Archaeological Survey - Wyoming State Archaeologists Office

Ceramic Petrography

Thin-section microscopy has proven most useful in the analysis of archaeological ceramics. Yet, not all pottery is suitable for thin-section analysis. There is a common misconception that petrography is used to identify clay. On the contrary, standard thin-section, polarized light microscopy is used to identify sand-sized inclusions (.063 mm – 2 mm) in the clay body. Coarse silt (.03 mm – .063 mm), though visible is frequently too small to identify with confidence and everything smaller than .03 mm is thinner than the thin-section and therefore not visible. Since the analysis is incumbent on the identification of sand-sized inclusions it is essential that the sample submitted contains enough inclusions to provide a statistically viable sample particularly when resource provenance is the goal of the analysis. If the ceramic assemblage is characterized by sparse inclusions then the size of samples to be thin-sectioned must be as large as possible, around 22 mm x 42 mm. This requires that the sample be cut parallel to the vessel surfaces rather than perpendicular as is typically done. We will cut each sample to the optimal size for a nominal fee (50 ¢). Please contact us if you would like to have a ceramic sample analyzed but are unsure whether it is appropriate for a resource provenance study.

Some ceramic traditions are characterized by crushed shell, grit (crushed rock) or grog (crushed pottery) temper. Typically there are sands naturally occurring in the paste, but not necessarily. In such circumstances it may be prudent to first have a ceramic thin-section analyzed prior to submitting sediment samples for a provenance study. For instance, a ceramic sample can be thin-sectioned and a Basic Paste Characterization analysis conducted. The analysis will identify the type and volume of temper/inclusions present as well as other useful technological variables of the sample and provide data as to the feasibility of a resource provenance study.

Thin-section microscopy is a destructive analytical technique. Many curation facilities as well as federal and state agencies have regulations or guidelines pertaining to destructive analysis. It is the client’s responsibility to ensure that all regulations and guidelines are followed when submitting archaeological material for thin-sectioning.

Resource provenance studies are not suitable in all regions due to homogenous mineral compositions of sediments. Please review the Information on Resource Provenance Studies to ascertain whether a provenance study is appropriate for the assemblage in question. In cases where sand petrofacies have already been identified in a region, such as southern Arizona there is no need to submit sediment samples as the results of published studies can be used for provenance determinations. Likewise, our office has samples from many drainages in Wyoming, eastern Colorado and western Nebraska. Please call ahead to see if we already have a sample.

At least one, and preferably two or more sediment samples are required if you would like us to evaluate whether a ceramic sample was manufactured using locally available sediments. If the goal of the analysis is to develop a petrofacies model of a region or locale then 10 or more sand samples from multiple adjacent drainages is typically required. As with all statistical analyses, the larger the sample the higher the confidence one can place in the results. Please review the Resource Provenance Studies section for more information on petrofacies modeling.

Sediment samples should consist of sand from a fluvial source, such as a sandbar or wash, within the same drainage and as close as possible to the site from which the pottery was recovered. In order to obtain a representative sample, dig a trench two to four meters wide and 20-30 cm deep or several holes of the same depth and place the excavated sand on a tarp and mix thoroughly by hand. Once the pile of sand is well mixed screen a couple liters through a 2 mm sieve (No. 10) to remove gravel and granule making sure to collect all the sediment that passes through. Next screen the sample through a 63 µm (No.230) to remove silt and clay. Submit approximately ½ liter of the sand along with UTM coordinates and a map of the sample location. A sample of clay may also be useful if naturally occurring clay deposits are present in the area. A 50-60 ml sample will suffice. Again please submit the UTM coordinates and a map of the sample location. If you do not have a set of standard geological sieves we will process the sample for a $5 fee. Please call ahead if you have any questions or concerns.

Ceramic samples should be individually bagged, labeled, wrapped in shock-absorbent packing material and placed in a box. Please ensure that the package is labeled “Fragile.” Sediment samples should be clearly labeled and placed in a sealed container, or in two (double bagged) ziplock-type bags. Please include the Petrographic Services Order Form, a map showing the location of the site and/or sediment samples as well as a table or list of the UTM coordinates.

