Blades and blade-cores have been recognized as classic Hopewell diagnostic artifacts since the earliest investigations.  It is only within the past twenty-five years, however, that these materials have been consistently and systematically analyzed.  Much of this relatively recent research has focused on establishing the degree and kind of cultural-historical relations among different groups of Hopewell blade producers based on a comparison of metric attributes.  Another trend has focused on determining the function or purpose of this class of artifacts, primarily using use-wear analyses.  A third avenue of inquiry has focused on the economic organization of Hopewell blade production, especially as it pertains to the more general problem of craft specialization.  The focus of this project is in the third area.

The Site and the Artifacts

            The Turner Workshop site is a large, approx 1 acre, lithic scatter in a plowed field located on a terrace just south of the Little Miami River in Hamilton County, Ohio.  The site is in view of the Turner Earthworks and the Turner Parallel Walls, and surrounded by several other archaeological sites.  The Turner Workshop collection was gathered in the 1960s and 70s by Jim and Clyde Theler.  Unlike most amateur collectors the Thelers collected all material, not just the high value diagnostics and complete specimens.  The collection is dominated by blades and blade-cores with 584 blades, 6,319 blade fragments, and 333 blade-cores.  It is the largest known Hopewell blade/blade-core collection from a single site.

            A blade is an elongated flake.  Blade production makes use of removal face topography to maximize cutting edge per unit volume of raw material.  Standardized morphology is a necessary consequence of a blade prod tech.  To document the presence of a blade technology at a given site requires the presence of a blade-core, or a prepared nucleus of raw material that bears evidence of systematic removal of elongated flakes.  A blade will, by definition, bear the scars of a prior series of removals on its dorsal surface.  These prior removal scars, or facets, will be parallel to the axis of force for the artifact.  The lateral margins of the blade are expected to be roughly parallel to each other and the axis of force.  While some definitions require a 2:1 L/W ratio, not all products of a blade industry will exhibit this characteristic.

The Problem

            This project sought to answer the question: Did the Hopewell Blade industry employ craft specialists?  There are several definitions of craft specialization, but that put forth by Michaels (1989:142) is preferred here.  The aspects of his definition that are most important to this paper are: standardized process, mass production, restricted access, and exchange.

            The products of a specialist are expected to be more highly standardized than those of non-specialists.  However, standardization in isolation is not sufficient evidence for the archaeological identification of specialization (Yerkes 2003).  There are at least 2 ways in which production specialization limits the amount of variation introduced into the final products.  First, by limiting the number of producers (Perlès and Vitelli 1999:96) inter-producer variation in minimized.  Second, a high rate of production facilitates motor habit mapping (Roux 2003:770); thereby minimizing intra-producer variation. 

The research hypothesis of this project is summarized in the following sentence: “The Ohio Hopewell blade industry does not...represent a specialized craft production system (Nolan 2005:36).”

Testing the Hypothesis

The research hypothesis was tested in two ways: first I conducted an intra-Hopewell comparison to determine whether the products of different groups of Hopewell knappers were statistically distinguishable from one another.  This comparison was a test for differences in location among the Turner Workshop site and the several “Localities” in and around the Liberty earthworks.  This collection is one of the largest that has been analyzed in detail and reported (1620 blades and fragments, 159 blade-cores). Second, I compared relative variability within the Turner Workshop sample against that of a collection from a known specialized context (Tula) and a non-specialized context (Joss).  Aside from the fact that the level of production specialization for these two sites is known, these collections were selected for comparison because of there large sample sizes and the fact that their metric attributes had been recorded.

            The alternative statistical hypothesis for the Turner Workshop/Liberty comparison was non-directional with an α = .001.  This stringent α was used for this and all subsequent tests due to the large samples sizes available for comparison.  For the blade-core weight comparison I obtained a sample of 107 blade-cores from the Robert Phillipi workshop at Flint Ridge.  This comparison utilized a One-Way ANOVA with the Null and alternative Hypothesis shown below:

H0: μTW = μLib

H1: μTW ≠ μLib

α = .001


H0: ΤTW = ΤLib = ΤFR = 0

H1: at least one Τ ≠ 0

α = .001

For the relative variability comparison among different industries I utilized the CV adjusted for bias.  CV* is the sample standard deviation divided by the sample mean multiplied by a correction factor (Sokal and Braumann 1980).  For the Tula/Turner Workshop comparison the alternative hypotheses were directional as I expect that Turner Workshop represents a lower-level of specialization than Tula, and therefore Tula will exhibit lower CV*s:

H0: CV*TW = CV*Tula

H1: CV*TW > CV*Tula

α = .001

t∞ = 2.88


For the comparison with Joss I have no a priori reason to suspect that the Turner Workshop will be more variable than Joss, or vice versa, thus the alternative hypothesis is non-directional and the critical t-value is greater for this comparison than that of the previous comparison:

H0: CV*TW = CV*Joss

H1: CV*TW ≠ CV*Joss

α = .001

t∞ =3.291


The Intra-Hopewell comparison utilized a pooled t-test for the attributes of length, width, thickness, platform width, platform thickness, and a One-Way ANOVA for the blade-core weight.

