Summary of Project Results
Christopher Carr, Professor
Department of Anthropology
Arizona State University
Tempe, AZ 85287-2402
This project had three broad goal. First was to identify the nature of the surface materials that are found on Hopewell copper artifacts and that possibly were used in the process of creating works of art on the artifacts. Second was to identify the artistic technical processes, themselves. Third was to develop a systematic, integrated set of digital photographic techniques for effectively recovering, enhancing, and displaying the works of art.
Through the work of a team of eleven researchers, who specialize in the areas of archaeology, remote sensor systems, digital image photography and enhancement, applied metallurgy, mineralogy, petrography, paleoethnobotany, and prehistoric textile analysis, each of these three goals were achieved. The following specific accomplishments and professional contributions were made with NCPTT funding.
(1) Preservation. A large number of Hopewellian copper artifacts were preserved for the artistic imagery rendered on them and the surface materials found on them through the taking of ultra-high resolution, color, near-infrared, and midrange-infrared digital photographs of them.
(a) Color Digital photographs. Ultra-high resolution (3360 x 2253 pixel) color digital photographs were taken of 219 sides of Ohio Hopewellian copper items bearing artwork or thought to bear artwork. Of these, 122 sides were of breastplates, 22 sides were of headplates, and 75 were of celts. The items come from a diversity of sites (11) dispersed over south-central to northeast Ohio and represent a range of natural and archaeological formation and preservation processes. In total, the items also bear the full range of kinds of inorganic and organic materials known to occur on Ohio Hopewellian copper items. The sample of items photographed is approximately 10% larger than that proposed.
(b) Infrared photographs. Near-infrared (.715 – 1.1 microns) and mid-range infrared (1.0 – 1.8 microns) digital photographs were both taken of 263 sides of the copper items, including all of the 219 sides photographed in color. The greater breadth of the infrared sample reflects the decision to photograph more than the proposed number of item- sides that do not seem to bear much or any indications of artwork, and to explore the power of IR in revealing artwork essentially not visible to the naked eye. Examples include copper surfaces that appear largely uniform in their corrosion, copper surfaces that are entirely hidden by a uniform textile wrapping or textile-pseudomorph wrapping, and intensely burned (cremated) copper surfaces. The sample of items photographed is approximately 30% larger than that proposed.
(2) Identification and Mapping of Inorganic Materials. The mineralogy of inorganic surface materials was identified, described, and mapped exhaustively for 19 sides of 11 copper breastplates, 5celts, and 3 headplates, and in more focused regions for an additional 25 sides of 21 breastplates, 2 celts, and 2 headplates. A wide range of copper corrosion minerals and other relatively fine-grained minerals were observed, including: cuprite, malachite, azurite, chrysocolla, turquoise, hematite, and hydroxyapatite (bone, calcined). Some minerals had multiple, distinct kinds of morphological variants. Other materials found on the artifacts include soil, mica, mother-of-pearl, calcite or aragonite, a powdered bone and calcite pigment, a powdered talc/serpentine applique, and ash. In all, 21 distinct kinds of surface materials were observed. Identifications of these substances were made by 10X and 30X binocular microscopy, as well as by reference to areas of copper artifacts known by previous chemical and physical determinations to bear these materials.
Only one item-side (B031A) was found unequivocally to evidence painting as the means by which art works were created on the copper artifacts. A yellow pigment probably comprised of bone and carbonate (identified by chemical assays done prior to the NCPTT research) was evidenced by six macroscopic and microscopic characteristics: its homogeneous, very fine grain; its lack of chemical interaction with the underlying copper substrate, causing it to flake off; its unnatural flat top surface; shrinkage cracks within it; possible drying lines within it; and its limit to a well-defined area with sharp boundaries. The item also bears a white serpentine or talc substance that was applied as a powder rather than a pigment in a liquid vehicle, within well constrained lines. Three other items – B052A, B053A, and C023A – had a yellow material that is chemically similar to that on B031A and that probably was the same paint, but its nature as a paint could not be confirmed by its visual properties. C023 also bore hematite powder that was well bounded and that was probably applied, either as a paint or powder.
Copper minerals (corrosion) were the minerals found most commonly on the artifacts. Malachite was most common, followed by azurite. Materials that occurred on most artifacts and that occupied a moderate to a large portion of the area of each include stalactite-form malachite, cauliform malachite, malachite grown through an organic material forming a pseudomorph, malachite grown after cuprite, stalactite-form azurite, iridescent cuprite, and cuprite in situ.