Works Cited and Suggested Reading

Chayes, Felix

1954 The Theory of Thin-Section Analysis. Journal of Geology 62:92-101.

Chayes, Felix and H. W. Fairbairn

1951 A Test of the Precision of Thin-Section Analysis by Point Counter. The American Mineralogist 36:704-712.

Dickinson, William R.

2001 Petrography and Geological Provenance of Sand Tempers in Prehistoric Potsherds from Fiji and Vanuatu, South Pacific. Geoarchaeology 16:275-322.

Dickinson, William R. and Richard Shutler, Jr.

1979 Petrography of sand tempers in Pacific Island potsherds: Summary. Geological Society of America Bulletin Pt. 1, 90:993-995.

Donahue, Jack, David R. Watters and Sarah Millspaugh

1990 Thin Section Petrography of Northern Lesser Antilles Ceramics. Geoarchaeology 5:229-254.

Dye, Thomas S., and William R. Dickinson

1996 Sources of Sand Tempers in Prehistoric Tongan Pottery. Geoarchaeology 11:141-164.

Ferring, C. Reid and Timothy K. Perttula

1987 Defining the Provenance of Red Slipped Pottery from Texas and Oklahoma by Petrographic Methods. Journal of Archaeological Science 14:437-456.

Green, G.N.,

1992 The Digital Geologic Map of Colorado in ARC/INFO Format: U.S. Geological Survey Open-File Report 92-0507. Electronic document, http://mrdata.usgs.gov/geology/state/state.php?state=CO, accessed December 2013.

Heidke, James M., Elizabeth J. Miksa and Henry D. Wallace

2001 A Petrographic Approach to Sand-Tempered Pottery Provenance Studies. In Ceramic Production and Circulation in the Greater Southwest: Source Determination by INAA and Complementary Mineralogical Investigations. Edited by Donna M. Glowacki and Hector Neff. Costen Institute of Archaeology, Monograph 44. University of California. Los Angeles.

Josephs, Richard L.

2011 A Petrographic Analysis of Glenwood Locality Ceramics, Mills County, Iowa. Central Plains Archaeology 13

Lintz, Christopher and Kathryn Reese-Taylor

1997 Migrations, Trade, or Replicated Ceramics: Petrographic Study of Collared Rim Sherds from the Texas Panhandle. Bulletin of the Texas Archeological Society 68:273-300.

Lombard, J

1987 Provenance of Sand Temper in Hohokam Ceramics, Arizona. Geoarchaeology 2: 91-119.

Lovering, T.S. and E.N. Goddard

1950 Geology and Ore Deposits of the Front Range Colorado. Geological Survey Professional Paper No. 223, Washington D.C.

Middleton, A.P., I.C. Freestone and M.N. Leese

1985 Textural Analysis of Ceramic Thin Sections: Evaluation of Grain Sampling Procedures. Archaeometry 27:64-74.

Miksa, Elizabeth J., Mary F. Ownby and Carlos P. Lavayen

2012 Petrographic Analysis of Pottery from Honey Bee Village. In Life in the Valley of Gold: Archaeological Investigations at Honey Bee Village, a Prehistoric Hohokam Ballcourt Village. Edited by Henry D. Wallace. Desert Archaeology. Electronic document, http://www.desert.com/analysis/ceramicp.html, accessed January 2014.

Ownby, Mary F.

2012 Petrography of Great Basin Brown Ware from the Yamashita Sites, Moapa Valley, Southern Nevada. Desert Archaeology, Petrographic Report No. 2012-02. Submitted to the University of Nevada Las Vegas. Copy available as electronic document http://www.desert.com/analysis/ceramicp.html, accessed January 2014.

Page, Michael K.

2009 The High Plains Upper Republican Reconsidered: Stylistic and Petrographic Analyses of Central Plains Tradition Ceramics from the High Plains. Unpublished M.A. thesis, Department of Anthropology, University of Wyoming, Laramie.

