Data Recovery Excavations at the Maple Swamp (38HR309) and Big Jones (38HR315) Sites on the Conway Bypass, Horry County, SC
by John Cable, Kenneth Styer and Charles Cantley, New South Associates, Stone Mountain, Ga. 1998
These sites are located inland in Horry County. HR309 is on the edge of a swamp (Poplar Swamp actually- not Maple). HR315 is located on the edge of a large area of swamps and Carolina Bays. This is a relatively small and diffuse site they interpret as a men's hunting camp. HR309 was more heavily occupied.
In the ceramic analysis chapter John Cable sets out his task succinctly. He says Stan South's taxonomy successfully describes the ceramic variability in broad terms, but can be refined. Prior to South little or no work had been done. John wants to build on more recent work. He says that efforts at establishing a sequence have suffered because people have relied too heavily on surface decoration. He will use a multivariate attribute based approach to address the problem. He has written elsewhere (Cable 1993) that paste groups are discernible that are as reliable as surface decoration, and which in fact, are found with many different surface decorations. Thus analysis must consider both variables together to make finer distinctions and build a more sensitive sequence. He sets up three major ware groups: Thoms Creek/ Refuge; Deptford/Cape Fear; and Hanover. Stallings and complicated stamped wares are also present, but in small amounts.
This report introduces his multivariate analysis technique in detail. He looked at sixteen variables:
Ceramic Form refers to body part, not vessel shape. That is, body, rim, base etc.
Exterior Surface Treatment. He identified 23 variations, including four indeterminate varieties. He has plain, unsmoothed plain, simple stamped- paddle and dowel- check and liner check stamped, complicated stamp, shell scraped, fine, regular, heavy and wide spaced cord marked, fabric impressed, random, reed, and reed drag and jab punctate, and finger pinched.
Paste Variables. This is a major subgroup. He feels that paste recipes are the result of a series of conscious decisions, rather than a random result of available clay. Other researchers credit local clay with more importance. He feels that intensive multivariate analysis will allow evolutionary change to be monitored. He says four major paste characteristics are related to these choices: temper, grain of temper, density of temper and hardness of paste. He sets up four variables that he says are easily identifiable: background sand inclusions, other inclusion types, inclusion density and paste hardness.
He notes five modes of background inclusions: fine or none, well sorted medium, heterogeneous medium to coarse, very coarse and granule to pebble sized inclusions.
Other temper inclusions include crushed quartz, subangular pebbles, and coarse sand again. In this case I think he means occasional coarse grain inclusions in numbers too low to be considered “temper.” He splits clay tempers into sherd and grog in fine (>1mm) and coarse <1mm) grades. He also identified fiber, shell and “hole.” Hole tempering is what remains when calcium based tempers (like limestone or shell) leach out and leave a void.
Density is an ordinal variable. He splits sand, “hole” and shell into low (0-15%), medium (15-30%) and high (>30%). Grog/sherd is split into low (<5%), medium (5-15%) and high (>15%). For fiber, ,15% is low and 15-30% is medium. There is no “high” value.
Hardness. He doesn't like using the Moh's Scale, as suggested by Shepard (1964: 114-116) because of the variability in inclusions and uneven surfaces. He developed the “Cable Thumbnail Test” to address this. Essentially if the sherd edge is easily broken off it is friable, or soft. Moderate is, well, moderate and hard is harder. Results may vary from researcher to researcher obviously.
He takes all of the paste variables and uses them to determine a Coarseness Grade for the sand tempered wares. He adds .3 or .6 to the Wentworth value if coarse background inclusions are present, and .6 if granular/pebble are present (ie, fine= 1,0, 1.3, 1.6) . Not sure why, but there's no “very coarse” group.
In assemblage terms he says Big Jones has finer inclusions, lower inclusion density and softer paste, reflecting the dominance of Thoms Creek there, while Maple had more late material.
Metric Variables. Length, width and thickness were measured. Length and width were used to determine sherd size, and make statements about site formation. Interestingly, at the two sites sub-plowzone sherds tended to be larger, but there was a wide size range. Plowzone sherds from both sites were smaller and more homogenous in size, showing that plowing does, indeed, break up the sherds. Common sense, yes, but its nice to have hard evidence too.
The Erosion variable is one that some might call “sherd condition.” It is broken into “extremely,” “Moderately” and “crisp.” His decision to not consider residuals and extremely eroded sherds in the analysis is seen to detract from the usefulness of this variable, but when they are added in it is clear that small, and eroded sherds are more common in the plowzone, reflecting the size variable.
Vessel Form and Function Variables relate primarily to rim morphology and decoration. They include rim type, lip form and neck orientation. Most rims were plain, with slightly thickened exteriors being next most common. This occurs in Thoms Creek, but more often is a Deptford/Cape Fear group trait. He identifies ten lip forms that are variants of the basic rounded and flattened shapes. Thoms Creek/Refuge lips are more often rounded while Deptford/Cape Fear has more squared/flattened rims. Neck Orientation is broken into four categories: straight, everted, flaring, and incurved. The early wares are straight and incurved, while the later wares flare more. Hanover is more often straight, however. He does limited vesselization, identifying eleven vessels. Three were bowls, two were conoidal jars, and the remainder were conoidal jars/bowls. Both jars were Deptford, and the bowls were Stallings (1) and Thoms Creek (2). His final variable in this group was rim decoration. Altogether 50 sherds were decorated.. This was much more common on Deptford (34), and Hanover (11) than Thoms Creek (5). Cord marking was the most common on Deptford (17) and Hanover (9), with simple stamping (14), fabric impressed (6) and tick marks (3) also included. The Thoms Creek was either simple stamped (4) or ticked (1).