Art works on the items that did not bear paints or powdered pigments were most probably created by patination. Several observations support this conclusion. First, almost all the inorganic materials within areas thought to represent artistic compositions are copper corrosion products with natural patterns of growth. Second, no evidence was found for particular kinds of corrosion having grown unintentionally under particular kinds of paints, nor was corrosion growing through paints observed. Third, soil that adhered to the objects seldom had corrosion growing into it. This suggests that the corrosion developed prior to burial of the objects in the ground. In contrast, in experiments made by Pimentel and Carr reported above, soils that served as vehicles for applying acids to copper plates to produce patinas had corrosion that grew through them and that made them difficult to remove from the patinas. Fourth, half of the copper items surveyed for organic remains on them had textiles, and additional items had other absorptive materials on their surfaces. Patinas may have developed naturally under such materials if they had been applied to the copper artifacts as decorative cutout images– cutouts are a well-developed form of Hopewell art – or if they had been applied with acids to the artifacts in order to intentionally produce patinas of those shapes on the artifacts. Fifth, 32% ( n = 6) of the 19 item-sides thoroughly examined had corrosion with a mesa-like growth habit, suggesting their formation underneath a growth-limiting boundary, such as organic cut-out shapes that were applied for decoration and unintentionally produced patinas, or that were applied soaked with acid in order to in intentionally produced patinas. Sixth, several artifacts exhibited “drying lines” or “meniscus lines” that would have developed around the edges of acid or water-soaked organic cut outs. Seventh, no sharp boundaries were found between corrosion minerals of different kinds, suggesting that corrosion was not gathered as a pigment, mixed with a liquid vehicle, and painted on the artifacts. Where boundaries did occur, they were more diffuse and related to meniscus effects. Eighth, one item – B050 – bore an apparently complex artistic composition and had an equally complex mineralogy, with 9 distinct kinds of surface materials that were spatially distributed to form the art work. Several other items had 7 or 8 distinct kinds of inorganic surface materials, supporting their artistic rather than natural origin – one would expect only one or a few kinds of corrosion to develop naturally on an artifact of limited size. Nineth, no significant differences were found in the kind of corrosion products that occurred on opposite sides of an item, which might have been the case if the corrosion were natural and different moisture and other corrosive conditions existed above and below the item in the ground.
In general, the three different classes of artifacts – breastplates, celts, and headplates – had similar ranges of minerals on them. However, certain exceptions possibly point to different kinds of artistic patination having been intentionally produced on different kinds of artifacts used in different social and ritual contexts. Azurite was absent from the celts examined, but occurred on two-thirds of the headplates and one-third of the breastplates studied. Cauliform malachite occurred twice as frequently on breastplates and celts (two- thirds of them) than on headplates (one-third of them). Malachite bubbles (perhaps from the burning of patinated items when they were decommissioned) occurred on two-thirds of the celts examined but on none of the headplates and breastplates studied. Malachite pseudomorphs of organic materials were found slightly more often on breastplates than on celts and head plates.
No inorganic surface materials beyond burnt bone significantly correlated with items bearing cremated remains. This suggests that art works made by different processes were not distinguished from one another in their ritual form of decommissioning.
(3) Identification and Mapping of Organic Materials. The organic surface materials on the entirety of both sides of 77 copper artifacts, including 59 breastplates, 14 celts, and 4 headplates, totalling 154 sides, and from seven sites, were identified with stereobinocular microscope from 7 – 30X and then mapped, for comparison to their color and infrared photographic responses. The sample is about 30% larger than that proposed, which was made possible by the team paleoethnobotanist contributing extra effort without cost to the grant.
Eleven general categories of organic materials were identified on the artifacts: textiles composed of the yarns from Group 1 fibers (herbaceous plants such as Indian hemp or milkweed, for example), leather/hide, feathers, fur, bark, carbonized wood, uncarbonized seeds, bast, monocot stems, cut-up plant stems, a plant-fiber plaster of a kind, and unknown organics. Textiles, leather, and wood charcoal were the organic materials most commonly observed, with up to 40 or 45% of the surveyed copper objects from a site having them. Feathers and bark were about half as common, and fur about a third as common. Organic materials turned out to be much more common on the copper artifacts than had originally been thought, which can be related in part to some of them likely having been used in the process of making art works on copper, by collage or patination, as described below.
Two copper artifacts and possibly a third were found to have traces of paints on their textiles: red and olive green in one case (B034),, and red in one or two others (B044, B079). The olive green pigment may have been an example of a paint made from gathered copper corrosion; green paint is known to have been applied to other, free examples of textiles.
