Stallings in the lowest pottery producing levels, but we also find textile marked wares in the same levels.

Textile Group 4 may be the equivalent of Coe's (1964) Badin type, which, because of its stratigraphic position he believed was early. He compared it to Thoms Creek in terms of paste. But if the only difference between Textile Group 3 and Group 4 is that it has a low density temper of fine to coarse sand instead of a high density of fine to medium sand, accurately and consistently sorting the two will be a problem. Microscopic examination on a sherd by sherd basis may be effective, but the labor involved may make this a question that few will consider worth the effort. And, if G4 is Badin, what is G3? Is there a valid difference at the cultural or individual potter level, or is it a random element such as the clay source that affects the sand density? Also, if G4 is Badin here, are all cord marked wares with that paste Badin? Do similar wares found on the Savannah date to the same time and represent the products of the same, or closely related groups, or is it a factor of the limited combination of ingredients that went into the paste and decoration recipe?

Textile marking has been called "the Northern tradition" since the 19th century (Holmes 1903), and indeed, it is much more dominant in the north than it is to the south and west. But this does not mean that there is no textile marking in the south and west- Dunlap, Cartersville (link) and others appear to be coeval with Deptford and they are found throughout Georgia. As seen at Minim Island, cord marked and fabric impressed sherds had pastes identical to Check Stamped Deptford wares, leading to the conclusion that they were in fact Deptford. But in North Carolina, where check stamping is less common, the same ware is called "Deep Creek" which Joe Herbert argues is the same thing as "New River" (Herbert 2003). Deptford cord marked has been identified in Beaufort County repeatedly (see Trinkley ed. 1989; Brooks 1982). Chester DePratter's "Chatham County Cord Marked" with its "compact paste containing an abundance of coarse grit" (DePratter 1991: 179) sounds a lot like "Deptford" and was found in context with Deptford and Wilmington.

When the differences between types are so subtle, and type definitions are so broad, it is difficult to state that a given cord marked, sand tempered sherd is one type or another with any confidence. With textile marked wares the consideration of context is essential. Though most sites that we find have been plowed, logged and bedded, or otherwise disturbed, we should pay close attention to stratigraphy and try to obtain empirical dates where possible.

The type names used for textile marked wares by researchers in the state are introduced below. As Joel Gunn (1995) put it, whether these are valid or not is not the question. These are the definitions people have applied, and it is our task to interpret them and assess them for use in our own work. Thus if we see that an out of state researcher has come in and for the first time ever identified wares as "Cartersville Check Stamped" we must assume they mean its a coarse sand (or "grit") tempered ware that is the local equivalent of Deptford.

 

Badin (Coe 1964, 1995 (1964 types at the link)) This ware has “a very fine sand temper with occasional pebble inclusions” similar to, but slightly post dating Thoms Creek. Empirical dates are lacking, but Coe believed it developed around 1000BC. Surfaces (images) tend to be fabric marked, cord marked, net impressed or “impressed with a mat-like “wicker” fabric” (Coe 1995: 154) though a plain variant was also noted (Coe 1964: 29). This type name is not widely used, but based on the description alone, it appears to be present at the Kolb Site (38DA75).

Berkeley (Cable 2002 in Cantley ed. 2002) The Berkeley series appears to be the same as Hanover I, having a sandy paste with grog tempering. This is seen as developing from Deptford, as check stamped surfaces are common (35.9% at 38SU136, 141). Cord marking (42.6%) is the most common treatment, however, and fabric impressing is third (23.7%). Simple stamping is seen in a few cases. “The modal paste for the Berkeley series sample is a compact, gritty clay body with abundant medium and medium-to-coarse background sand inclusions” (Cable 2002: 222).

