Up-"dating" the Stubbs Cluster, Sort of....

Frank L. Cowan

F. Cowan & Associates

Ted S. Sunderhaus

Robert A. Genheimer

Cincinnati Museum Center

Ohio Archaeological Council © 2003

Three seasons of intensive excavation in scattered portions of the Stubbs Earthworks site complex (33WA1, 33WA256, 33WA257, 33WA258, 33WA260, and 33WA362) in Warren County, Ohio uncovered large numbers of wooden architectural remnants and documented a bewildering array of Hopewellian building forms (Cowan et al. 1998, 1999; Sunderhaus et al. 2001). The striking densities of wooden structures immediately adjacent the earthworks, coupled with their architectural variety, demand that we pay close attention to all available chronological controls as we attempt to interpret the site. The results of 18 previously reported radiocarbon age estimates (Cowan and Sunderhaus 2002) indicate considerable contemporaneous variability in Ohio Hopewell wooden architecture. However, that series of radiocarbon dates also showed us that some structures were very much earlier in age than nearby artifact and feature associations suggested. Had we not invested in multiple radiocarbon dates for the circular Structure 11, for example, we would not have had a clue to the presence of Early Woodland architecture in at least one portion of the site. Given those experiences, we continue to whittle away at placing the known elements of the Stubbs Earthworks complex into an intelligible temporal framework.

Four accelerator mass spectrometer (AMS) radiocarbon dates are now available for two previously undated wooden structures excavated at Stubbs during the 1998 field season (Cowan et al. 1998). In addition, four conventional radiocarbon samples were recently analyzed from the adjacent Smith site (33WA362), one of the Stubbs Cluster of Hopewell sites excavated in 2001; those samples targeted the Smith site structure and a large, artifact-rich pit (Sunderhaus et al. 2001). These new age estimates help establish the period of ceremonial use of the large Stubbs Earthworks. In addition, age estimates for post mold fills from the three structures bolster already compelling evidence that contemporaneous Ohio Hopewell architectural forms were highly varied.

The chronological controls and other detailed data developments being brought to the study of the Stubbs Earthworks complex are due, in part, to several excavation strategies we decided upon at the very beginning of the project in early 1998. Faced with very large areas of investigation (about 86 acres in 1998 and several additional acres in 2001) and a very limited daily labor force (although supplemented by the intermittent assistance of hundreds of wonderful volunteers), we elected to maximize feature exposure and excavation. This decision necessarily meant "back- loading" the processing of excavation results. Thus, we exposed many large sampling transects, shovel-shaved and trowel-scraped those transects (again and again and again), and mapped and excavated nearly a 1,000 features in 10 months of field work in 1998 and 1999 and almost another 100 over the course of a month at Smith in 2001. Instead of screening feature fills through 1/4-inch mesh hardware cloth, we elected to bag all feature fills, excepting some from the very largest pits, as flotation samples. All in all, we collected more than 1,600 bags of feature fill! The downside of this field strategy is that, after all this time, the Cincinnati Museum Center's excellent corps of archaeology lab volunteers is still floating bags of Stubbs dirt, and we are now making hard decisions about which of the remaining samples must be floated and which can be water-screened. One of the benefits of that early decision is that we are developing an unprecedented corpus of context-specific archaeobotanical data and have many potential samples for problem-specific radiometric dating.

Marge Schroeder, a paleoethnobotanist at the Illinois State Museum, has thus far analyzed charcoal samples from 52 features (971 liters of feature fill, water-separated and then floated again in a zinc chloride solution) from Stubbs and 21 features (145 liters) from Smith for their botanical content. Most of the features analyzed at present were post molds since we chose to place our radiocarbon priorities on the wooden architecture. We selected the features to be dated, based on charcoal quantities and the botanical content of features, and Schroeder selected and prepared the charcoal specimens for submission to the radiocarbon laboratories. The reported AMS samples for Stubbs, run in the fall of 2002, were analyzed by Beta Analytic, Inc., while the conventional dating samples, run in the summer of 2003, were analyzed by the Illinois State Geological Survey's radiocarbon laboratory. All radiocarbon samples were measured for 12C/13C ratios and are here reported as "corrected" for that variation.

