THERMAL AND ENVIRONMENTAL EFFECTS ON OBSIDIANGEOCHEMISTRY: EXPERIMENTA - PDF document

THERMAL AND ENVIRONMENTAL EFFECTS ON OBSIDIANGEOCHEMISTRY: EXPERIMENTAL AND ARCHAEOLOGICALM. STEVEN SHACKLEYArchaeological XRF Laboratory, Phoebe Hearst Museum of Anthropology, 103 KroeberHall, University of California, Berkeley, CA 94720-3712, USADepartment of Anthropology, 232 Kroeber Hall, University of California, Berkeley, CADO NOT CITE WITHOUT PERMISSION OF AUTHORS ABSTRACTRecent EDXRF compositional studies of thermally altered archaeological obsidian from anumber of late period sites in New Mexico and Arizona suggested that extreme thermalalteration may have been responsible for the depletion of elemental concentrations in themid-Z x-ray region; a region where the most sensitive incompatible elements for thediscrimination of archaeological obsidians reside. A stepped heating experiment subjectingbetween 500° C and 1080°C indicated that at temperatures over 1000°C extreme mechanicalchanges occur, but the elemental composition in the mid-Z region does not vary greatlybeyond that expected in typical instrumental error. It appears that the apparent depletion ofelemental concentrations in the archaeological specimens is due to EDXRF analysis ofsurface regions where melted sands in the depositional matrix become bonded to the surfacebe removed before analysis.KEYWORDS: OBSIDIAN, EXTREME THERMAL ALTERATION, NORTH AMERICANSOUTHWEST, SITE DEPOSITIONAL EFFECTS, ENERGY DISPERSIVE X-RAYFLUORESCENCE Recently, a number of obsidian studies in pre-Classic Salado and Hohokam, as wellas northern Rio Grande contexts have focused on the potential effects of pre-depositional andpost-depositional burning on the trace element chemistry of archaeological obsidian(Shackley 1998a; Steffen 1999a, 1999b). These studies, while informative, were notconducted in controlled laboratory conditions focused on thermal threshold rates todetermine at which temperature, if any, trace element composition my change significantly(c.f. Skinner et al. 1997; Trembour 1990). Our purpose here is to discuss the results of acontrolled laboratory experiment focused specifically on the thermal effects onarchaeological obsidian within a background of archaeological applications in the AmericanSouthwest, and an understanding of thermal gradients in silicic melts. The results presentedhere, of course, are likely applicable anywhere.ARCHAEOLOGICAL BACKGROUNDIn the past few years large scale archaeological projects in Arizona and New Mexicohave, as part of problem domain generation, integrated archaeological obsidian studies intoanalytical research (see Bayman and Shackley 1999; Peterson et al. 1997; Shackley 1995,1999, 2000; Simon et al. 1994). Evident for over 60 years is the periodic and often culturallyproduced pre-depositional and post-depositional burning of obsidian artifacts (Gladwin et al.1938; Shackley 1988, 1990). Cremation, common in pre-Classic Hohokam and Mogolloncontexts is the most obvious vector for the pre-depositional effects, but post-occupationalburning of rooms and entire sites is also responsible for surface modification of obsidianartifacts (Foster 1994). Gladwin and Haury’s excavations at Snaketown in predominatelypre-Classic contexts are the best known studies where cremations were common and artifacts burned to varying degrees (Doyel 1996; Gladwin et al. 1938; Hoffman 1997; Haury 1976;Figure 1 here). Recent analyses of pre-Classic and Classic period burned obsidian artifacts,often projectile points from these contexts, have indicated significant variability in the sourceelement chemistry inconsistent with typical rhyolite glass composition (Cann 1983; Petersonet al. 1997; Shackley 1998a). Analysis of artifacts from burned contexts in Rooms 15 and 16of the Upper Ruin at Tonto National Monument indicated partial to nearly completedepletion of trace elements in three of 19 specimens (Shackley 1998a). All of these TontoRuin specimens, like the Snaketown artifacts, exhibited a thin layer of melted