Minim Island Site

The Minim Island Site- 38GE46

Minim Island is located in the Santee Delta about five miles from the open ocean. It is bounded by the North branch of the Santee and Duck and Minim Creeks. The Intracoastal Waterway cuts across the old rice fields beginning at the north end of the site. This has resulted in considerable erosion, and was the impetus for the testing (Drucker and Jackson 1984) and data recovery (Espenshade and Brockington 1989). I had the good fortune to work on both phases of work as a field tech, and on the first phase as a lab tech. The large, varied and fairly well stratified assemblage provided me with a good learning experience, and beyond that (ie, what was in it for me) is a valuable collection for us all to consider.

Speaking generally the stratigraphy consists of a shell midden about one meter thick over a dark soil midden with nearly no shell that is another 50cm thick. Chris Espenshade thought that he detected three living surfaces based on artifact density. Bioturbation- mostly by humans- has caused a degree of mixing, as sherds of the dominant Deptford Check Stamped ware are found in nearly every level. That said, the type is most common in the lower half of the shell midden, with relatively few in the upper or lower levels, except in features.

Fabric Impressed coarse and very coarse sand tempered wares are found mostly in the top 60cm. Cord Marked wares are less common and vary more, but there is a strong presence at the shell midden/soil midden interface, just as there is with Deptford Check Stamped, which suggests that they were contemporary. When we started the testing Ron Anthony and Leslie Drucker, who trained under David Phelps at East Carolina wanted to use type names Phelps had introduced for North Carolina fabric and cord marked wares: Deep Creek and Mt. Pleasant. What quickly became clear, to me at least, was that the bodies of the fabric marked and check stamped wares were identical. That is, if you looked at the interiors only you couldn't separate them. So is it valid to consider cord marked ware “Deep Creek” while otherwise identical check stamped ware is “Deptford?”

This, I suppose, is fundamental to my ambiguous feelings about pottery “types.” Does the difference in decoration/surface treatment indicate social identity? Is a motif a sign of a particular potter's mindset? Is the mixing of wares a sign of human interaction? Or did groups traveling through the area 50 years apart simply happen to find the same clay source? It's 2009 at this writing, 27 years after I first went to Minim Island and I still don't have an answer. In recent years some researchers have been calling virtually all coarse to very coarse sand tempered wares “Deptford” - ie, Deptford Cord Marked. If splitting types too finely is bad, then lumping them is also.

At minimum here's what we do have though. Historic material and complicated stamped, as well as late fine cordmarked wares are found in the top few levels only. Fabric impressed wares extend to about 60cm below surface. Check Stamped wares are concentrated in the lower levels of the shell midden and a little below the interface to about 120cm. They are clearly associated with cord marked wares at the interface, and with fabric impressed sherds above. Thoms Creek and what is called “Refuge” are concentrated in the soil midden below the Deptford wares. Corrected carbon dates of 1170 and 1440BC were obtained for the soil midden. Dates ranging from 679BC to 314BC were obtained for Deptford Check Stamped wares, and two dates of AD220 were obtained from the mixed fabric/check levels Espenshade characterizes as Deptford/Deep Creek. The 679BC Deptford date was obtained from a feature dug into the soil midden, and may be on the early side, though it is not outside of the range of dates obtained elsewhere for Deptford.

An Archaeological Study of the Minim Island Site: Early Woodland Dynamics in Coastal South Carolina.


Christopher Espenshade and Paul Brockington 1989

This will not be a full review of the report. I refer the reader to the original. As usual with CRM reports the level of effort was constrained by time and financial constraints. If it was someone's dissertation, and they had spent several years working on it the results would be much stronger. But the data is there still if anyone wants to take a crack at it. Pottery is our concern here, but the report has chapters on zooarch, ethnobotany, palynology, human osteology and more.

Pottery and analysis techniques are introduced in Chapter 2. All sherds larger than 3cm square received a preliminary analysis consisting of sorting by aplastic inclusion type and density and decorative motif, and grouping sherds by type. This was refined a bit after the secondary analysis, but Chris notes that the groups were pretty distinctive and the changes were minor. He uses the Wentworth scale, but lumps fine and very fine together. In this phase of analysis some 7,397 sherds from levels and 902 from features were examined. The sherds from carbon dated contexts from the testing phase were also examined.

In the secondary analysis he chose 185 larger sherds, preferably rims, that he felt represented the full range of variation in the collection. An interesting research project would be to take a more “blind” approach by analyzing the entire collection from a sample of the units and seeing whether Chris's results can be refined. In this phase he sent the sherds to Michael Trinkley and Chester DePratter to get their take on the collection. This illustrates the regional bias we are confronted with. Although they mostly agreed - there was an 87% correlation rate- there were instances where Trinkley called sherds “Deep Creek” while DePratter called the same thing “Deptford.” Since, as stated above, some of the “Deep Creek” decoration were on a “Deptford” body Espenshade compromises by calling some sherds “Deptford/Deep Creek.”