Stubbs Earthworks (33WA1)

Stubbs Structure 1 (Figure 1) was a 5-by-8-meter quadrilateral structure with rounded corners and four asymmetrically placed interior support posts (Cowan et al. 1998). By our estimates of the location of the now-invisible Stubbs Earthworks embankments, Structure 1 appears to have been located very close (perhaps just 20 meters) to the outside of the southern wall of the rectilinear Stubbs Earthworks enclosure. Of the seven post mold fills thus far studied for their botanical content, we chose charcoal from two widely separated post molds, Features 42 and 48, for AMS analysis. The results of the two age estimates (Table 1) are within one standard deviation of each other, and we take the average, 1785 +/- 30 radiocarbon years B.P. to be the best single estimate of the structure's age. Calibrated for known secular variations in the production of the 14C isotope (Ramsey 2000), the most likely calendar date for the structure lies between A.D. 210 and A.D. 330 (59.9 percent of the distribution) at one standard deviation, or between A.D. 130 and A.D. 340 (95.4 percent) at two standard deviations. That date range is very close to those of other quadrilateral structures at the site.

Figure 1.  Stubbs Structure 1.  Dated post molds are labeled.
Figure 1. Stubbs Structure 1. Dated post molds are labeled.

Table 1. Radiocarbon dates for Structures 1 and 3 at the Stubbs Earthworks site and Structure 1 and Feature 37 at the Smith site.


Structure shape Structure Feature
(post molds)
Lab number Radiocarbon
age BP
(corrected for
C12/C13 ratio)
average age BP
per structure
1 42 Beta-166639 1750 +/- 40 1785 +/- 30
48 Beta-166641 1820 +/- 40
3 80 Beta-166642 1730 +/- 40 1785 +/- 30
202 Beta-166640 1840 +/- 40
1 14 ISGS-5438 1890 +/- 70 not averaged
23 ISGS-5437 1690 +/- 70
Smith pit -- 37
ISGS-5440 1820 +/- 70 not averaged
ISGS-5441 1420 +/- 70

Stubbs Structure 3 (Figure 2) is of a unique form in Ohio Hopewell, to the best of our knowledge, although it may be somewhat similar to Middle Woodland structures known from the Six Flags site (9FU14) in southwestern Atlanta, Georgia (Kelly 1979:2, Figures 1.2, 1.3). It is a circular structure of 8-meters diameter with paired inside-and-outside posts. The close one- to-one match of the paired posts argues against an interpretation of a replacement building enlarging upon a previous one. Whether the interior circle of posts represents an interior bench or a series of support posts, we are not sure; there doesn't seem to be much systematic difference in their diameters or depths. AMS dates were run from charcoal in the fills of two widely separated postholes, Features 80 and 202. The two age estimates (Table 1) diverge by more than one standard deviation but are well within two standard deviations of statistical error. We are inclined to accept the average of the two age estimates as the best single approximation of the true age of the structure: 1785 +/- 30 B.P. Translated into calendar years, that radiocarbon age estimate equates to the same date ranges as Structure 1: between A.D. 210 and A.D. 330 (59.9 percent of the distribution) at one standard deviation, or A.D. 130 and A.D. 340 (95.4 percent) at two standard deviations.

Figure 2.  Stubbs Structure 3.  Dated post molds are labeled.
Figure 2. Stubbs Structure 3. Dated post molds are labeled.

Despite radically different architectural forms, Stubbs Structures 1 and 3 are of identical age, within the levels of precision and accuracy afforded by radiocarbon dating. Hopewell architecture was remarkably varied. The accumulating evidence from the Stubbs Earthworks site demonstrates that such architectural variability is not time-dependent but relates to other, not as yet understood, functional or social variables.