material, likelyfrom the surrounding matrix. As we shall see, this latter attribute is the operative issuehampering reliable trace element compositional studies, not necessarily direct hightemperature effects.THE NATURE OF SILICIC MAGMA COOLING BEHAVIOR AND CHEMISTRYAs a background to understanding both the modal trace element composition of silicicglasses and temperature properties, a slight digressive discussion of melt temperatures will beuseful. Magmas erupted on the earth’s surface are quite hot and dangerously explosive,particularly silicic magmas, so there have been few direct studies (Carmichael et al. 1974).Macdonald and Gibson’s (1969) measurement of the peralkaline obsidian at the Chabbieruption in Ethiopia in 1968 and Carmichael’s (1967) estimates are the most appropriate here(see also Buddington and Lindsley 1964; Table 2 here). These measurements are made withmineral geothermometers using two minerals (usually titanomagnetite and ilmenite) toestimate the liquidus temperature of the silicic lava; by theory the equilibration temperatureof the mineral pair closely approximates the liquidus temperature (Buddington and Lindsley1964; Carmichael et al. 1974:6; Hildreth 1979). Those shown in Table 2 are considered upper limits, and for this exercise the temperature that we would expect to see physical andpossibly chemical changes. Given these data, our initial firing began at 500° C. The processof volatilization and subsequent removal of some compounds such as water and silica isapparently not an intervening variable (Hildreth 1979, 1981).HIGH TEMPERATURE EXPERIMENTAL PROCEDURESThirteen samples from five different obsidian sources in the greater AmericanSouthwest and northwest Mexico were heated and analyzed. All samples weremegascopically aphyric; no megascopically observable phenocrysts. For each obsidiansource, at least two nodules were sampled in order to establish a source differentiationbaseline beyond that previously reported (see Shackley 1995).Thirteen nodules from five known obsidian sources in the Arizona, New Mexico, andnorthern Chihuahua were split to obtain fresh surfaces and avoid contamination duringanalysis (Table 1, Figure 2). The five sources include both peraluminous and mildlyperalkaline lavas in order to attempt to cover the spectrum of trace element variability typicalof silicic glasses (see Cann 1983; Mahood and Hildreth 1983; Hildreth 1981; Shackley1988). Each flake was weighed, measured, optically scanned, and analyzed using EDXRFprior to any heating for baseline comparative data (Table 1). Additionally, for each flake, theanalyzed surface was recorded and all future XRF analyses were performed on the samesurface. The Spectrace 400 instrument used in the Department of Geology and Geophysics iswell reported and instrumental settings and laboratory standards are reported elsewhere (seeDavis et al. 1998; Shackley 1995, 1998b; http://obsidian.pahma.berkeley.edu/tontobs/anlysis.htm). A summary is included in theAppendix herein.HeatingObsidian samples were heated using a Blue Electric Furnace in the Petrography Lab,Department of Geology and Geophysics, University of California, Berkeley. The kiln waslined with ceramic plating and linked to a digital thermometer to accurately monitortemperature. To maintain a constant heating temperature, the kiln thermostat was checkedand adjusted manually throughout each heating session. The same obsidian samples wereheated during each session, and each was weighed and examined for physical changesfollowing heating. After every heating session, flakes were submitted to EDXRF analysis.The samples were subjected to five heating sessions (Step 1 through Step 5) of increasinglyhigher temperatures.The kiln was pre-heated to 500°C, and samples placed loosely on theceramic plate inside the kiln. The kiln was closed and monitored until the temperature againreached 500°C. It took 30 minutes for the temperature to return to 500°C. Samples wereheated at 500°C for one hour. After one hour, the kiln was turned