Different kinds of copper artifacts vary in the kinds of surface organic materials that they bear. Breastplates had organic materials of all kinds more often than celts.
Breastplates were especially favored with textiles, twice as commonly as celts. Different sites varied in the kinds of organic materials that commonly occurred on their copper artifacts. Seip had greater percentages of copper artifacts with Group 1 plant textiles, bark, and wood charcoal. The Hopewell site had greater percentages of copper artifacts with leather/hide and monocotyledon leaf fragments. Edwin Harness mound had greater percentages of artifacts with feathers. These differences in the materials found on the artifacts by artifact type and by site probably reflect, respectively, first the different ritual functions and social contexts of use of the different artifact types and the somewhat different manners in which they consequently t types were decorated, stored, and/or decommissioned, and second, the different rituals of decoration, storage, and decommissioning practiced by different, neighboring Hopewellian communities. Differences in the frequency and means of copper patination among artifact types and among communities is one possibility.
Most sides of most breastplates bear a diversity of organic materials, rather than only one kind. Of 60 breastplates examined in detail or scanned, 52 (86.7%) had two or more distinct kinds of materials. This pattern is in line with the observation that organic materials appear to have been arranged on copper items as collages, to produce artistic compositions.
In all, a broad range of explanations can be offered for the observed occurrences and spatial distributions of organic materials on the surfaces of Hopewellian copper artifacts. These possibilities, which are not mutually exclusive, include: (a) cutouts of textile, hide, fur, and/or feathers that were directly and intentionally applied to the artifacts to form collages; (b) cutouts of the same materials soaked with mild acids and applied to the copper artifacts to produce patinas in those shapes; (c) textiles, hide, fur, feathers, and/or plant masses that were spread over a copper artifact and then removed in places by sanding or carving them away to produce positive or negative images; (d) small flowers, seeds, cut up stems, pearls, shell beads, or cremation remains that were intentionally placed on the objects during decommissioning rituals, forming imagery or not; (e) feathers that were arranged in various directions over the entire surface of an item to create an artistic composition; (f) textiles, hide, fur, feathers, and/or plant masses that were layered on top of a copper artifact during a decommissioning ritual, but not with the intent of creating imagery; (g) textiles, hide, textiles with feathers possibly attached, and hide with feathers possibly attached, which were used to wrap copper artifacts with patinas for decommissioning and burial, followed by the differential preservation of the materials in locations of different patinas; (h) textiles, hide, fur, and feathers that were components of clothing and against which copper artifacts with patina were laid during burial, followed by the differential preservation of the materials in locations of different patinas; (i) textiles, hide, or textiles with attached fur or feathers that were used to wrap plain copper artifacts without patina images; (k) textiles, hide, fur, and feathers that were parts of clothing and against which copper artifacts without patinas were laid during burial; (l) large expanses of hide or fur that were placed over copper artifacts and painted; and (m) any of the above processes combined with differential preservation or erosion in the ground or during cleaning. Of these multiple ways in which organic materials may have come to be present and preserved on Hopewellian copper artifacts, none was common in the sense of pertaining to the majority of the154 artifact sides examined.
(4) Identification and Mapping of Textiles and Cordage. The textiles and cordage on 132 sides of 66 copper breastplates, celts, and headplates were identified for the structural properties of their yarns and/or weaves with a 7X stereozoomscope and mapped for their locations on the items. This sample is smaller than that proposed, by 68 item-sides, because a good percentage of textiles and cordage on items in the Ohio Historical Society collections had been converted to pseudomorphs comprised of corrosion. This condition prevented firm structural observations. In addition, the textile identification process turned out to be more time consuming than originally planned.
Textile constructions were limited to four kinds: oblique interlacing, spaced 2-strand twining, alternate pair twining, and spaced alternate pair twining. Textiles were found on breastplates and celts with approximately equal commonality (approximately half the artifacts of each type examined) and not at all on headplates in the Ohio Historical Society. (One headplate at the Ross County Historical Society had much cordage on both sides.) Most copper artifact sides with fabrics present on them (48 of 68; 71%) had fabrics on less than 50% of their area. Very rarely did fabrics cover 80% of the area of a side. One-third of the 66 artifacts had textiles on both sides, one-third on only one side, and one-third had no textiles.
