Setting the Stage

The Maxwell Museum's copy of the Mimbres Photo Archive was used by many researchers, including Marit Munson. Munson's research (published in Reference Munson2000; see also Munson and Hays-Gilpin Reference Munson, Hays-Gilpin, Nelson and Hegmon2010) explored issues of gender and status as represented in the Mimbres paintings. She presented a multistaged analysis (described in more detail below) that allowed her to identify several gender-specific activities.

Hegmon teaches Southwest archaeology and often used Munson's article as an example of clearly presented analysis. McGrath was an undergraduate student in one of those classes. As an exercise to learn about research analysis, he duplicated Munson's procedures and applied them to part of the MimPIDD dataset, using only vessels with at least site-level provenience that are housed in public institutions. McGrath found fewer markers of gender and fewer gendered activities, and his overall results did not support Munson's conclusion about the clear division of labor. This article explores the reasons for these different results and their broader implications.

The Two Analyses

The analyses, done by Munson and then duplicated by McGrath, used identical procedures but were applied to different datasets. They involved a multistep process of identifying sex, gender markers, and then activities. Here we explain each step and the different results reached in each case; the details are summarized in Table 1.

Table 1. Summary and Comparison of the Two Analyses Done by Munson and McGrath.

Databases. Munson considered all bowls in the Mimbres Photo Archive (ca. 6,500), including those professionally excavated and those with no provenience in private collections. McGrath drew on the larger MimPIDD collection but considered only bowls with at least site-level provenience owned by public institutions. Currently (September 2016), MimPIDD has more than 10,500 bowls and 3,776 fit McGrath's criteria. There is some overlap between the two samples in that the professionally excavated bowls in Munson's sample would also have been in McGrath's sample.

Munson found 170 bowls (2.6 percent of her sample) that depict humans, for a total of 326 human individuals (Reference Munson2000:130). McGrath found 104 (2.7 percent) bowls depicting humans for a total of 139 human individuals. Neither author discussed chronological control, but since representational designs in general and depictions of humans in particular are most often found on what is called Style III pottery associated with the Classic period (Shafer and Brewington Reference Shafer and Brewington1995), both analyses likely included the same chronological period.

Sex. The depictions of some individuals included sexually explicit genitalia, breasts, or beards. Munson found that sex could be determined for 123 (38 percent) of the depicted individuals in her sample, including 78 males and 45 females, for a male:female ratio of 1.7. McGrath could determine sex for 43 (31 percent) of the depicted individuals in his sample, 32 males and 11 females, for a ratio of 2.9. The difference in the frequencies of sexed individuals is statistically significant (χ ² = 25.98, p < .05), although the difference in the sex ratios is not.

Gender Markers. The next step was to examine the sexed individuals to identify gendered traits. Munson identified a series of traits that might be gender-specific and found 440 instances of them, including 177 on sexed individuals. Four were associated primarily or exclusively with males (eye masks, hair knots, plaited hair, and head feather) and three were associated with females (string apron, head band, and leggings). For example, MimPIDD 2104 (Figure 1) shows two males, each with a horizontal stripe across the eyes, which Munson called an eye mask. She used these results to identify a total of 136 men and 68 women, as well as 12 individuals who had traits of both genders. McGrath's analysis with the smaller more controlled sample found 187 instances of these gendered traits, 56 on sexed individuals. He found support for only some of the gender-specific traits identified by Munson (eye mask, hair knot, and head feather on males, string apron on females). McGrath identified a total of 50 men, 21 women, and two individuals with traits of both genders. The difference in the gender ratios and the different frequencies of individuals with mixed gender traits are not statistically significant.

Figure 1. MimPIDD 2104 from Swarts Ruin. Note that the human figures are anatomically male and they wear eye masks, thought to be associated with men. © President and Fellows of Harvard College, Peabody Museum of Archaeology and Ethnology, PM# 24-15-10/94502 (digital file# 99090022).

Gendered Activities. Finally, both analyses considered the activities of the gendered individuals they identified, which Munson called “gendered roles and statuses” (Reference Munson2000:134). She found a consistent association of certain activities and tasks with gender, “with men more often hunting or involved in ceremonies and women predominately responsible for child care. Women are also depicted carrying burdens with greater frequency than men” (Reference Munson2000: 138). In contrast, McGrath found a less clear division of labor. While he found that men were associated with hunting and ceremonies, his sample revealed no clear association of women with any particular activities.

For example, MimPIDD 2781 (Figure 2) shows a figure that can be identified as a man because he wears an eye mask, and he is depicted as carrying something, possibly a person or large animal. Munson had concluded that carrying was primarily a woman's activity, although McGrath was not able to duplicate that result. The well-known cover of Engendering Archaeology (Gero and Conkey Reference Gero and Conkey1991) shows a Mimbres bowl that depicts a pregnant woman carrying a pronghorn. That bowl (MimPIDD 3898) is in the collection of the Western Colorado Museum, but unfortunately its provenience and excavation/collection history are unknown, so it was not included in McGrath's analysis. It is depicted as Figure 5 in Munson (Reference Munson2000:136).