Espenshade looked at 16 variables. The following list and definitions are directly from the report:

“The secondary analysis examined 16 decorative or technological attributes:

1. degree of overstamping

2. presence/absence of coil breaks

3. interior surface treatment

4. shoulder form

5. rim shape

6. rim production step

7. rim diameter

8. thickness

9. core configuration

10. core percentage

11. paste color (s)

12. aplastic type

13. aplastic size

14. aplastic shape

15. aplastic density

16. paste texture

In addition, rim profiles were drawn for all rim sherds of the secondary sample. The discussions of the significance of each attribute follows. It should be noted that the attributes were measured through replicable means, and that results will accurately reflect the utility of published type descriptions for Minim Island.

Overstamping was classified as absent, present only at paddle edges, and extensive. Overstamping is closely related to subjective calls of sloppy versus well executed, and it has been hypothesized as a possible means for distinguishing Deptford and Refuge simple stamped sherds.

The relative frequency of coil breaks directly reflects the technology utilized in vessel construction, and also indicates the thoroughness of coil compaction. Coil breaks were recognized as latitudinal breaks paralleling the rim, with the break often exhibiting a regular concave or convex surface. .Interior surface treatment refers to the method utilized in finishing the inside of a vessel.

Interior surface treatment may be the only macroscopically observable difference between two types of similar paste and surface treatment. The nature of interior surface treatment was recorded with regard to the presence/absence of tool marks, the presence/absence of fingerprint grooves, and the degree of aplastic exposure or cover.

Shoulder form was recorded to provide a gross indication of vessel morphology. The form was be classified according to the type, location, and degree of vessel wall inflection. In the present study, the vessel forms were either a minor variation of the straight shouldered vessel (no major shoulder inflection) or the hemispherical form (gradual, regular inflection at the shoulder).

Rim attributes have been shown to vary with both cultural and functional factors. Rim shape refers to the form of the most extreme portion of the vessel wall; it has been referred to as "lip form" by other researchers. Rim shape was classified by the basic shape of the rim, the presence/absence of lipping, and the presence/absence of stamping or scraping .

Because there can be variability in rim form on a single vessel, rim production step was also recorded. Rim production step reconstructs, as feasible, the method utilized to produce a -given rim. Rim production step complements the rim shape attribute, because many different production steps can result in a given shape.

Rim diameter is a metric attribute which is a rough indicator of vessel size. Rim diameter was recorded for sufficiently large rim sections through use of a diameter chart.

Thickness is related to durability and vessel size. This attribute is often diagnostic of a particular ceramic tradition, suggesting that thickness directly reflects cultural norms of how a pot should look. Thickness was measured only on rim sherds, and it was measured consistently 3 cm below the rim.

The location and extent of a dark carbon retention core in a sherd cross-section is related to paste composition, aplastic content, firing practices, and possibly post-firing use. Core configuration was classified as the relative position of the dark grey, tan, and/or red paste bands,

The core percentage is a direct measure of the degree to which the organics in the past have been oxidized and burned out. Core percentage was measured as the percentage of the sherd cross-section comprised of black or dark grey paste.

The dominant dark grey, tan, and red paste colors (as appropriate) were recorded. Paste color (s) can reflect the composition and firing history of a vessel. Paste colors can indicate relative degree of oxidation and presence/absence of ferrous minerals. If firing practices were relatively consistent through time at a site, paste color may be the only indication of a change in clay resource selection, Colors were classified by the Munsell color system, for a fresh break viewed under fluorescent lighting.

The aplastic content of each sherd was evaluated under 60power magnification. Aplastic type reflects the possible source of aplastics (temper vs. natural inclusions). Aplastic type can be related to vessel function or the working characteristics of the clay, and was possibly utilized as a cultural identity marker. Aplastic types recorded included quartz, rose quartz, and grog (non-sherd clay lump).

Aplastic shape is indicative of the source of the aplastics and preparation process (if any). For example, the angular edqes of quartz aplastics in the Yadkin series pottery -have been interpreted as the result of crushing vein quartz for temper, Aplastic shape was recorded as round, sub-angular, or angular.

Aplastic size and aplastic density affect the clay workability and the physico-chemical characteristics of the fired vessel, and may also be related to cultural identity. Temper size was classified according to the Wentworth size system, through comparison with prepared clay and temper bars. Aplastic density was recorded as the number of aplastic grains within a consistent microscope view.

Paste texture is indicative of both the clay resource utilized and the firing history of a vessel. Paste texture will change as a firing progresses, and this attribute can reflect the degree of oxidation and the firing temperature. Paste texture was examined under 60 power magnification, and was classified according to the constituents of the paste (very fine vs. grainy) and their degree of compression (compact vs. open).

An additional element of the ceramic analysis was limited experimentation with the unfired clay recovered from Feature 10.Because it was suspected that the material represented pottery clay, several standard ceramic tests were conducted. The wire test was done, and linear drying shrinkage was calculated, following Shepard (1964). The aplastic content was recorded as in the secondary ceramic analysis.” (Espenshade and Brockington 1989:26-30)

So this stands as a good example of a well planned and presented analysis format. The distribution of surface treatments is addressed in Chapter 5, and then he returns to a discussion of the 16 variables in Chapter 6. I'll summarize Chapter 6 results first.

