5 DISCUSSION

The results of the many analytical techniques performed for this study are summarized in tables 2–8. The only red pigment identified was red iron oxide. Green pigments were identified as the basic copper carbonate, malachite, and the copper silicate, chrysocolla. Although the malachite was sometimes used alone, chrysocolla was found only in combination with other pigments. Azurite, another basic copper carbonate, was identified as the blue pigment, but both blue paint samples also contained significant amounts of chrysocolla and the hydrated yellow iron oxide, limonite. Yellow paint samples were identified as primarily limonite. Black pigments were identified in all but one case as being charcoal black. The white ground layers were in all cases calcite-based, while the cream-colored ground was calcite, clay, or a mixture of the two materials. It should be noted that varying degrees of calcareous material were detected in almost all of the pigment and ground samples.

According to Shepard's study, the calcite found in most of those pigment samples may represent the addition of slaked lime that has recarbonated upon exposure to the atmosphere, hence binding the pigments (Shepard 1946). It is possible that lime washes were intentionally mixed with the pigments, a method somewhere in between the painting techniques of true fresco and a secco. Since the calcite crystals can grow in the last stage of evaporation, they would already be at contact points between the mineral grains, and very little calcareous material would be needed (Torraca 1982). With further study, the actual nature of the binding mechanism may become apparent. It is quite possible that the addition of gum binders may not have been necessary.

While many questions have been answered in this study concerning the materials and manufacture of stuccoed tripod vessels, the identity of the artisans still has not been confirmed. As stated earlier, some scholars have proposed that stuccoed tripod vessels were first constructed and fired in ceramic workshops and then decorated as a subspecialty of the mural painting workshops (Evans and Berlo 1992). Conides (1997) believes it more likely that ceramic artisans incorporated stucco decoration into their repertoire. Supporting evidence for this theory includes comparison of decorative motifs. Conides has found much stronger similarities between the execution of some design elements on both stucco and planorelief decorated vessels than with murals. Examples that she cites include feathers, hands, and circular forms. For instance, hands are depicted on both ceramic forms as rectilinear and with unbending fingers, a representation not typical of mural paintings.

Since this study showed that the use of cinnabar is not as diagnostic as previously thought in distinguishing ceramic artisans from mural painters, sherds that were identified as having clay grounds were further investigated. The use of clay as a ground has been found in examples widely distant in Mesoamerica (Shepard 1946). Clay base coats are also very similar in concept to the application of decorative slips on the surfaces of vessels. Their presence on some but not all of the sherds in this study could represent a bridge between the traditional techniques of potters and the development of applying stucco to ceramics. The examples of pigmented calcite ground are very similar in appearance to clay but likely would have been even more durable. Perhaps they represent an improvement in technology, while still maintaining visual continuity.

Examination revealed that more of the roughened sherds are, indeed, associated with calcite-based cream grounds than with clay-based ones. These data are in no way conclusive, however, especially since the ceramic surface is not visible on three sherds. As well, more of the nonroughened sherds also are associated with calcite grounds. This finding could be partially explained by the fact that not all vessels after 550 A.D. were roughened. Stucco was still applied in some cases over vessels with other completed decorative finishes (Conides 1997). Clearly, 10 sherds do not constitute a large enough sample size for a correlation to be made. This question would make an interesting topic for further research.