off and the kiln dooropened. Samples cooled inside the kiln for 30 minutes.The obsidian samples were then weighed and analyzed using EDXRF. No weight orchemical changes were detected. Samples were also visually inspected for physical changes.No physical changes were apparent after heating at 500°C.The kiln was pre-heated to 700°C, and samples placed loosely on theceramic plate inside the kiln. The kiln was then closed and monitored until the temperaturereached 700°C. It took 15 minutes for the interior kiln temperature to return to 700°C. Samples heated inside the closed kiln at 700°C for one hour. After one hour, the kiln wasturned off and the door opened. Samples cooled completely inside the kiln.After cooling, samples were again weighed and analyzed using ED-XRF. No weightor chemical changes were detected. Minor physical changes were noted in one sample:Vulture #2. This sample exhibited a band of white discoloration and minor vesiculation on asmall section of the flake’s cortical surface. No other changes were noted.The procedure for Step 3 was identical to that of Steps 1 and 2.Samples were placed loosely on the ceramic plate in the pre-heated kiln. It took 30 minutesfor the internal kiln temperature to return to 800°C. One sample, Burro Creek #2, crackedfrom heat stress when placed on the heated ceramic plate. Samples were heated at 800°C forone hour and then allowed to cool completely inside the kiln with the door open.After cooling, samples were again weighed and analyzed using EDXRF. No weightor chemical changes were detected. Minor physical changes were noted in three samples:Vulture #3 exhibited minor vesiculation and a white discoloration along one edge of theflake. Cow Canyon #1 showed a reddening of residual cortical material on the dorsal surfaceof the flake. The dorsal surface was not analyzed using EDXRF. Antelope Wells #2exhibited melting and vesiculation of cortical material along the flake edge. Again, thecortical surface was not analyzed using EDXRF.Due to thermal cracking of the one sample during Step 3, minorprocedural changes were enacted during Step 4. In Step 4, the kiln was pre-heated to 350°Cand samples were then placed on the ceramic plate inside the kiln and the door closed. Theinternal temperature was then raised to 940°C. It took one hour for the internal kilntemperature to reach 940°C. The samples remained inside the kiln at 940°C for an additional hour. After heating, the kiln was turned off and samples were allowed to cool inside the kilnwith the door closed until the temperature reached 600°C, at which point the kiln door wasopened and the samples cooled completely.Again, samples were weighed and analyzed using EDXRF. No weight or chemicalchanges were noted after heating at 940°C. Upon visual inspection, no additional physicalchanges were noted.In Step 5, samples were placed in a cold kiln to avoid thermalfractures. It took 90 minutes for the internal kiln temperature to reach 1080°C. Sampleswere heated at 1080°C for one hour and then allowed to cool in the kiln with the door closedfor 45 minutes until the temperature reached 600°C. The kiln door was then opened andsamples cooled completely.Severe physical changes were observed in all samples after heating at 1080°C for onehour (Figure 2). Both Antelope Wells samples melted, and all other samples exhibited severevesiculation due to off-gassing. Because of melting and expansion of the obsidian samples,some samples fused together or fused to the ceramic plate inside the kiln, making accurateweight measurements impossible. For the two samples that were not fused with the ceramicplate, Vulture #3 and Government Mountain #1, no weight changes were apparent. Giventhis, it seems reasonable to conclude that no heavy compounds came out of solution due toheating. Chemical changes, as shown through EDXRF, will be discussed below.Only minor physical changes, limited to thin edges and cortical surfaces, wereapparent from heating prior to Step 5 at 1080°C. Heating to 1080°C caused severe