Figure 3 summarizes the calibrated date distributions (at one standard deviation) for eight of the Hopewellian house-like structures on the terrace immediately surrounding the earthwork enclosure. Quadrilateral Structures 1, 4 and 5 have date ranges that overlap quite considerably at one standard deviation and are not statistically differentiable from the age estimates for the double-posted circular Structure 3 or the C-shaped Structure 6. Within the level of precision of radiocarbon dating, these diverse architectural forms are essentially contemporaneous. The quadrilateral wall-trench Structures 19 and 21 are dated, as yet, only by a single conventional radiocarbon date of wood charcoal from a small pit that fortuitously intruded into the overlap of two previously abandoned wall-trench structures. More radiocarbon age estimates will be necessary to get a better handle on the actual dates of the wall-trench structures that were so abundant in Transect 27. The post molds and wall-trench fills of those several structures were remarkably depauperate in charcoal, and we are presently trying to isolate flotation samples that will permit AMS dating. Circular Structure 2 is the most obvious outlier in this distribution, and it gives us pause. The illustrated date range is based on the average of similar radiocarbon dates from two post molds but ignores a much earlier date, 1890 +/- 70 B.P., from another post mold (reported in Cowan and Sunderhaus 2002:Table 1). Secular time-calibration of the average of the two younger age estimates push the date range rather later than we would have expected, and it may be necessary to reexamine this anomaly by AMS-dating additional post molds to test alternative hypotheses.

Figure 3.  Calibrated radiocarbon dates for Stubbs house-like structures.
Figure 3. Calibrated radiocarbon dates for Stubbs house-like structures.

Our interest in refining the chronological relationships of these structures is not limited to the problem of architectural variability. Such chronological refinement goes to the heart of such interesting questions as how long an individual Ohio Hopewell earthwork complex may have been in active use and how its use may have related to that of other neighboring earthworks. In this case, how does the Stubbs Earthworks relate, chronologically, socially and functionally, to the nearby Fort Ancient site and Fosters Works, as well as to other Hopewell earthworks in the Little Miami River valley?

Smith (33WA362)

If radiocarbon dating the various elements of the lower-terrace Stubbs Earthworks complex has experienced reasonably few hitches, a consistent dating of elements at the presumptively associated Smith site has proven much more frustrating. The Smith site (Genheimer 1996, 1997; Sunderhaus et al. 2001) is located on a high, narrow terrace remnant immediately overlooking the Stubbs Earthworks. Emergency salvage excavations in advance of subdivision construction in 2001 focused on three areas of investigation. One area revealed a complete Hopewellian house-like structure, while a second area showed evidence for many repetitive reoccupations with a chaotically dense array of post molds and pit features. The third area of salvage investigation focused on a very large pit feature that contained extraordinary quantities of lithic, ceramic, charcoal and bone debris (Sunderhaus et al. 2001). Our recent efforts to obtain consistent radiocarbon age estimates for Structure 1 and the large Feature 37 just go to show that radiocarbon dating can often give inconsistent and perplexing results.

Smith Structure 1 (Figure 4) was an 8-by-8-meter, single-posted quadrilateral structure with rounded corners (double-posted at the northeast and northwest corners), a very large central post and six additional, smaller interior posts. A very wide doorway was located in the south wall, and a possible second, narrower doorway occurred along the north wall near the northeast corner. The excavators' impressions were that the structure may have burnt down, based on the observation that the wall postholes generally contained quite a bit of charcoal while the interior support and/or partition posts contained much less charcoal. That interpretation appears to be borne out by analysis of the wood charcoal remains from flotation samples of 20 of the post molds. Most individual post molds contained wood charcoal attributable to only a single genus of wood, and exterior wall posthole samples tended to contain more charcoal than interior postholes. Of the studied post mold samples, nine posts were oak, seven or eight were hickory, one (possibly two) were ash, and one was black walnut. Only one post mold (Feature 14, adjacent the possible northern doorway) had a clear mix of wood types (hickory, black walnut, oak and "ring porous"), suggesting a more general accumulation of midden debris deflating into the posthole. Five floated post mold fills contained carbonized nutshell fragments, and these were dominated by hazelnut (Corylus), although hickory and walnut were also present. Small numbers of carbonized seeds were also included in the mix of sediments that deflated into four of the post molds. Five of the seeds remain unidentified, but five are identified as maygrass (Phalaris). Hazelnuts were a preferred nut food among many Middle Woodland populations in the Midwest (e.g., Asch and Asch 1985:349-354), and maygrass is well-known as one of the starchy cultivated seeds of the Eastern Agricultural Complex (e.g., Smith 1992).