Page, Michael K., and Charles A. Reher

2013 A Petrographic Analysis of Dismal River Micaceous Pottery: Products of Southwestern Trade or Local Production? Paper presented at the 71^st Annual Conference of the Plains Anthropological Association. Loveland, Colorado.

Solomon, M.

1963 Counting and Sampling Errors in Modal Analysis by Point Counter. Journal of Petrology 4:367-382.

Stoltman, James B.

1991 Ceramic Petrography as a Technique for Documenting Cultural Interaction: An Example from the Upper Mississippi Valley. American Antiquity 56:103-120.

Tweto, Ogden

1979 Geologic Map of Colorado. Geologic compilation cartography by R.E. Schoenfeld. Department of the Interior, United States Geological Survey prepared in cooperation with the Geological Survey of Colorado, Reston, Virginia.

Van Der Plas, L. and A. C. Tobi

1965 A Chart for Judging the Reliability of Point Counting Results. American Journal of Science 263: 87-90.

There are two commonly used methods of analyzing ceramics in thin-section. Modal analysis, or standard point counting, is preferred by many because it allows the modes (volumetric percentage) of identified rocks and minerals to be calculated with precision (Chayes and Fairbairn 1951; Chayes 1954; Heidke, Miksa and Wallace 2001; Lombard 1985; Middleton et al. 1985 ). In standard point counting a two- dimensional equidistance grid is first established, typically at an interval equal to or greater than the diameter of the largest inclusion in the thin-section. This procedure prevents single grains from being counted multiple times and further allows the reliability of the results to be calculated (e.g. ±5%) at a 95% confidence interval (2s) using the Van Der Plas and Tobi (1965) method. This is the technique used for our Basic and Advanced Ceramic Paste Characterization services (see Petrographic Services).

The second technique, or rather group of related techniques, is grain frequency counting, of which there are three approaches (Dickinson 2001; Middleton et al. 1985). The first is areal counting where a thin section is gradually moved below the crosshair eye-piece and all grains that fall completely within the field of view are identified, measured and counted. Second, is the ribbon (traverse) method where all inclusions that pass completely within a predefined transect, measured by the reticle in the microscope eyepiece, are size-graded, identified and tallied. Lastly is line counting where all grains that are intersected by the horizontal reticle are measured, identified and counted. Frequency counts do not produce volumetric estimates of grain proportions, but are quickly performed and comparable (Dickinson 2001). Moreover, ceramics with sparse and/or large (>.5 mm) inclusions cannot be economically analyzed by point-counting because the total number of identified inclusions is too small to allow for statistical comparisons. For instance, only 200 points can be counted on a thin-section of a sherd that has a maximum inclusion size of 2 mm. If the sample has a large volume of inclusions, say 25%, then only 50 inclusions could be counted for an optimal 20 mm x 40 mm sized thin-section. Similarly, a sample with only 5% very-fine to medium-sized inclusions would require a grid spacing of .5 mm and 2,000 counts in order to identify only 100 inclusions. Given these limitations, traverse counting is our preferred method for resource provenance studies.

Both techniques produce similar and replicable results for well-sorted sands. However, results of the two methods may differ significantly for poorly sorted sands (Dickinson 2001; Middleton et al. 1985), a problem that can be obviated by a research design focusing solely on certain grain sizes. Other methodological parameters can be altered, such as simply ignoring large grains during modal analysis allowing more points to be counted on a thin-section, or using a grid that is not equidistant. We encourage prospective customers to contact us with questions and concerns and assist us in devising an appropriate research design.

There are two basic approaches to “sourcing” archaeological ceramics through thin-section, polarized light microscopy (Heidke et al. 2001). The simplest method relies on the identification of one or more geographically restricted, easily distinguishable rocks or minerals. This approach, referred to as the “key grain” method (Dickinson and Shutler 1979; Heidke et al. 2001), though potentially effective and economical is only practicable where rare rocks or minerals are present. Examples of rare rocks and minerals in the Rocky Mountain west include anorthosite from the Laramie Anorthosite Complex in southeastern Wyoming (Page 2009, Stanley 1976) and leucite from southwestern Wyoming and north-central Montana, respectively.