physicalchanges to the obsidian samples, quite expectable given the predictive data on silicic magma extrusion temperatures. Minor physical changes began after 700°C in the range of extrusiontemperatures predicted by Carmichael (1967) and others. Due to melting and fusion of theobsidian samples inside the kiln at some temperature over 940°C, weight measurements werenot available for most of the samples. However for the two samples that were not fused, noweight changes were apparent.CHANGES IN ELEMENTAL CHEMISTRYWhile physical changes in the glass samples were abrupt and extraordinary, moreimportantly, the elemental chemistry exhibited no significant changes with a few importantexceptions. For most of the samples, there was no statistically significant changes in traceelement chemistry between ambient and the temperature beyond the melting point of siliciclava (ca. 1000°C), above that expected and typical in the instrumental variability of EDXRF(see Davis et al. 1998).Table 3 exhibits the measured elemental chemistry at ambient through all heatingsteps to 1080°C (see also Figure 4). Those elemental changes over 10% are shown in boldand underline. These changes are not necessarily related to the most obvious physicalchanges and do not correlate with modal chemistry (peraluminous versus peralkaline), orother samples analyzed here from the same source. Most intriguing is the complete depletionshowed no significant change. This is not immediately explicable, nor necessarily importantarchaeologically as we will argue. The Vulture 3 specimen gained over 30 ppm (about a19% change) in rubidium, although this may be related to analysis of a small amount ofceramic material incorporated into the obsidian at the last step as discussed earlier (Figure 4).The only significant shift in elemental composition was in one of the mildly peralkaline glasses from Antelope Wells (Table 3). Both rubidium and zirconium were depleted; 20%for rubidium and 17% for zirconium. The three-dimensional and biplots of the datagraphically indicate this change.Figure 5 graphically displays the conundrum presented by the Antelope Wells data.One of the samples was affected such that source assignment be a problem, however,given that only rubidium and zirconium were affected, source assignment could be confidentin a typical assemblage of archaeological obsidian in the southern Southwest. What is moreof a concern is the effect on only one of the samples. Sample AW-1 is well within the rangeof variability on these two elements for Antelope Wells. While Antelope Wells is distinctivein the Southwest north of the border, recent research in the basin and range region ofnorthern Chihuahua indicates a number of peralkaline obsidians used in prehistory that havesimilarly high proportions of iron and zirconium (see Shackley 1995, 1999). This couldcause a problem in this region, particularly since surveys and geoprospection are in theirinfancy in the Basin and Range region of northern Chihuahua unlike the portion of theSouthwest north of the border (Shackley 1995, 1999). As we will argue, however, pragmaticconsiderations make this apparent problem, less of an issue.SITE DEPOSITIONAL ISSUES AND HIGH TEMPERATURE INCORPORATIONOF SURROUNDING MATRIXNot surprisingly, the high temperature experiments suggested that materialincorporated into the glass can modify expected trace element composition. And while weare arguing that high temperature modification of artifact quality obsidian will notnecessarily inhibit confident assignment to source, another physical change will cause As mentioned earlier, artifacts subjected to high temperatures are relatively commonin the Southwest, particularly in pre-Classic and Classic contexts in central Arizona due toinclusion in cremation, domestic trash burning, or deliberate or accidental domestic housefires. The most well known examples are those from cremation contexts such as the obsidianpoints recovered during excavations at Snaketown (Figure 6). Based on the experimentsdiscussed