Cape Fear (South 1976; Cable and Cantley 1998; Anderson 1982 (link)). Stanley South's Cape Fear series has been criticized as being too broad (Cable and Cantley 1998). Indeed, as it can contain nearly all sand tempered textile marked wares there is much to be said for refining the definition. South said that the paste had a high percentage of sand, with occasional large inclusions, but not enough to consider them any more than incidental. He noted that cord marking and fabric impressing had a reverse relationship when compared with Hanover. Cord marked dominates in Cape Fear, with fabric impressed more common in Hanover. He noted a loose heavy cord and a tightly twisted fine cord. Rigid warp fabric includes both heavy and fine fabric varieties. Both cord and fabric marked rims and upper vessel interiors were noted.

Joe Herbert (2009: 16, 17) narrows the definition to exclude net impressed wares, and feels Cape Fear has a lower proportion of sand in its paste than New River. He thinks cord marking that is perpendicular to the rim, and rigid warp fabric impressions are key characteristics. Earlier "cord marking is more often parallel or haphazardly oblique" (Herbert 2009a: 17) and earlier fabric tends to have a flexible fibrous warp. Joe says this is the same ware that Phelps would call "Mount Pleasant." In 2003 Herbert thought Cape Fear dated between 400BC and 400AD, but in 2009, with 16 empirical dates in hand, he narrows the range to 300BC to 300AD.

In South Carolina ceramics called Cape Fear have received carbon dates in nine cases. These all range between 611 and 1534 AD. Five of these were from Mattassee Lake (Anderson et al 1982), and they were internally consistent, ranging from 611-775AD. At Sandy Island (Clement et al 2001) the dates were highly variable, with two from charcoal at 860 and 841AD, and two from shell at 1523 and 1534, (median with Calib 5,1, reservoir effect applied or 1158 and 1224 without). Either way everything they called Cape Fear is later than Anderson's Cape Fear, and all of these are later than everything Joe Herbert calls Cape Fear. So again the importance of keeping in mind that the clustering of attributes that we use to define types and ware groups can occur, and reoccur over time and space is illustrated.

Deep Creek (Phelps 1983) Joe Herbert (link) placed this in his New River series, while others have considered it a “Deptford” correlate (see Deptford and Cape Fear). David Phelps described it as having a moderately compact to friable paste, with coarse sand temper making up 40-50% of the body fabric. Cord marked, fabric impressed, net impressed and simple stamped surfaces are found. It was thought to date between 1000BC and 1AD initially (Phelps 1981a). Joe Herbert has 12 dates for Deep Creek/New River that range between 1865 and 160BC. At Minim Island oyster shell from levels containing pottery called Deep Creek received two carbon dates of AD233 (Calib 5.1, median, no reservoir effect, or median AD613 with). These were found in context with Deptford check stamped wares with the same paste characteristics, and they continued to be made well after the Deptford occupation. Thus some call similar or possibly identical pottery “Deptford cord marked” or Deptford fabric impressed” (Trinkley ed. 1991; Poplin et al 2004).

Dunlap (Jennings and Fairbanks 1940; (link)) This is a fabric impressed ware found in North Georgia. The vessel walls are thin. The thread of the fabric is heavy.

Hanover (type discussion link references: South 1976 (link), Herbert 2003 (link), Cable and Cantley 1998, 2005 (link). Morphologically Hanover and Wilmington grog tempered wares run together, with no absolutely clear boundaries. Although there are phases within the group, Hanover and Wilmington are essentially the same ware. This causes confusion in dating and cultural affiliation assignment. For instance, following South (1976) David Anderson believed Hanover was earlier, and that a North-South chain of adoption was indicated. As a result he saw no problem with grog tempered wares with dentate stamping being considered "Refuge." Subsequent research has shown that there was a fluorescence of Hanover that was in fact much later than the first appearance of grog tempered wares in the south.

Leaving the Refuge variant aside, sherds identified as “Hanover” have received the earliest carbon dates, with some as early as 230BC, uncorrected (South and Widmer 1976). However, South's dates, when calibrated with Calib 5.1 and with correction for the reservoir effect applied (see Thomas 2008 link), have intercept dates of AD237 and AD270. Without the reservoir effect calibration the dates are 199BC, and 121BC. Thus the earliest dates may not be valid.