            The formulas for the statistical tests for differences in relative dispersion were taken from Sokal and Braumann (1980).  The t-test is modified as shown here:

CV*1 – CV*2

Sv*1 –Sv*2


The standard error of the difference is computed as shown below:



√(Sv*12 + Sv*22)


            The results of the Intra-Hopewell comparison are shown in Table 1.  The last row shows that all of the attributes from these two samples are different from each other at a .001 level of significance.



H Len

NH Len








































p <







Table 1.  Comparison of blade metrics: Turner Workshop, Liberty.


            The ANOVA for blade-core weight among the three sites (Flint Ridge, Liberty, Turner Workshop) demonstrates that discarded blade-cores are highly significantly different from one another (Table 2).  An a posteriori Bonferroni test shows that Turner Workshop is significantly different from both of the other sites, while they are indistinguishable from each other (Table 3).



Sum of Squares


Mean Square





















Table 2.  ANOVA for differences in weight among Phillipi, Liberty, and Turner Workshop blade-cores.


















Table 3.  Blade-core weight Bonferroni results.


            When comparing CV* from Tula and Turner Workshop the Null hypothesis was not rejected for length and TbBOF, but was handily rejected for the remaining 3 attributes (Table 4).






































p <**






Table 4.  Comparison of CV* for blade attributes: Turner Workshop, Tula.


            The Joss/Turner Workshop relative variability comparison results include a little bit of everything (Table 5).  Like the previous comparison, the length CV* is not significantly different between these 2 sites.  Interestingly the Joss CV* is nearly identical to the Tula CV* (22.63). CV*s for the attributes of width and thickness are lower for Joss than Turner Workshop, and both of these diff are highly significant.  In fact, the Joss width CV* is lower than the Tula width CV* (26.16), and the Joss and Tula thickness CV*s are again nearly identical (33.68, 33.57, respectively).  Thus, it seems that length, width, and thickness may not be sensitive indicators of level of craft specialization in all blade industries.  The most intriguing result is that of the platform angle comparison between Turner Workshop and Joss.  The CV* for this attribute at both of these sites could not be distinguished from each other statistically.  Remember, this attribute for the Tula/Turner Workshop comparison resulted in the largest t-statistic computed for this entire project (24.32).
































p <





Table 5.  Comparison of CV* for blade attributes: Turner Workshop, Joss.


            Other evidence for the level of specialization represented by the Ohio Hopewell blade industry is found in the blade-core shapes.  Blade-core shape provides direct evidence of the reduction process followed for a given production event.  If the process were standardized within and among sites of an industry, variation in blade-core form would be limited.  This is not the case for the Ohio Hopewell blade-cores.  Greber et al. 1981 name nine distinct shapes for the Liberty blade-cores, in addition to several blade-cores that didn’t fit into any of their shape categories.  In addition to the variation represented by the different shapes I discovered in my analysis of the Turner Workshop blade-cores that there are several possible ways to end in any given shape present in the Turner Workshop sample.

            This situation is at least somewhat similar to that for the Dorset blade industry, where multiple blade-core shapes have been documented.  It is also a distinct contrast with the Tula blade industry where there is only a single blade-core shape.  The lack of variability in Tula blade-core shapes is indicative of the high level of process standardization present at the site as described by Healan (2002, 2003).

Another source of evidence of level of specialization in a blade industry is the amount of errors present in the finished products, specifically the frequency with which the blade producers achieve the presumed desired termination type: feather.  The frequency of errors reflects in part the level of standardization of the process; if the actions involved in the production of an artifact do not change within and among production episodes motor-habit mapping is facilitated, thus the knapper is less likely to commit an error.  The Turner Workshop blade producers failed to reach the desired end result 43% of the time, while the Tula craftsmen committed a knapping error only 9% of the time (Figure 1).  This draws another stark contrast between the Post-Classic craftsmen of Tula and the Ohio Hopewell blade producers.

Figure 1. Error frequency comparison: Turner Workshop, Tula.

            It appears that the Hopewell blade industry as a whole was not concerned with minimization of errors to the same extent as the craftsmen of Tula.  It would appear from the small collection of sites shown in Figure 2 that the Turner Workshop blade producers were more skilled than most Hopewell blade producers with the only success rate over 50%.