above, some of these artifacts must have been subjected to temperatures near orover melting point. Most importantly here, are the examples that while not exhibitingphysical evidence of melting, are coated with material incorporated into the surface at nearmelting temperatures (Figure 6). .However, we recently analyzed an obsidian assemblage from two rooms of the UpperRuin at Tonto National Monument in Tonto Basin, central Arizona (Shackley 1998a). Bothrooms were subjected to what appears to be a high temperature fire, probably sometimeduring occupation. Three of the 22 samples analyzed were pieces of debitage that exhibitedvarious degrees of surface accumulation from the surrounding matrix, one completelycovered. As you can see in Table 4, two of the samples could be assigned to the Superior(Picketpost Mountain) source with reservation due to partial depletion of trace elementconcentrations, and one appeared nearly completely depleted in trace elements even though asmall break indicated that it was indeed obsidian (Figures 7 and 8).What is apparent here is that while we were initially concerned that high temperatureswere exclusively responsible for the depletion of trace element concentrations, the depletion apparent and due to the limitations of EDXRF. Energy Dispersive XRF at the 30kVtube voltage used in these analyses penetrates the surface only approximately 4-5 micronsm). Therefore, any significant surface accumulations will be analyzed rather than the glass itself. Either the surface must be cleaned, the artifact broken to present a unobstructedsurface, or not analyzed at all. Newer EDXRF technology, such as Kevex’s Omicron™instrument that can analyze very small areas, may ameliorate this problem in some artifacts.SUMMARY AND RECOMMENDATIONS: THE PRAGMATIC APPROACHAt least two conclusions can be derived from these experiments relevant toarchaeological applications of EDXRF analysis of archaeological obsidian. First, thereappears to be no significant change in elemental composition up to temperatures above1000°C, particularly for peraluminous silicic glasses. This is predictable given recent theoryand practical experiments in the understanding of silicic melt temperatures. Second, the realproblem lies in the interaction between those artifacts that were subjected to hightemperatures and accumulated surrounding matrix on the surface combined with theanalytical limitations of EDXRF. But are these issues really causing significant problems inthe use of obsidian compositional data in addressing archaeological problems? In thisexperimental analysis of 13 samples, only one exhibited significant changes in the traceelement composition such that source assignment became hazardous. Indeed, this AntelopeWells sample could still be assigned to source with some degree of confidence using up tofive or six of the other EDXRF measured elements that were not affected. In the case of theUpper Ruin assemblage from Tonto National Monument, only three artifacts were affectedby surface accumulation and only one could not be assigned to source.What we conclude is that melting temperatures have no significant effect on theelemental composition of obsidian (at least those elements of interest here), but the surfaceaccumulation of surrounding matrix on some artifacts can affect our ability to assign artifactsto source. This latter issue can be ameliorated by using more advanced technology, removing the coating in some manner, or eliminating that artifact from the analysis. So, thephysical changes that occur due to extreme heat do not necessarily present a problem inassigning source provenance.ACKNOWLEDGMENTSThis is an expanded version of a paper presented in the session: The Effects of at the Society for California Archaeology Annual Meeting,Sacramento, April 1999. The research was funded in part by a grant to Shackley from theStahl Endowment, Archaeological Research Facility, University of California, Berkeley. Ourthanks to Tim Teague, Staff Research Associate in the