Joe Herbert (2003) and John Cable (1998) have posited two phases of Hanover in North and South Carolina that appear to be valid on the Pee Dee at 38DA75 as well. Hanover I is given a date range of 400-800AD. Hanover II dates from 800-1500AD, though the latest dates may be specious. It is the most common pottery type at the Kolb Site (38DA75) and has received four dates there that range between 837 and 1132 AD. The dates in the project database range between 55 and 1566AD (Calib 5.1, median). There are five dates between 55 and 285AD, and thirteen between 491 and 1356AD (average 818AD). Dates for wares identified as Wilmington fit well with the later group, ranging from 515-1384AD (average 1090AD).

The earlier phase has a sandy paste, with finely crushed grog. It is more often cord marked, and decorations include fabric impressed and check stamped. Hanover II has a fine, silty paste -Anderson's fs/clay paste- with little sand and more coarse grog. It is most often fabric impressed. In more recent and detailed analysis John Cable identified seven paste variants which should have temporal significance, but unfortunately empirical dates could not be obtained (Cable et al 2005: 120). Hanover appears to be focused on the coast at about the NC-SC line, but is found well inland in the Cape Fear and Pee Dee drainages.

Grog tempered wares are less common inland on the Santee and Savannah, but are found on the coast as Wilmington and St. Catherines wares. The St. Catherines variant tends to be more compact, finer ground bits of grog, and finer cord, but the distinction should be made with proper consideration to context, rather than on a sherd by sherd basis (DePratter 1978; Thomas et al 2008).

Distribution maps show a curious result. With only a single exception in Kershaw County (Cable 2009) pottery called Hanover is only found north of Charleston, while Wilmington is only found to the south.

Mossy Oak (Wauchope 1966 (link)) This is a name that is no longer in use. It was originally thought to be an early ware. A simple stamped variant was thought to equate to simple stamped Deptford, but subsequent work found that it actually dated after 1000AD. Mark Williams suggests this be re-named to Vining Simple Stamped. A cord marked variant was identified by Wauchope that did not carry over to the Vining series, but Williams recommends that the name not be used. He does not suggest what to call the cord marked variant however.

Mount Pleasant (Phelps 1983; Herbert 2003) Mount Pleasant is thought to evolve from Deep Creek. David Phelps thought it was the same as Cape Fear (Herbert 2003: 189). It tends to have fine to medium sand temper, with occasional large inclusions (Herbert 2009: 12). Phelps thought that net impressing increased and incising was also seen for the first time. He assumed a 300BC to 800AD range, but only had dates for AD200-800 (actually 265+/-65 to 890+/-90 or 200-980AD). Herbert has 24 dates associated with Mount Pleasant that push the date to as late as 1300AD (Herbert 2009: 27) throwing the type as conceived originally into question.

New River (Loftfield 1976; Herbert 2003). New River was first defined by Tom Loftfield and was later re-defined by Joe Herbert, who brings the Deep Creek and Lenoir types into the fold. It is a sand tempered ware that is most often cord marked, followed by fabric impressed. He says it has a high density coarse sand temper and suggests "grit" (defined as very coarse to granule size) inclusions (Herbert 2009:8). Cord marked, fabric impressed, net, simple stamped and plain types are noted. Phelps proposed three periods for his Deep Creek, but Herbert does not believe the seriation is valid at this point.

Moser et al (2009:198) note that it is easy to mistake one of the many lithic tempered wares for New River however. Herbert has twelve dates that range from 1865-130BC (Herbert 2009: 10). Moser et al got a TL date of 3850+/-220BP (1900BC). Moser et al obtained a carbon date of 2810+/-70, or about 950BC (median, applying Calib 5.1). Cord marked (56%) and textile marked (net, fabric impressed - 23.1%) were the most common surface treatments at 31NH747 (Moser et al 2009: 198).