Figure 2. Hopewell blade error frequencies. Note: a) Greber et al. 1981, b) Ruby 1997.


            In summary, the Hopewell blade industry products are highly variable in continuous and categorical attributes.  The Turner Workshop is variable especially in platform attributes.  There is no evidence of production process standardization.  The process varies from site to site as evidenced by blade-core weight, and within each site as evidenced by blade-core shape.  The success rate of individual blade removals is much lower than expected if specialists were involved in the production of this class of artifacts.  This indicates low levels of production process standardization and low levels of skill involved in the Ohio Hopewell blade industry.  In conclusion, it appears that craft specialists were not employed in the Ohio Hopewell blade industry.



Works Cited

Greber, N’omi, Richards S. Davis, and Ann S. Dufresne

1981    The Micro Component of the Ohio Hopewell Lithic Technology: Bladelets.  In The Research Potential of Anthropological Museum Collections, edited by A. E. Cantwell, J. B. Griffin, and N. A. Rothschild, pp. 489-528.  Annals of the New York Academy of Sciences 376.  New York.


Healan, Dan M.

2002    Producer Versus Consumer.  In Pathways to Prismatic Blades: A Study in Mesoamerican Obsidian Core-Blade Technology, edited by K. G. Hirth, and B. Andrews, pp. 27-36.  The Costen Institute of Archaeology, University of   California, Los Angeles.

2003    From the Quarry Pit to the Trash Pit: Comparative Core-Blade Technology at Tula, Hidalgo, and the Ucareo Obsidian Source Region.  In Mesoamerican Lithic Technology: Experimentation and Interpretation, edited by K.G. Hirth, pp. 153-169.  The University of Utah Press, Salt Lake City.


McGhee, Robert

1970    A Quantitative Comparison of Dorset Culture Microblade Samples.  Arctic Anthropology VII(2): 89-96.


Michaels, George H.

1989    Craft Specialization in the Early Plostclassic of Colha.  Research in Economic Anthropology, Supplement 4 pp. 139-183.


Nolan, Kevin C.

2005    The Ohio Hopewell Blade Industry and Craft Specialization: A Comparative Analysis. Unpublished Master’s thesis, Kent State University.


Odell, George H.

1994    The Role of Stone Bladelets in Middle Woodland Society.  American Antiquity 59(1): 102-120.


Parry, William J.

1994    Prismatic Blade Technologies in North America.  In The Organization of Prehistoric North American Chipped Stone Technologies, edited by P Carr, pp.   87-98.  International Monographs in Prehistory, Ann Arbor, MI.


Perlès, Catherine, and Karen D. Vitelli

1999    Craft Specialization in the Neolithic of Greece.  In Neolithic Society in Greece, edited by P. Halstead, pp. 96-107.  Sheffield Studies in Aegean Archaeology, 2.  Sheffield Academic Press Ltd., England.


Pi-Sunyer, Oriol

1965    The Flint Industry.  In The McGraw Site: A Study in Hopewellian Dynamics, edited by O. H. Prufer, pp. 60-89.  Cleveland Museum of Natural History.


Roux, Valentine

2003    Ceramic Standardization and Intensity of Production: Quantifying Degrees of Specialization. American Antiquity 68(4):768-782.


Ruby, Bret J.

1997    The Mann Phase Hopewellian Subsistence and Settlement Adaptations in the Wabash Lowlands of Southwest Indiana.  PhD. Dissertation, Indiana University.


Sokal, Robert R., and Carlos A. Braumann

1980    Significance Tests for Coefficients of Variation and Variability Profiles. Systematic Zoology 29(1):50-66.


Yerkes, Richard W.

1990    Using Microwear Analysis to Investigate Domestic Activities and Craft Specialization at the Murphy Site, a small Hopewell Settlement in Licking County, Ohio.  In The Interpretive Possibilities of Microwear Studies, edited by B. Gräslund, H. Knutsson, and J Taffinder, pp.167-176.  AUN 14 (Uppsala,   Societas Archaeologia Upsaliensis).

1994    A Consideration of the Function of Ohio Hopewell Bladelets. Lithic Technology 19(2): 109-127.

2003    Using Lithic Artifacts to Study Craft Specialization in Ancient Societies.  In Written in Stone: The Multiple Dimensions of Lithic Analysis, edited by P.N. Kardulias and R. W. Yerkes, pp. 17-34.  Lexington Books, Lantham, MD.


Latest News

by Kevin Schwarz on May 10, 2021
by Eric Olson on April 30, 2021
by Al Tonetti on February 11, 2021