By virtue of having relatively large quantities of charcoal, Features 14 and 23 (Figure 4) were selected for conventional radiocarbon dating. As noted above, Feature 14 appears to bound a small doorway and contained a mix of wood charcoal of different genera, hazelnut and black walnut nutshell fragments and both maygrass and other unidentified seeds. Its radiocarbon age estimate (ISGS-5438) was 1890 +/- 70 B.P., slightly earlier than most Stubbs dates, but well in keeping with the structure's clearly Hopewellian character. Feature 23, on the other hand, yielded a much younger date. Feature 23 contained the most wood charcoal of any of the Smith Structure 1 samples (and the submitted radiocarbon sample was quite large) and consisted solely of hickory wood (on the basis of the identification of 20 wood charcoal specimens larger than 2 mm) with no nutshell or seeds. Its radiocarbon age (ISGS-5437) is reported as 1690 +/- 70 B.P., quite late in the apparent Hopewell sequence. The difference between the two samples, which should have been roughly same-aged, is greater than would be expected by random radiometric counting differences at two standard deviations. Consequently, we are not statistically justified in averaging the two sample results, and, empirically, we are not justified in arguing for acceptance of either date over the other. More radiocarbon dates will be necessary to define a central tendency in age measurements for Structure 1.

Figure 4.  Smith Structure 1.  Dated post molds are labeled.
Figure 4. Smith Structure 1. Dated post molds are labeled.

Feature 37 was a most extraordinary pit in our experiences with Ohio Hopewell archaeology. Not only was Feature 37 unusually large, ca. 1.5 meter wide and nearly a meter deep, it contained very large quantities of almost all things Hopewellian -- scores of bladelets and many heat-fractured biface fragments of diverse, exotic lithic materials, sherds from at least 23 vessels of various kinds, and thousands of mica scraps (Sunderhaus et al. 2001). It also contained lots and lots of charcoal. A cinch to date, right?

Although Feature 37 was excavated in a hurry under very adverse archaeological circumstances, we managed to retrieve 42 liters of feature fill for flotation and botanical analysis. The recovered sediments included approximately 106 grams of charcoal. Of a representative sub- sample, 20 wood charcoal fragments larger than 2 mm in size were dominated by oak with a small quantity of elm. Large numbers of carbonized nutshell fragments were found within the charcoal, approximately 78 percent of which were hickory (Carya), with the remainder Juglandaceae (hickory and/or walnut). Given the large quantity of feature fill and its richness in other contents, it is not surprising that the studied sample included many of the common members of the Eastern Agricultural Complex, maygrass (Phalaris), goosefoot (Chenopodium) and knotweed (Polygonum), as well as blackberry (Rubus) and, of course, bedstraw (Galium), a weed common in disturbed areas.

Given the large quantity of charcoal and the very interesting sample of Hopewellian artifacts, we elected to run two radiocarbon samples to get a good chronological "fix" on this feature. One sample was composed of wood charcoal and the other composed solely of nutshell, an annual growth that should be as contemporary with the burning event as one could ask. Should any discrepancy arise in the dating results, we would expect the sample composed of nutshells to be a little bit younger and a more reliable date than the wood sample, which could include somewhat older wood. Well, we predicted the trend right, but the difference between samples exceeded all expectations. The wood sample, ISGS-5440, returned a result of 1820 +/- 70 B.P., a most pleasing result. On the other hand, the nutshell sample, ISGS-5441, yielded a radiocarbon age estimate of 1420 +/- 70 B.P.! The 400-year mean difference between the two samples is an order of magnitude greater than we could have predicted on the basis of any expectations about prehistoric firewood collecting behaviors. The nutshell date is certainly post-Hopewellian, although the context is certainly not. Fortunately, our 42-liter flotation sample from this feature contains much more material from which we can submit additional radiocarbon samples, once we think of a strategy that would provide an unambiguous resolution to this extraordinary discrepancy.