John explains that focusing on surface decoration blinds us to small, but significant variations in paste recipes that can be used to refine our seriation and understanding of change and time. “It is an unfortunate but undeniable fact of the Southeastern prehistoric ceramicists life that the variation relied upon to distinguish types is continuous.” (pg 267). He feels that the variation is due to technological change, not function. Some might argue that this is almost always random, and don't try to account for it. Thus a given “type” can have fine to very coarse temper in some researchers formulations, for instance). He points out the borrowing of types from adjacent states, but notes that there is no scientific evidence (that is rigorous, replicable comparisons) that conclusively proves that this is true. He sets out to at least provide a basis for comparison by applying statistics. As someone who is 95% statistically illiterate I will leave the nuts and bolts to interested readers. It looks convincing though. He identifies three distinct clusters in Thoms Creek /Refuge and three in Deptford/Cape Fear, and two in the Hanover series. Hanover 1 has more sand and a more compact paste than Hanover 2. He thinks there may be a third that is only hinted at by the small assemblage. He makes a case for D/CF 3 and H1 being successive.
Systematics and Sequence
He discusses South's s typology, and then discusses the Maple/Jones sequence. He thinks Stallings/Thoms Creek/Refuge arrived relatively late in the north coast. He thinks the Stallings is comparable to Sassaman's group III, which he thinks dates to 3500-3200BP. But only a few sherds from a single plain vessel were found. He discusses the three TC/Ref phases identified at Spanish Mount. The Horse Island Phase dates from 3700-3300BP. It is primarily punctate decorated- shell, drag and jab, reed) and plain wares. The Awendaw Phase dates 3300-3000BP and has more plain and shell scraped wares, as well as finger pinched and finger smoothed. The Minim Island phase (3000-2700BP), he says, is basically the central-north coastal variant of Refuge. The main difference between Awendaw and Refuge at Minim Island, he says, is that finger pinched decorations are replaced by dentate stamps. The Maple Swamp and Minim Island occupations are contemporary, he thinks.
The Hanover-Wilmington-Cape Fear issue. John discusses Badin, Yadkin, and Deep Creek from NC and other regions, noting an early popularity of cord marking which gives way to a dominance by fabric marked wares. He notes that the “Yadkin” assemblage from SU83 is only “Yadkin Like” and that it appears to be the same thing as the Deep Creek at Minim Island (which I said when I saw the ceramics at the time). He notes that in NC grog tempered pottery does not precede the coarse sand tempered wares. He denies David Anderson's (1982) assertion that Hanover/Wilmington and early Deptford are associated, and uses the lack of H/W at Minim Island as evidence. He cites several carbon dates for Hanover that place it in the AD400ish range, before Wilmington appears in the mouth of the Savannah sequence. There are contemporary sand tempered and grog tempered wares throughout the 300BC to 700AD period however, and even later if we include the St Catherines and Carteret series.
John believes that the “Yadkin-Like” pottery some have identified should be given a name of its own as it is a distinct, transitional ware. (pg 318).
North Coast Regional Sequence. He begins by pointing out the incompatibility of the NC paste based vs SC/Ga surface decoration based classification systems and the problems this causes when attempts are made to integrate the two in a region of “significant trait overlap” (pg 319). His chronology summarizes previous statements. He sees the assemblages beginning with Stallings 3, at about 3500BP and transitioning to Thoms Creek/Awendaw and Refuge. He feels that north coast Refuge should have a name of its own. This is followed by Cape Fear I and II. He has dropped the “Deptford” at this point, though he notes CF I correlates with the D/CF 1 cluster, and early Deptford and “Deep Creek II” (pg 322) Cape Fear II correlates with D/CF cluster 2. Hanover 1 and D/CF cluster 3 correspond. He says that D/CF 1 is the same as Phelps' Deep Creek, while D/CF 2 and 3 are the same as Mt Pleasant. Hanover I correlates with D/CF 3. He proposes a date range of AD500 to 800 for Hanover II, though more recent dates appear to extend that range a good bit. Two dates around AD900 have been obtained at 38DA75 (Kolb site) and other dates post date AD1200 (Southerlin et al).
In summary, this is a truly worthwhile effort at analyzing the pottery and establishing a chronology for the north coastal region. Although he acknowledges this in the text and discussions, his chronology chart gives a set sequence where type a replaces type b without overlap. This is a problem because the literal minded among us will not recognize the inherent overlaps, and local variation, and may try to argue that site A dates earlier of later than site B by strictly applying the dates on the chart. Enough people do this with, say, the Late Archaic/Early Woodland transition that it is worth mentioning.