XRF and Thin Section Labs,Department of Geology/Geophysics, UC, Berkeley, for his constant help. Greg Fox, WesternArchaeological and Conservation Center (WACC), National Park Service, Tucson,graciously allowed us to re-analyze the Tonto Ruin samples. Paul Fish, Arizona StateMuseum secured funding and permission to analyze the Snaketown assemblage.REFERENCESBayman, J.M., and Shackley, M.S., 1999, Dynamics of Hohokam obsidian circulation in theNorth American Southwest. Buddington, A.F., and Lindsley, D.H., 1964, Iron-titanium oxide minerals and syntheticCann, J.R., 1983, Petrology of obsidian artefacts. In The petrology of archaeological artefacts, (eds. D.R.C. Kempe and A.P. Harvey), 227-255. Clarendon Press, Oxford.Carmichael, I. S. E., 1967, The mineralogy of Thingmuli, a tertiary volcano in easternIceland. The American Mineralogist Carmichael, I.S.E., Turner, F.J. and Verhoogen, J., 1974, . McGraw-HillBook Co., New York. Davis, K.D., Jackson, T.L., Shackley, M.S., Teague, T., and Hampel, J.H., 1998, Factorsaffecting the energy-dispersive x-ray fluorescence (EDXRF) analysis ofarchaeological obsidian. In Archaeological obsidian studies: method and theory, (ed.M.S. Shackley), 159-180. Advances in archaeological and museum sciences 3.Plenum Press, New York.Doyel, D.E., 1996, Resource mobilization and Hohokam society: analysis of obsidianartifacts from the Gatlin Site, Arizona. KivaFoster, M.S. (ed.), 1994, The Pueblo Grande project: material culture. Soil systemspublications in archaeology (4). Phoenix, Arizona.Gladwin, H.S., E.W. Haury, E.B. Sayles, and N. Gladwin, 1938, Excavations at Snaketown:material culture. University of Arizona Press, Tucson.Haury, E.W., 1976, The Hohokam, desert farmers and craftsmen: excavations at Snaketown,. University of Arizona Press, Tucson.Hildreth, W., 1979, The Bishop Tuff: Evidence for the origin of compositional zonation insilicic magma chambers. Geo. Soc. of Am., Special Paper, 43-75.Hildreth, W., 1981, Gradients in silicic magma chambers: implications for lithosphericmagmatism. J. of Geophys. Res., 10153-10192.Alliance Formation and Social Interaction during the Sedentary. Ph.D. dissertation, ArizonaState University. Ann Arbor: University Microfilms.Macdonald, R.,. Gibson, and I. L, 1969, Pantelleritic obsidians from the volcano ChabbiMahood, G., and Hildreth, W, 1983, Large partition coefficients for trace elements in high-silica rhyolites. Geochim. et Cosmochim. ActaPeterson, J., Mitchell, D.R., and Shackley, M.S., 1997, The social and economic contexts oflithic procurement: obsidian from Classic-period Hohokam sites. Shackley, M. S., 1988, Sources of archaeological obsidian in the Southwest: anarchaeological, petrological, and geochemical study. Shackley, M.S., 1990, Early Hunter-Gatherer Procurement Ranges in the Southwest:Evidence from Obsidian Geochemistry and Lithic TechnologyArizona State University. University Microfilms, Ann Arbor. Shackley, M.S., 1995,Sources of archaeological obsidian in the greater American Southwest:an update and quantitative analysis. American Antiquity Shackley, M.S., 1998a, An energy-dispersive x-ray fluorescence (EDXRF) analysis ofarchaeological obsidian from rooms 15 and 16, Upper Ruin (AZ U:8:48), TontoNational Monument, Central Arizona. Report prepared for the WesternArchaeological and Conservation Center, National Park Service, Tucson, Arizona.Shackley, M.S., 1998b, Geochemical differentiation and prehistoric procurement of obsidianin the Mount Taylor Volcanic Field, northwest New Mexico. Archaeological ScienceShackley, M.S., 1999, Obsidian sourcing, in An archaeological investigation of Late Archaic (eds. R.J. Hard and J.R. Roney),, Center for Archaeological Research, University of Texas,Shackley, M.S., 2000, Source provenance and technology of obsidian projectile points andother artifacts from Snaketown, Central Arizona: ethnic and linguistic relationshipsamong the Sedentary Period Hohokam. Report prepared for the Arizona StateMuseum, University of Arizona.Simon, A.W., McCartney, P.H., and