Figure 2. MimPIDD 2781 from the Galaz Ruin. The sex of this figure is not obvious, but because of the eye mask it is classified as a man. Mimbres bowl, earthenware with slip and pigments, 3¾ × 8¼ in. Collection of the Frederick R. Weisman Art Museum at the University of Minnesota, Minneapolis. Transfer from Department of Anthropology. 1992.22.169.

Summary. Individually, the differences found in each stage of the analyses are fairly minor and few are statistically significant. There is, however, a consistent trend in the nature of the difference, demonstrated by the summary data in Table 1. Of the bowls that depict humans, those in Munson's sample have more individuals and those individuals more frequently (38 percent vs. 31 percent) have sexually explicit characteristics. There are more gender-specific attributes in Munson's sample. As a result, a larger percentage (66 vs. 52) of individuals in Munson's sample are gendered and a larger percentage (5.6 vs. 2.7) have a mix of gendered attributes. Finally, there is a more clear division of labor in Munson's sample, a division that conforms to both cross-cultural patterns and many contemporary expectations.

Although not definitive, these results are suggestive of sample bias. Munson's sample included everything in the Mimbres Photo Archive, including unprovenienced material in private collections, while McGrath's was more controlled. It is likely that the latter is more representative of the overall universe of Mimbres bowls, and our conclusion that there are differences between the samples is not surprising. More interesting is the nature—and directionality—of those differences and the way they might affect archaeological interpretations.

Collection Sources

The collections analyzed by Munson and by McGrath included material gathered for different purposes and in different ways, and it would be surprising if they are not different. Munson's sample included material excavated systematically by archaeologists, as well as various private collections assembled in unknown ways that almost certainly included looting. McGrath's sample included only material excavated systematically and housed in public institutions. Here we explore the nature of those differences by comparing decorated bowls in two subsets of MimPIDD. Called the Controlled and Uncontrolled samples, these were selected to emphasize differences between different kinds of collections. Whereas there was some overlap between Munson's and McGrath's samples, the two samples considered here are entirely different.

The Controlled Sample includes material excavated from known documented contexts and housed in public institutions. Because it includes only documented contexts, it is smaller and even more controlled than McGrath's sample, which included bowls for which there was only site-level provenience. Specifically, the controlled sample includes material from the following sites: Cameron Creek (Bradfield Reference Bradfield1931), Galaz (Anyon and LeBlanc Reference Anyon and LeBlanc1984), Mattocks (Gilman and LeBlanc Reference Gilman and LeBlanc2016), Swarts (Cosgrove and Cosgrove Reference Cosgrove and Cosgrove1932), NAN Ranch (Shafer Reference Shafer2003), Saige-McFarland (Lekson Reference Lekson1990), and Wind Mountain (Woosley and McIntyre Reference Woosley and McIntyre1996). Some of these were excavated carefully over the course of several years; others were salvaged with archaeologists literally working in front of bulldozers. Research goals included understanding the site, its architecture, and its occupational history, and excavations produced pottery from all kinds of contexts. Some of the vessels were restored, and it is possible that paint was enhanced as part of the restoration; however, it is unlikely that restoration involved “embellishing” the designs or creating entirely new designs (the distinction between restoration and embellishment is discussed further below). Thus the controlled sample is assumed to be generally representative of Mimbres pottery overall and to include only authentic designs.

The Uncontrolled Sample includes vessels with no provenience information for which the owner is either unknown or is listed as “private;” many of these were probably looted. Such collections were included in Munson's analysis. Private collections were probably assembled in various ways. Some collectors might have been eager to obtain any and all Mimbres vessels, others might have wanted one of each kind of design, and others might have wanted only the best or the most interesting. Given that the Mimbres tradition is famous for its engaging representational designs, it is likely that these were specially sought by some—perhaps most—collectors. It is also likely that some of the designs were modified to increase their visual appeal for collectors and are therefore not authentic.

In order to examine differences between the two samples, the relative frequency of different kinds of designs is shown in Tables 2 and 3. Only bowls associated with the Classic period (with Style II/III or Style III designs) are included in these tallies. As expected, representational designs, particularly those that include humans, are significantly more common in the uncontrolled sample.

Table 2. Occurrence of Representational Designs in the Controlled and Uncontrolled Samples, Including Only Classic-Period Bowls.

χ ² = 134.6, p < .05

Table 3. Occurrence of Designs that Depict Humans in the Controlled and Uncontrolled Samples, Including Only Classic-Period Bowls.

χ ² = 11.43, p < .05

The results of this comparison between the controlled and uncontrolled samples should serve as a sort of warning for research involving mixed collections. However, the comparison does not fully explain the differences between Munson's and McGrath's results, since their analyses focused primarily on the association of various attributes, rather than the overall frequency of those attributes. That is, having more or fewer representational designs should not affect the association of string aprons with females or hunting with men, if those associations are in fact consistent. Other factors must be at work.