Yadkin (Coe 1964, 1995 (link); Oliver 1992; Anderson et al 1996 (link)). Joffrey Coe defined the Yadkin type at the Doerschuck site. There it was superior stratigraphically to Badin ware. It differs from Badin both subtly- the cordage for cordmarking was tighter wound, and the fabric impressions were finer- and profoundly- sherds came to be tempered with large quantities of crushed quartz. Coe estimates that 30-40% of the body is made up of these chunks, which range in size from 1-8mm. Surfaces are mostly cord marked and fabric impressed, with the cord impressions often being smoothed over. Interior surfaces were smoothed but not burnished. A minority type was Yadkin linear check stamped. This had the same appearance, except for the stamping, and Coe believed it was made locally.

Yadkin has become a source of confusion at least partially as a result of Coe's statements that some 46 sherds had “clay temper” added to the quartz and that “one other sherd with this clay-quartz tempered paste was decorated with a dentate stamp” (Coe 1964: 30). This has caused some to correlate Yadkin and Refuge, and assign it a very early date as a result. But the Yadkin type has even worse problems, as pottery that was “Yadkin-like” (Anderson 1982) in that it had large inclusions came to be considered true “Yadkin” (Blanton et al 1986; Trinkley and Adams 1993; and many others).

Add to this the clay tempering issue and the general similarity of the major pottery series discussed by Herbert (2003, 2009) and it is evident that we should apply Coe's definition strictly, and not use the Yadkin name unless the sherds are densely tempered with crushed quartz. "Yadkin-like" sherds tempered with rounded granules or pebbles should not be called "Yadkin." This is Joe Herbert's conclusion (2009a) though I arrived at the same conclusion independently.

Wando (Poplin 2005- link images at link) "Wando" is a name given to limestone or marl tempered pottery found in the Charleston Harbor drainage system. People were aware of it for years, but no one gave it a name until Adams and Trinkley identified the check stamped and cord marked types in 1993. Later fabric impressed, brushed, incised, punctate and simple stamped variants were identified (Poplin 2005). Interiors are occasionally shell scraped. Vessel walls are 6-8mm thick (thicker than Hamps Landing) and vessels include straight sided bowls and jars. Though decorations are comparable to Deptford / Wilmington, which he calls "Middle Woodland" he cites five carbon dates that range from 615-1140AD (intercept, calibrated by Poplin). He looked at over a hundred assemblages in the area, and found Wando associated with "Middle" and Late Woodland" assemblages regularly, but a seriation chart shows they tend to occur in the upper levels, above Deptford. He suggests that cord and fabric marked decorations occur early, and check and simple stamped later. He concludes that Wando ceramics were the product of a specific group. and that this is a local type. With only a single exception all ceramics identified as Wando have been found in the Charleston area (map link). Comparisons are often made with the similarly tempered and decorated Hamps Landing type defined near Wilmington, but that ware is thought to be much earlier.

White Oak (Loftfield 1976) Joe Herbert has folded this into his Townsend series. The type appears to be the same as South's Oak Island and Phelps' Colington.

Wilmington (Anderson et al 1996 Caldwell and Waring 1939; DePratter 1991; also link). See Hanover discussion above. Wilmington is defined broadly, and includes eight to twelve “types” in Georgia (link). St. Catherines evolves from Wilmington, which seems to evolve from Deptford. DePratter (link) places Wilmington into two phases. The earlier is AD 500-600 and is seen as essentially the same as Deptford, with newly introduced grog temper. His Wilmington II phase, AD600-1000, sees the abandonment of check and complicated stamped variants and more emphasis on cord marked, plain, brushed and fabric marked surfaces.

The best carbon dates are from St. Catherines Island, where Wilmington had a two sigma range of AD310-780 (highly calibrated). The relatively few (14) clear Wilmington dates in the project database range from 515-1384AD (Calib 5.1, median). This is a fairly mixed bag, however, as five are pre-1000AD, four are between 1072 and 1198AD, and five are from the 1300s. The first group is compatible with the St. Catherines Island dates, but the middle and late groups would be contemporary with St. Catherines and Savannah, respectively. Clearly in depth research like that of Joe Herbert in North Carolina, and David Hurst Thomas (et al) in Georgia is needed to sort this out, as there are obviously phases and local variations that are going unnoticed.