Concluding Thoughts

One might sometimes fondly recall the "good old days" when, after excavating a large, complex site, one could "date" it with one or, maybe, two radiocarbon age estimates. The resulting date(s) would then either be assumed to establish a meaningful chronological fix for the site (or even for a whole phase), or the date(s) might be "rejected" if they didn't meet prior expectations. We are rather tempted to do that with the present Smith site dates: accept the two older dates and reject the others as "too young" due to some inscrutable radiocarbon "black magic." Such a strategy, of course, simply affirms prior biases and assumes that which must be demonstrated. Strong chronological controls are a baseline necessity if we are to learn about the populations that created and employed the Stubbs Earthworks, or other earthwork complexes, and as we try to understand the social and functional relationships among earthwork complexes. Despite occasional frustrations with the random uncertainties of radiocarbon dating, we are glad we elected field strategies that maximized the possibility of multiple age estimates for as many contexts as possible to create a sound baseline for addressing more interesting research questions.

References Cited

Asch, D. L., and N. B. Asch
 1985  Archeobotany. In Smiling Dan: Structure and Function at a Middle Woodland Settlement in the Illinois Valley, edited by B. D. Stafford and M. B. Sant, pp. 327-401. Kampsville Archeological Center Research Series 2, Center for American Archeology, Kampsville, Illinois.

Cowan, F. L. and T. S. Sunderhaus
 2002  Dating the Stubbs "Woodworks" [HTML Title]. Available: http://www.ohioarchaeology.org/cowan_03_2002.html Date of use: Oct 25 2003.

Cowan, F. L., T. S. Sunderhaus, and R. A. Genheimer
 1998  Notes from the Field: An Update from the Stubbs Earthworks Site. The Ohio Archaeological Council Newsletter 10(2):6-13.

 1999  Notes from the Field, 1999: More Hopewell "Houses" at the Stubbs Earthworks Site. The Ohio Archaeological Council Newsletter 11(2):11-16.

Genheimer, R. A.
 1996  Bladelets are Tools Too: The Predominance of Bladelets Among Formal Tools at Ohio Hopewell Sites. In A View from the Core: A Synthesis of Ohio Hopewell Archaeology, edited by P. J. Pacheco, pp. 92-107. The Ohio Archaeological Council, Columbus.

 1997  Stubbs Cluster: Hopewellian Site Dynamics at a Forgotten Little Miami River Valley Settlement. In Ohio Hopewell Community Organization, edited by W. S. Dancey and P. J. Pacheco, pp. 283-309. The Kent State University Press, Kent, Ohio.

Kelly, A. R.
 1979  Hopewellian Studies in American Archaeology: Hopewell after Twenty Years. In Hopewell Archaeology: The Chillicothe Conference, edited by D. S. Brose and N. Greber, pp. 1- 2. The Kent University Press, Kent, Ohio.

Ramsey, C. B.
 2000  OxCal Program v3.5. Oxford Radiocarbon Accelerator Unit, University of Oxford. www.rlaha.ox.ac.uk/orau/06_01.htm.

Smith, B. D.
 1992  Rivers of Change: Essays on Early Agriculture in Eastern North America. Smithsonian Institution Press, Washington, D.C.

Sunderhaus, T. S., R. E. Riggs, and F. L. Cowan
 2001  The Smith Site: A Small Hopewell Site Overlooking the Stubbs Earthworks [HTML Title]. Available: http://www.ohioarchaeology.org/sunderhaus_10_2001.html. Date of use: Oct 25 2003.