“Embellished” Images: Faking Mimbres Designs

A recent volume (Lazrus and Barker Reference Lazrus and Barker2012) considers the impact of looting on archaeologists' knowledge of the past. Their emphasis is on the Old World, though they also discuss the issue in North America. Most researchers who work with Mimbres pottery are aware of some designs that do not look quite right. However, lacking the kinds of authentication techniques used by art historians, most are also reluctant to put these impressions in print, both because of a lack of certainty and because, as Brody remarks, “in a litigious society it is unwise to publically call anything “fake” without technical evidence” (Reference Brody2004:xvi).

One important study done in the field of museum conservation provides such technical evidence (Lee and Khandekar Reference Lee and Khandekar2012). Their work, done in consultation with Mimbres expert Steven LeBlanc, was prompted by the speculation that “due to the current market value of Classic Mimbres vessels, a large number of authentic vessels have had their original design modified by transforming a simple central element to a more complex figurative design, resulting in a value increase of one to two orders of magnitude” (Reference Lee and Khandekar2012:218). Indeed, at least one Mimbres archaeologist has been told that bowls with “blank” interiors (i.e., with framing lines around the bowl's rim but no marks in the center) are worth a considerable sum because they can be modified by adding new designs. Lee and Khandekar looked for such modification using a technique called pyrolysis-gas chromatography-mass spectrometry, which can detect modern synthetic paint on Mimbres bowls. They studied 25 bowls, five from the Princeton Art Museum and 20 from private collections. The bowls all had designs that appeared, stylistically, to be genuine; however, the bowls’ history, what art historians call “provenance,” was unknown. These bowls were chosen because they were good candidates for having been modified and thus were useful for Lee and Khandekar's research; the sample was not intended to be representative of Mimbres pottery overall.

Lee and Khandekar (Reference Lee and Khandekar2012) discuss what they call “restoration” and “embellishment” although they acknowledge that the distinction is not always clear. Restoration might involve reattaching broken fragments as well as “inpainting” to revive faded images, although restoration is not intended to manipulate designs. Embellished, in contrast, “is used to describe bowls that have had the original imagery manipulated in a deceptive manner” (Reference Lee and Khandekar2012: 219). All 25 bowls they studied were thought to be authentic Mimbres pieces, with at least some authentic designs. The question they asked is whether the designs had been embellished with modern paint. Answering this question involved taking and testing many samples of paint from each bowl, and they specifically focused on possibly suspect designs, including paintings of humans. They concluded that, of the 25 bowls they studied, more than half were possible fakes; seven had “embellished” designs and six more had “questionable” designs.

Replicating Mistakes

The problem of unprovenienced and sometimes fake artifacts is, sadly, well-known in classical archaeology. The Brooklyn Museum recently designed an exhibit intended to reveal that about one-third of its collection of Coptic sculptures is fake (German Reference German, Lazrus and Barker2012). These artifacts are the basis of what German calls the “invention” of Minoan and Mycenaean religion in which the conclusions of early scholars, based on illegitimate collections, were reproduced in later works. There are even suggestions that the invented history might have influenced later fakes, which reproduced and falsely legitimized the invented history. While there is no reason to think that archaeologists’ understanding of Mimbres culture is similarly tainted, there are suggestions that modified paintings may have falsely emphasized and thus effectively invented some aspects of Mimbres art.

One of the bowls analyzed by Lee and Khandekar (Reference Lee and Khandekar2012) is from a private collection and illustrated as their Figure 1 (MimPIDD 9330). It shows four human figures surrounding a table or blanket; they all wear head feathers and thus would be classed as men by both Munson and McGrath. Its configuration is similar to MimPIDD 1308, which was illustrated in Brody (Reference Brody1977:170, Figure 114). Lee and Khandekar focused analysis on these figures “because similar small human figures are known to have been embellished in other bowls on the market,” and they indeed found modern paint in the figures (Reference Lee and Khandekar2012:222). Thus it is possible that whoever modified the painted design on this bowl had seen similar figures on other Mimbres bowls (bowls that themselves might or might not have themselves been modified) and copied them onto this one. It is also possible that the head feather was chosen because it is indicative of “Indians” in popular culture.

McGrath's analysis revealed one intriguing and problematic bowl (MimPIDD 10386) that unfortunately cannot be illustrated here because it is in a private collection. Fortunately, MimPIDD includes photos of this bowl before and after “restoration.” It was a large fragment of about one half of a bowl that showed a human figure with arms from about the chest down. The figure wears leggings, which were found to be associated with women by Munson but not McGrath. The final restored and apparently embellished bowl is whole and depicts an entire human, including head. That figure is wearing an eye mask, which was found to be associated with men. In other words, the embellishment of this figure may have (inadvertently) created a mixed gender figure.