Textiles could, in the abstract, have come to occur on the copper artifacts in several ways: (a) by the artifacts being wrapped in a cloth or placed in a bag; (b) by the artifacts having lain against clothing or shrouds worn by the deceased; (c) by textiles having been cut out into shapes that were applied to the artifacts to decorate them; and (d) by textile cutouts having been soaked in weak acid and applied to the artifacts to create mineral images through copper patination. The idea of artifacts wrapped in bags is not relevant to most of the copper artifacts examined because most do not have textiles on both sides of them. The same is true for the idea of the artifacts having lain against clothing. In addition, neither of these ideas is supported by the small area of a copper artifact that textiles typically cover. The latter fact provides greater support for the idea that the textiles were cutouts applied to the artifacts for decorations or to create copper patinas in their shape.
(5) Quantitative Modeling of Corrosion Processes. Ten Pourbaix thermodynamic quantitative models of corrosion development were created. The models allowed an evaluation of whether the several to many mineral species of copper corrosion found on single copper breastplates, celts, and headplates likely developed naturally after burial in soil or, instead, were copper patinas that were induced intentionally by Hopewellian artists to create artistic compositions. The models show that only one compound will be chemically stable at a given pH and electrical potential, and that thus it is unlikely that more than one copper corrosion compound or at most two will form on a given copper artifact naturally. If chemical environmental conditions and equilibria change over time, the previously formed corrosion will become unstable and a new corrosion will form in its place. Reasonable explanations of the diverse copper corrosion minerals found on the copper artifacts, from a chemical corrosion perspective, include painting with corrosion pigments or patination. Examination of the surface materials on the artifacts showed that painting seldom occurred, leaving intentional patination as a logical explanation.
(6) Examination of Breakage Patterns of Copper Artifacts for Intentional, Standardized Means of Decommissioning Them from Their Ritual and Social Uses. In the course of detecting imagery on the copper breastplates, it was noticed that some broken ones (n = 49) were similar in shape to the images of animal and human heads applied to such plates by patination, painting, and/or collage. Many of the breaks occur at odd angles that do not appear natural. It was hypothesized that these breastplates were decommissioned by purposefully breaking them into specific, culturally-preferred forms. This practice had already been witnessed by C. Carr in other media, including fabrics, stone bifaces, and shell artifacts.
To explore this possibility, a sample of 20 of the 49 plates was selected for further study, including 16 breastplates and four headplates. The edges of each selected item were examined by microscope at 10 – 20X to determine whether the broken edges had as much corrosion development as the unbroken edges, or whether they appeared to be fresh breaks (made at the time of excavation or thereafter). The kinds and amounts of corrosion on the broken edges were also compared to the kinds and amounts on copper plates that had been mechanically or electrolytically cleaned of corrosion during their early curation. The latter served as a baseline for defining “recent” corrosion, i.e., that which might develop between the time of excavation of the artifact and the present.
The microscopic examination showed that most of the broken edges of the 20 plates were well-covered with cuprite and/or malachite and were very old, with the strong likelihood that they corroded in the soil and were produced prior to burial. It is concluded that breastplates and headplates were sometimes decomissioned by breaking them into specific animal and human head shapes that were culturally important.
(7) Innovations in Image Capture Methodology.
(a) Photomosaicing. Curved objects, including copper breastplates that were curved in length and usually width, as well as a few large celts with curved tops and sometimes bottoms, posed special photographic problems. The goal of creating a flat layout of each object with minimal distortion from parallax error could not be achieved with one photograph. To create a flat layout, multiple digital photographs were taken at different points along the objects curvature, perpendicular to it. Operationally, this was achieved for headplates by keeping the digital in one position and rotating the headplate about is approximate center of curvature on a styrofoam support so that each photographed section wasorientedparalleltotheplaneoffocusofthecamera. Celts,havinglesscurvature,were repositioned more subtly between photographs. The photographs of the series for each artifact were then spliced together to form a single, flat layout using Adobe Photoshop’s image scaling, rotating, skewing, and stretching routines – a procedure from satellite imaging called photomosaicing.
(b) Lighting. Photography light stands commonly orient two lights at 180o from each other and 45o from the object plane. This tends to remove shadows on an object that are produced by its surface texture. To the contrary, we wished to document the relief of the surfaces of artifacts photographically because surface texture is an important diagnostic characteristic of the organic and inorganic materials on the artifacts and, thus, critical to the definition of art works comprised of these materials. To include information on material relief in the digital photographs taken, without shadows from relief overwhelming the photograph and the recording of material color, we oriented both lights at 45o from the object plane, as usual, but oriented the two lights at 135o from each other: 90o from the length of the object on its “west side” and 45o from the length of the object on its “east side.” This lighting orientation produced shadows from relief features that were oriented in essentially any direction on the object. The orientiation of the lights 45o from the object plane ensured that much of the color of the object was also visible and not obscured in shadow.