Overstamping was found with all stamped/fabric type treatments except Deptford simple stamped.

Coil Breaks were uncommon on Thoms Creek (21%), but common on Deptford (64%) and Refuge (50%) and a little less so on Deep Creek (41%).

Interior Surface Treatment: several variants identified. Thoms Creek had the only burnished examples, and tended to be shell scraped or well smoothed, leaving no marks. The latter treatment was also common on the other four ware groups he uses.

Thickness: shell scraping was used to thin the walls on Thoms Creek, which at 6.0mm was considerably thinner, on average, than the other groups. Refuge was 7.8mm, Deptford 9.3mm, Deep Creek 8.7mm. The sample of Pee Dee sherds consisted of three sherds, which is too small to draw conclusions, but they were thin also- 6.0mm.

Rim Diameter could only be extrapolated for two vessels, both of which were about 32cm in diameter. These were Thoms Creek and Deptford vessels.

Rim shape varied little in all groups. Most rims were simple rounded or flattened, with occasional lipping.

Rim Production Step seems to show continuity from Thoms Creek to Deep Creek, with thirteen variants identified. Thoms Creek had the most variation. Marks on top of the rim were common in Thoms Creek and Deptford. These were usually made with the paddle used to stamp the body and then smoothed over. Thoms Creek sherds were sometimes scraped with shell and notched, and both Thoms Creek and Refuge were shell impressed in a few cases.

The shoulder form for most vessels in all five groups was essentially flat, indicating a preponderance of straight sided vessels. However, about 45% of Thoms Creek vessels had rounded shoulders, suggesting hemispherical, or bowl forms.

Core percentage showed considerable variation in all of the types, with the exception of the Refuge Plain and Dentate Stamped which had non-oxidized cores in 88% of the sherds analyzed (n=16).

Core Configuration and Color are considered together. Thoms Creek and Refuge were similar, with tan/dark (exterior/interior) and tan/dark/tan cross sections most common. Deptford tended to be redder, which suggests possible changes in clay sources and aplastics, and/or a change in vessel use which resulted in frequent exposure to fire. Deep Creek was more often grey, while the Pee Dee sample had gray and tan examples.

Aplastic Content for the most part consisted of clear to white subangular quartz. A few examples had rose quartz sand, but generally granule size pieces were uncommon, even in the preliminary analysis. Grog temper was surprisingly uncommon, with only seven of the 8,000 odd sherds having this temper. Thoms Creek was tempered with very fine (7), fine (21), medium (14), coarse (14) and very coarse (1) sand. In the methods Espenshade stated that very fine to medium had been grouped. If so that would mean nearly 75% of the Thoms Creek had “medium” temper though a fair amount had coarse temper as well. Refuge ware tended to have coarser temper- coarse= 10, very coarse = 5. There five medium, one fine and one example with only “natural” sand. Deptford temper was predominantly coarse (16) or very coarse (54) though a few (4) had fine to medium sand in the paste. Deep Creek had a similar temper profile, suggesting continuity. Temper density is generally higher in Thoms Creek and lower in the later wares, but there is considerable variation.

Paste Texture included four variants: very fine open, very fine compact, grainy open and grainy compact.

Summary Thoms Creek tends to be thinner, with shell scraped interiors, fewer coil breaks, and more hemispherical forms. Refuge has a coarser temper and thicker walls. Deptford is thicker and has considerably coarser temper. Espenshade says that Deptford and Deep Creek were “similar across the board” (p. 167). He says that there is some evidence of continuity in the wares across time, but no clear evolution of one type to another, with the exception of Deptford to Deep Creek.

The horizontal distribution of the different surface treatments and decorations is considered in Chapter 5. Espenshade presents 13 simplified wall profiles with data values from four units overlaid by level. Because of disturbance by later occupants there is a bit of mixing. In one unit (#9) check stamped sherds are found in all but the lowest two levels but in the others it disappears by Level 12 (of 21 maximum), at the base of the shell midden. Thoms Creek and Refuge decorations are likewise found throughout the profile in small numbers, but all are concentrated in the soil midden. With only two exceptions all fabric impressed wares are in the top seven levels. Cord marked wares are a little more variable but there are concentrations at the shell midden / soil midden interface in all four units. Complicated stamped wares and historic artifacts are limited to the top four levels.

In conclusion, the Minim Island site confirms with stratigraphic and empirical evidence ideas about the relative dates of major pottery types that had largely been inferred in the past. It shows that Thoms Creek and Refuge are distinct types with shared traits. Espenshade points to the long coexistence of Thoms Creek and Stallings, and suggests that Refuge may have grown out of Stallings. Deptford does not clearly evolve from Thoms Creek and Refuge, but a similar Deptford / Deep Creek evolution is present. Espenshade suggests that in both cases distinct groups were interacting, whether through intermarriage or cooperative subsistence tasks, and the shared traits drifted in that manner.

On a personal note, the two Minim Island projects were a lot of fun and very illuminating. The people I worked with on both projects were bright and good to be around. Some had better boat driving skills than others, but we all survived.