The second approach to “sourcing” archaeological ceramics is the petrofacies method. This method is predicated on the principle that sand, or detritus, is composed of weathered or eroded particles of parent rock that have entered into a sedimentary transport system such as a stream. The mineral and lithic composition of sand within any stream basin reflects the mineral and lithic composition of the parent material from which it was derived. The petrofacies method was originally devised to identify and differentiate sandstones. It was quickly realized that the same principles that allow for the identification of sandstone petrofacies could be used to determine the provenance of sand-tempered pottery. Dickinson, (2001 and citations therein) in a series of studies conducted on archaeological ceramics from the South Pacific, showed that the provenance of sands within pottery can be determined with a relative degree of confidence. Lombard (1987), a student of Dickinson, later applied the petrofacies method to the Tucson Basin of southern Arizona where about 30 petrofacies have been identified in a relatively small area (Miksa et al. 2012). Unfortunately, despite the utility of the approach relatively few modern sand petrofacies have been systematically identified for archaeological applications. Lombard (1987) and Heidke et al. (2001) provide a thorough and useful outline of the process of creating a petrofacies model. Please contact us if you are interested in or would like assistance with developing a sand petrofacies model.

OWSA is currently working to build petrofacies models for the Front Range of Colorado, southeastern Wyoming and western Nebraska (Page 2009; Page and Reher 2013). Initial results indicate that there is patterned compositional variability between and within the North Platte, South Platte, and Arkansas River basins, as well as between and within several first and second order streams within the North and South Platte River drainages (Page 2009; Page and Reher 2013). As our database grows it may soon be possible to assign pots from eastern Colorado and Wyoming and perhaps Kansas and Nebraska to specific petrofacies.

The petrofacies approach is not suitable for all regions or ceramic traditions. Portions of the Central Plains of Nebraska and Kansas, for instance, have homogenous mineral compositions resulting from millions of years of lateral migration of a few large streams that drain the central Rockies. Sands in these regions are comprised predominantly of monomineralic quartz and feldspars grains with relatively few distinguishable lithic grains. The presence of key grains can in certain circumstances be used to determine where a sample did not originate (Page 2009), but specific provenance areas are not typically assignable. Similarly, grit (crushed rock), grog and shell tempered ceramics may not contain enough sand to allow for a provenance determination. Or, in the case of grit tempering the mineralogy of the rock that was crushed may mask the underlying mineral composition of the naturally occurring sands present in the paste.

Even in the absence of defined petrofacies models it is possible to glean valuable information from petrographic analyses (Ferring and Perttula 1987; Josephs 2011; Lintz and Reese-Taylor 1997; O’Malley 1981; Ownby 2012; Stoltman 1989, 1991). For instance, Page (2009) found that only 22.8% of a sample of 35 Central Plains tradition sherds from four sites in southeastern Wyoming was produced using sands that were locally available. The remaining 27 samples in Page’s study could not be assigned to a petrofacies but they could be identified as extralocal. Moreover, the results of the study called into question several hypotheses regarding the settlement pattern and nature of the Central Plains tradition (ca. AD 1000 – 1400) occupation of the High Plains (Page 2009). Similarly, another study using 19 Dismal River (protohistoric Plains Apache) sherds from one site in Nebraska (n=14) and four sites (n=5) in southeastern Wyoming revealed that pottery was produced from a variety of sources within the North and South Platte River basins (Page and Reher 2013). Again, the provenance of the samples could not identified with confidence, but the presence of pots not produced from locally available resources could be determined. Furthermore, the study undermined an often cited assertion that all micaceous pottery on the Plains was imported from the southwest (Page and Reher 2013). In short, the utility of petrographic studies is not limited to areas with defined petrofacies.