(8) Qualitative Evaluation of Digital Photographic Spectra and Digital Enhancement Procedures for Discriminating among Surface Materials and Discerning Artworks.
(a) Color Image Processing. Each of the 219 color digital images of the copper artifacts was prepared in Adobe Photoshop for image processing by changing its background to a uniform grey, and adding a rule of standard format. Contrast enhancements of two kinds were made for each photograph: a total histogram stretch, in which the histogram of all three bands of the RGB image were stretched at the same time, and an individual color band histogram stretch, in which the histogram of each band was stretched separately. The first contrast enhancement maintained the normal color balance of the image, which usually emphasized the green and/or blue channels, whereas the second contrast enhancement gave more equal balance to all three channels and de-emphasized the green and/or blue channels compared to normal. Both kinds of contrast enhancements considerably improved the visibility of art works on the artifacts. A third routine, histogram equalization, seldom proved effective in this regard, and was not pursued for the entire set of photographs.
Seven kinds of calculations of only two of the three color bands (R x B, R x B- inverse, R-inverse x B, R-inverse x B-inverse, B x G, B-inverse x G-inverse, and B – G) were made for both of the two kinds of contrast enhancements just discussed. These band calculations, of the many possible, were the ones found most effective on a trial-and-error basis in distinguishing and clarifying features in the art works. All seven of the bands were found important, in that different bands were helpful in distinguishing different kinds of organic and inorganic materials and, thus, different features within the art works.
(b) Infrared Image Processing. Each of the 263 near-infrared photographs and each of the 263 mid-range infrared photographs were enhanced in their contrast by a simple histogram stretch, using Adobe Photoshop. The images were sharpened with a sharpening filter before contrast stretching them. This order of the two operations was found to give images of slightly better contrast than the reverse.
The near-infrared and midrange infrared photographs both clarified the definition of art works on the copper artifacts, but different bands were more or less effective for different artifacts, apparently depending on the materials comprising their art works. In general, the near-infrared photographs corresponded more closely to the visible light, RGB photographs in the images that they revealed than did the midrange-infrared photographs, as was expected. The midrange-infrared photographs thus provided more information independent of the visible light bands than did the near-infrared photographs – a finding substantiated by the discriminant function analysis made of all five visible and infrared bands (see below).
(c) Hybrid Color and Infrared Images. Hybrid, color-infrared images were created in the GIS program, IDRISI, by replacing the R, G, and/or B channels of a photograph with a near-infrared band, a midrange-infrared band, and/or a calculated color band or occasionally a calculated color and infrared band. Hybridizing the digital photographic bands required that the three visible bands be reduced in their resolution closer to the resolution of the infrared bands, and that the three visible bands and the two infrared bands be resized and registered in relation to each other. These image modifications were necessary because the color, near-infrared, and midrange infrared images were captured with different sensor systems. Registration was accomplished in IDRISI, usually with a linear, rubber-sheeting transformation anchored in 15 points of correspondence among images. Occasionally, a quadratic transformation was required.
(d) Evaluation of Bands, Band Calculations, and Hybrid Band Combinations and Combination Procedures for the Effectiveness in Revealing Artistic Compositions. In general, the ease with which artistic compositions on the artifacts could be defined decreased from the Red to the Blue to the Green bands, and from the near-infrared to the midrange-infrared bands. Band calculations ranged in their effectiveness from high to low as follows: R x B-inverse and R-inverse x B-inverse, to B x G-inverse and B-inverse x G inverse, to R x B, to B x R-inverse and B – G. In general, R x B images looked very similar to the original RGB image and provided little new information, whereas other bands often brought out features not as visible or not visible in the RGB image. RxB-inverse usually looked very similar to B x G-inverse but often provided somewhat better image contrast. Likewise, R-inverse x B-inverse usually looked very similar to B-inverse x G-inverse and commonly produced better image contrast. The B – G band generally was grainy, and sometimes could not be contrast-stretched to give an image that wasn’t very dark and discriminating. The Red and NIR bands and combinations of these bands with other colors gave the artistic composition their greatest visibility.
