After completing the study of 134 rock art sites in the state, Diaz-Granados concluded that AMS radiocarbon dates were needed to place at least a selection of the rock art sites on a more secure chronological footing. Mississippian iconography in the Southeast has been discussed by several scholars (e.g., Brown 1989; Griffin 1952; Hall 1989; Howard 1953,1968; Knight 1989; Muller 1989; Waring and Holder 1945); however, it is not temporally well defined for Missouri. Because iconographic data, particularly in relation to diagnostic artifacts, are not temporally well defined in this region, six rock art sites were selected for dating.
The first site in the series is Picture Cave, which was chosen for its well-preserved pictographs containing ample amounts of black pigment as well as diagnostic motifs. In addition, Picture Cave features portrayals of story panels and representations of supernatural beings with associated cultural materials and symbolic motifs. Picture Cave is located on a high hill within a remote, wooded, karst area in east-central Missouri. Unfortunately, by the time professional archaeologists were notified of the existence of the pictographs and taken to the site, the cave floor had been seriously vandalized and an undetermined number of artifacts were removed. Judging from the dates inscribed on its walls, Picture Cave has had casual visitation for over a century, with a probable increase in vandalism during recent years as artifact values reached unprecedented levels. Our research team has been able to retrieve only a few small remnants of pottery, bone, and hematite from the surface of the disturbed areas.
A long-nosed god maskette panel at Picture Cave is estimated to date possibly as early as the Developmental Mississippian (A.D. 900-1150) according to Muller (1989:14). These maskettes (Figure 1) are believed to fit into the early expansion period at Cahokia (pre-Southeastern Ceremonial Complex), on the northern fringe of the Mississippian domain (Kelly 1991:73; Williams and Goggin 1956:58-59). However, this motif extends into later time periods. Diagnostic motifs at this cave, particularly the mace, are cited as good temporal markers by Muller (1989:15). He also includes the bi-lobed arrow as a definitive time marker for “Southern Cult” (mid-thirteenth century), and the ogee, both of which are common in Missouri rock art.
Other motifs such as concentric circles (shoulder tattoo or paint), and the falconid eye, are present in the pictographs at Picture Cave and associated with other Southeastern Ceremonial Complex (SECC) motifs. The AMS dates obtained are slightly older than the initial estimated range of A.D. 1050 to 1100 (Diaz-Grana’dos 1993:326). Experimental Procedure Until the results of van der Merwe et al. (1987), no direct dating of charcoal pigments in rock paintings had been reported. Ten years later, charcoal drawings on the walls of this Missouri cave site are believed to be the first in the Mississippi Valley region to be directly dated. With one exception discussed later, these charcoal pigment samples were processed by the plasma-chemical technique (Hyman and Rowe 1997a, 1997b and references therein), permitting direct radiocarbon analysis by accelerator mass spectrometry (AMS), leading to age estimates for the associated drawings. Sample Collection Five small samples of charcoal were taken from the rock paintings (only -. 1 mg of carbon is necessary for an AMS determination of the 14C). Latex gloves were worn during sample collection and subsequent handling. Samples were scraped with new scalpel blades onto clean aluminum foil. The sample’s location on the panel was noted and photographed. Of the five samples taken for processing at the Texas A&M laboratory, three contained sufficient carbon for AMS analysis. These processed samples were sent to the Lawrence Livermore National Laboratory’s Center for Accelerator Mass Spectrometry. The largest sample was split in two to provide us with
The analytical chemistry lab at Texas A&M University used the usual NaOH treatment (ultrasonication for -1 hr at 50 ? 5? C) to remove humic acids from charcoal. HCI treatment, also suggested as standard, is unnecessary with our technique and may even present some problems for dating rock paintings when calcium oxalate is present in the sample (Pace et al. 2000). The lab filtered the NaOH and the sample through binder-free borosilicate glass filters
that had been baked overnight at -600? C to remove organic contamination. Plasma extractions were run on the dried filtrate material on the filter.
The lab used radio-frequency-generated low-temperature (-150? C), low-pressure (-1 torr) oxygen plasmas, coupled with high vacuum techniques, to remove carbon from charcoal in the paint sample leaving the substrate rock and carbonate/oxalate accretions intact (Hyman and Rowe 1997a, 1997b and references therein). The plasma extraction system was cleaned with oxygen plasmas before sample insertion to rid surfaces of organic contamination.
After a sample was loaded into the plasma chamber, the chamber was evacuated to 104 torrs and then
filled with .2 torr ultra-high purity argon (99.999 percent). Low-power argon-plasmas were run on each sample to remove adsorbed CO2 that may have entered during sample insertion. This process works by the inelastic collision of the unreactive, but high-energy, argon atoms and ions with adsorbed CO2 molecules.
The lab repeated argon-plasmas until the amount of carbon desorbed as CO2 was < .001 mg, an amount that has a negligible effect on radiocarbon analysis. Following the final argon plasma for each sample, the system was pumped to _10-7 torr and left pumping for several hours. Vacuum pumps were then closed to the system and any rise in pressure was monitored for an hour or more to evaluate the extent of possible leakage of atmospheric CO2 into the plasma chamber. No significant leaks were found. The chamber was next filled with 1.0 torr ultra-high purity oxygen (99.999 percent), and the sample oxidized in a low-temperature plasma. The gaseous CO2 produced by the oxygen plasma oxidation of the charcoal in the paint samples was collected and its pressure was measured after the water produced in the plasma reaction was removed by freezing. The CO2 was then frozen at -1 940C with liquid nitrogen, sealed into a borosilicate glass tube, and sent to the Lawrence Livermore National Laboratory’s Center for Accelerator Mass Spectrometry for 14C analysis.
Validation of Technique
The validity of the analytical chemistry laboratory’s plasma-chemical technique was assessed by several
means. First, the chemists conducted an isotopic analysis of 14C-free samples (Albertite coal, International Atomic Energy Authority standard wood, Axel Heiberg wood, and commercial graphite) run at the Lawrence Livermore National Laboratory Center for AMS (Ilger et al. 1995:301) and at the Australian Nuclear Science and Technology Organisation (ANSTO) laboratories. Scientists at the ANSTO AMS Centre graphitized CO2 produced in the plasma
oxidation system from Albertite coal, Axel Heiberg wood, and commercial graphite 14C-free materials. Following AMS measurements on the ANTARES accelerator, it was concluded that the contamination added by the plasma system was small compared to the .9 Rtg modem carbon equivalent (Lawson, personal communication 1998). The chemists also compared their dates with samples of previously dated material (archaeological charcoal, Third International Radiocarbon Intercomparison wood, and ostrich eggshell) and found an agreement within statistical uncertainties (Ilger et al. 1995:301).
For the eggshell, the oxygen plasma was used to clean the sample; CO2 was removed for dating by phosphoric acid treatment. In addition, they also compared a series of their dates with archaeologically inferred age ranges; in virtually every case, the validity of the technique was supported (Hyman
and Rowe 1997b:7). And in some instances, we recovered enough of the paint sample to permit replicate analysis (samples 41VV75-29A & B) (Hyman and Rowe 1997a:64), samples 41VV75-37A-F (Hyman and Rowe 1997a:64; Ilger et al. 1996:410), samples 41VV75A-1-6 (Hyman and Rowe 1997a:64; Pace et al. 2000), and finally the duplicate
dates on Drawing #3 described in this paper. Do all these studies support the general validity of this technique with uncertainty?