Working with hybrid, color-infrared images allowed the exploration of the effectiveness of a number of operations in revealing the artistic compositions. First, no one or few bands or band combinations were always or usually effective in improving the clarity of the art works. For each different artifact, different bands were found optimal and were selected for use. Using information from all five bands was not found to be a good, general strategy, because this approach muddied the clarified information in some bands with the grosser information in others. Second, the same art work was often best explored with multiple combinations of bands; different features of a composition were brought out by different bands, presumably depending on the materials of which they were comprised. Third, it was found essential for effective display of the art works to place the most discriminating band in the R channel, the next most discriminating band in the G channel, and the third most discriminating band in the B channel. Third, unsupervised clustering gave clearer renditions of artistic compositions than did supervised clustering when logical units within a composition were internally heterogeneous or spiky. Fourth, cluster analysis helped to define artistic compositions better than palette boundary redefinition of a single- band image when logical units within a composition were internally heterogeneous and spiky. Palette boundary redefinition was found preferable when logical units were internally homogeneous. Fifth, palette blending was found better at rendering artistic compositions when their logical units were internally heterogeneous, whereas a palette defined with crisp boundaries among colors was found to better at defining compositions with internally homogeneous logical units. Sixth, cool and neutralized blue and yellow colors were found optimal color palettes for displaying images and for avoiding eye-jarring color contrasts. Seventh, principle components classification of pixels using multiple color and/or infrared bands was not found as helpful in clarifying artistic imagery as was cluster analysis or palette redefinition, when one or a few bands providing good definition of the art could be found initially. Principle components analysis was more helpful in clarifying art works when only bands with poor to moderate image definition were available.
(9) Developing a Quantitative Model for Identifying Surface Materials and Discerning Artworks. The color and infrared electromagnetic spectral differences among 52 kinds of surface materials that are found on Hopewellian copper artifacts and that often comprised elements of their artistic compositions were characterized and modeled digital photographically and statistically. The materials include cloth, leather, feathers, pigments, patinas, and various minerals. The spectral bands examined encompassed red, green, blue, near-infrared, and midrange infrared. Digital photographs of ten copper artifacts provided the samples of the materials and the source of the spectral bands. The goal of this work was to build a quantitative model that allows an assessment of two things: (a) the degree to which the various materials can be distinguished from each other by their color and infrared spectra and, thus, the artistic compositions on the artifacts can be clarified photographically, and (b) the particular portions of the electromagnetic spectrum that are most useful for making such discriminations.
The degree to which materials can be distinguished from each other photographically was estimated univariately and multivariariately. First, simple, single spectral band line plots were made of the spectral responses of the 52 materials for each of the five spectral bands. These plots show which materials are more or less alike in the R, G, B, NIR, and MIR responses. Second, the Euclidean distances between all pairs of the 52 materials were calculated, based on their “distances” from each other on the R, G, B, NIR, and MIR bands. The matrix documents which materials are more or less alike in the multispectral responses. The relationships among materials in the matrix could be summarized approximately by a multidimensional scaling plot or cluster diagram, although this was not done. Third, a canonical discriminant function statistical model based on all 52 materials and all five spectral bands was built. The model is very successful, in that it is capable of identifying all materials and discriminating them each other 89.9% to 100% of the time, that is, for 89.9% to 100% of the spectrally sampled pixels of a material. A Box M statistical test, Wilks Lambda, and the Chi-square statistic all indicate strong differences among the spectral responses of the materials and their distinguishability. The least well discriminated materials were primarily slight color or textural variants of a given general class of materials, such as malachites of somewhat different color or lighter and darker azurites. Three more serious misidentifications made occasionally by the model include: (a) organic hide and a light to medium olive green variant of malachite; (b) gum and a pine green variant of malachite; and (c) a dark variant of azurite and a pine green variant of malachite. Overall the results of the discriminant function analysis suggest that most features of artistic composition comprised of any of the 52 materials should be discernable with color and infrared digital photographic methods.
Of the five spectral bands used to build the discriminant function model, two were found most helpful in distinguishing among the 52 kinds of materials: green and midrange- infrared. These two bands dominated the first two discriminant functions, which together accounted for 92% of the brightness variability of the materials that distinguished them. The red and near-infrared bands were also found to be important to distinguishing among materials. These two bands correlated with and contributed significantly to three of the five derived discriminant functions – more functions than any other spectral bands correlated with. The blue band was found to have the least discriminating power. These generalizations about the utility of particular bands, of course, are based on all 52 materials simultaneously. A given pair of materials might be most distinguished by the blue band: for example, light and blue azurite patinas, or azurite and malachite. The need to tailor digital photographic enhancement procedures to the particular artifact rather than a large set of artifacts in general must be emphasized.
(10) Reconstruction of Hopewellian Artistic Methods for Creating Images on Copper. The mineralogical and chemical analyses of the copper artifacts, supported by the inventory of organic materials on them, have led to a reworking of our team’s understanding of how imagery was made on the copper objects. At the time of writing of the grant proposal, it was thought likely that spatial patterning in both the copper and non copper minerals was the result of painting with copper and noncopper pigments in some organic vehicle. The research team has, indeed, confirmed through microscopic examination that a few artifacts clearly have deposits of ground pigment particles on them – hematite, serpentine, and hydroxyapatite. Azurite and other copper corrosion minerals may also have been painted onto copper very occasionally. However, it appears that the great majority of images comprised of copper corrosion minerals were probably produced by patination, either intentional or incidental in nature.
Specifically, we suggest that some artifacts probably bear patinas that were produced intentionally by techniques common in copper artwork today. These techniques include (1) painting corrosive solutions in the form of images or parts of images directly on the copper artifacts, perhaps with the application of heat to accelerate the corrosion process before the maker’s eyes; (2) applying shaped cutouts of various organic materials (e.g., textiles, hide) to the copper artifacts and soaking the cutouts with corrosive solutions to produced patina images under the cutouts; and (3) applying naturally particulate or ground organic materials (e.g., seeds coats, ground up wood or other plant material) in a pattern on the copper artifacts and soaking the covered area with corrosive solutions to produce patina images under the covered area. The likely production of some patinas by these particular methods was brought to our attention by a master copper worker (David Pimentel, Professor and Chair of the Jewelry and Metals Department, School of Art, Arizona State University).
It is also possible that some of the patinas were incidentally produced. They may be a byproduct of the intentional application of decorative organic cutouts onto the copper surfaces, followed by natural patina development under the organic cutouts or paints, unintended for the eye. . The organic cutouts or paints then may have deteriorated, delaminated, and been lost, or been removed during excavation and/or museum curation.
The empirical observations that lead us currently to suggest patination development as a mechanism by which the images were produced and/or preserved are the following. (1) Most instances of copper corrosion minerals in image-bearing areas of the artifacts have natural mineral growth patterns, or natural growth patterns modified somewhat by the occurrence at one time of growth-restricting surface materials of some kind (organic materials?) over the corrosion. (2) Despite the natural growth patterns of the corrosion minerals in particular locals, there are three features of the corrosion surfaces that are clearly unnatural, cannot be explained by fully natural corrosion processes as we understand them, but can be explained by the intentional or incidental patination processes described above . These features are: (a) the occurrence of relatively sharp linear and curvilinear boundaries between areas of different corrosion minerals; (b) the homogeneity of the corrosion minerals within those bounded areas; and (c) the great diversity of kinds of corrosion minerals present on some single artifacts, despite their restricted size, compared to what normally would occur on a single piece of float copper in nature. (3) Organic surface materials on the artifacts were found to have affected the kinds of corrosion minerals that developed on them. (4) Organic materials of one kind or another occur on the copper artifacts in much higher frequency than previously thought, perhaps as high as 80% of the specimens. (5) Examples have been found of textiles that were cut into recognizable, repeated shapes, and pasted/laid on the copper artifacts. (6) Examples exist of free-standing Ohio Hopewellian fabrics that were used for mortuary canopies and/or shrouds and that were decommissioned by cutting them into shapes of the same kinds found on the copper artifacts, at sites where the copper artifacts also occurred.
To test the hypothesis that intentionally or incidentally made copper patina images were produced on Hopewellian copper artifacts, Pimentel and Carr made 42 test copper plates with 15 test patches each, documenting various combinations of salts, acids, and acid absorbant materials. All the materials are easily found in nature and were accessible to Hopewell Native Americans. The salts include table salt, crushed gypsum, crushed shell, and bone mean, as sources of corrosive chloride, sulfate, and carbonate, and phosphate ions, as well as crushed copper corrosion as seed crystals. For acids, pure red grape juice, apple juice, citric acid, cranberry juice, apple vinegar, white vinegar, urine, and three strengths of tannic acid as found in acorns we used. For the absorbant materials, we used hide, textiles, and feathers, which were found on the Hopewellian copper artifacts. In two weeks at room temperature, the test plates developed thick patinas of cuprite, malachite, chrysocolla, turquoise, and azurite – the minerals that predominate on Hopewellian copper artifacts. Corrosion impressions of feathers and textiles, which occur commonly on the artifacts, were also produced. Finally, a cuprite patina image of a bird-man creature, as found commonly on Hopewellian copper artifacts, was reproduced by cutting out its silhouette in textile, moistening it with a salt-acid mixture, covering it, and letting it stand for two weeks. The cuprite grew from the substrate naturally, but had crisp edges not usually found in natural corrosion, as in the case of the artistic compositions on Hopewellian copper artifacts.
This detailed, final report to the NCPTT has also been sent to and archived in the Ohio Historical Society, Columbus, OH.
(b) Publications and Professional Presentations Made.
Carr, Christopher, A. Lydecker, D Pride, S. Hoffman, J. Colwell, & J. Mitchell. 2000. Digital Photographic Preservation of Artworks on Ohio Hopewell Copper Artifacts. Paper presented at the annual meeting of the Society for American Archaeology, Philadelphia, April.
Wimberly, Virginia. 2001. Preserved Textiles on Ohio Hopewell Copper Artifacts: Clues for Artifact Relationships. Paper presented at the annual meeting of the Society for American Archaeology, New Orleans, April.
Wymer, DeeAnne. 2001. Organic Preservation on Ohio Hopewell Copper Artifacts: New Insights into Middle Woodland Ritual. Paper presented at the annual meeting of the Society for American Archaeology, New Orleans, April.
Wymer, DeeAnne. 2001. Organic Preservation on Prehistoric Copper Artifacts of the Ohio Hopewell. Paper presented at the annual meetings of the American Anthropological Association, 100th Anniversary, Washington, D.C., November.
Barron, Jeffrey W., and Christopher Carr. 2002. Hopewell Mortuary Artifact Analysis using Spectral Indices & Disciminant Analysis. Poster presented at the Mid-West GIS Conference, Kansas City, MO., April.
Carr, Christopher, Andrew D. W. Lydecker, Edward Kopala, Jeffrey S. Nicoll, Jeffery A. Colwell, Steven M. Hoffman, John Mitchell, Ann Yates, David Pimentel, Duane Simpson, and Jeffrey Barron 2002 Technical Studies of Artworks on Ohio Hopewell Copper Artifacts. Paper presented at the annual Midwestern Archaeological Conference, Columbus, OH, October 3.
Wymer, Dee Anne. 2002. The Value of Archival Collections: Organic Preservation on Hopewell Copper Artifacts. Paper presented at the annual Midwestern Archaeological Conference, Columbus, OH, October 3.
Wimberly, Virginia. 2002. Hopewell Fabrics: Evidence from Copper Artifacts. Paper presented at the Annual Midwestern Archaeological Conference, Columbus, OH, October 3.
Barron, Jeffrey W., and Christopher Carr. 2002. Hopewell Mortuary Artifact Analysis Using Spectral Indices and Disciminant Analysis. Paper presented at the annual meeting of the American Anthropological Association, New Orleans, November
Bernardini, Welsey, and Christopher Carr. 2005. Hopewellian Copper Celts from Eastern North America: Their Social and Symbolic Significance. In Gathering Hopewell: Society, Ritual, and Ritual Interaction, ed. by C. Carr and D. Troy Case, pp. 624-647; esp. pp. 626-631. Kluwer Academic Press, New York.
Wimberley, Virginia. 2005 Preserved Textiles on Hopewell Copper. In Perishable Material Culture in the Northeast, ed. by Penelope Ballard Drooker, pp. 69-85. University of the State of New York, The State Education Department, Albany, NY.
Wymer, DeeAnne. 2005. Organic Preservation on Prehistoric Copper Artifacts. In Perishable Material Culture in the Northeast, ed. by Penelope Ballard Drooker, pp. 69-85. University of the State of New York, The State Education Department, Albany, NY.
(c) Masters’ Thesis
Barron, Jeffrey W. 2005. A Determination of Hopewell Mortuary Artifact Composition Using GIS Modeling, Spectral Indices, and Discriminant Analysis. Unpublished Masters’ thesis. Department of Anthropology, University of Arkansas, Fayetteville, AR. January.
(d) Archiving of Project Images and Notes.
Compact disks of each of the 219 raw color images, 263 raw near-infrared images, 263 raw mid-range infrared images, and many hundreds of trial enhanced color images and enhanced hybrid color and infrared images have been made and archived at Arizona State University. The laboratory notes and maps produced by the mineralogist, plant and animal materials specialist, and textile specialist, who are members of this NCPTT research team, have been duplicated and archived with the Ohio Historical Center, Columbus.
(e) Media Announcements.
In late summer, 2000, the project was announced to the press and a full page Sunday cover story was published in the Columbus Dispatch, Columbus, OH. An interview about the project, for release to local and national presses at its closure, has been given to the Arizona State University Press Media Office.