Shackley, M.S., 1994, Lithic assemblage analysis, in Archaeology of the Salado in the Livingston Area of Tonto Basin: Roosevelt Platform , (ed. D. Jacobs), 739-760. Anthropological Field Studies 32, Arizona State University, Tempe, Arizona.Skinner, C.E., Thatcher, J.J., and Davis, M.K., 1997, X-ray fluorescence analysis and201, Surveyor Fire Rehabilitation Project, Deschutes National Forest, OregonNorthwest Research Obsidian Studies Laboratory Report 98-96, Corvallis, Oregon.Steffen, A., 1999a, When obsidian goes bad: forest fire effects on Jemez obsidian. Posterpresented at the 68 Annual Meeting of the Society for American Archaeology,Chicago, Illinois.Steffen, A., 1999b, The Dome Fire Study: extreme forest fire effects on Jemez obsidian.Paper presented at the Annual Meeting of the Society for California Archaeology,Sacramento, California.Trembour, F.N., 1990, Appendix F: A hydration study of obsidian artifacts, burnt vs. unburntby the La Mesa Forest Fire, in the 1977 La Mesa Fire Study: investigation of fire andrife suppression impact on cultural resources in Bandelier National Monument, (eds.D. Traylor, L. Hubbell, N. Wood, and B. Fielder), 174-180 28, National Park Service, Santa Fe, NewMexico. Table 1. Physical data for the experimental obsidian samples.Obsidian SourceSpecimenMax.LengthMax.WidthMax.ThicknessPre-heatedweightweight500°Cweight700°Cweight800°Cweight940°Cweight1080°C Vulture, AZ21.851.71.12.92.92.92.92.9additionalmaterial Vulture, AZ31.81.70.3111111Burro Creek, AZ12.82.10.853.83.83.83.83.8additionalmaterialBurro Creek, AZ22.72.20.53.13.13.13.13.1brokenBurro Creek, AZ31.71.50.3510.90.90.90.9brokenAntelope Wells,1(7-B-8)3.22.50.85.95.95.95.95.9ceramicfusedAntelope Wells,2(13-B-1)1.91.350.61.31.31.31.31.3ceramicfusedCow Canyon, AZ12.71.50.51.31.31.31.31.3additionalmaterialCow Canyon, AZ231.915.65.65.65.65.6brokenCow Canyon, AZ331.40.61.41.41.41.41.4brokenGovernment Mt.,14.73.11.0510.910.910.910.910.910.9Government Mt.,23.82.050.522222brokenGovernment Mt.,32.63.150.86.36.36.36.36.3broken At some point during heating to 1080° C, the ceramic sample base in the kiln shattered and some of this material wasincorporated into the melted glass. Designation for Antelope Wells sample splits also reported in Shackley (1995). Table 2. Estimated melt extrusion temperatures for various lavas (from Carmichael et al. 1974). Rhyolitetemperature underlined. Table 3. Elemental concentrations for the analysis of the five source standards at ambient through 1080°C.Bold and underlined concentrations are those Ti, Mn, Fe, and Rb – Nb that exhibited more that 10%change. All measurements in parts per million (ppm).TEMPSAMPLETiMnFeZnGaPbThRbSrYZrNb ambientV21115.1336.88412.536.417.926.815.6136.037.019.2128.218.9 V3945.4317.58229.743.820.827.832.2140.939.219.5121.718.4BC1569.5466.69088.744.123.740.637.3350.81.270.999.641.6BC2563.6462.89188.942.219.639.837.4351.63.469.696.243.7BC3476.0446.08882.950.220.237.133.5334.46.566.791.447.5AW11661.71044.123944.0191.724.649.641.5365.63.8136.81308.597.8AW22009.2971.121450.1161.025.542.033.8314.93.6119.31190.696.7CC11243.1589.810096.461.418.827.914.4157.3120.224.3144.120.5CC21143.2502.09421.644.016.722.517.1141.5113.525.7133.020.3CC31090.8492.49091.346.113.623.918.5143.6111.324.4130.718.5GM1510.7550.79513.757.922.335.918.2116.177.519.582.857.0GM2559.7578.110355.757.525.037.617.1117.979.719.279.558.8GM3526.7551.09512.155.322.835.716.7111.775.120.779.252.4500°CV21318.4348.18777.338.714.923.921.1143.439.117.9124.625.4V31024.5317.08177.032.318.726.926.4141.234.523.8131.021.6BC1620.5484.19091.543.020.937.338.6357.14.270.496.645.8BC2642.9496.69403.044.521.741.836.1355.73.369.297.244.1BC3536.9446.29040.044.117.841.439.5337.23.669.991.743.4AW11681.0999.623473.0174.523.245.340.6360.44.2134.51298.6105.0AW21769.4954.422128.2187.624.242.638.9312.54.6118.91153.193.7CC11119.3549.59969.198.926.930.418.8148.9111.529.8136.917.8CC21143.9425.18895.447.416.422.017.0137.3107.525.2129.921.1CC31193.4536.49970.754.120.629.318.4157.7119.024.9138.823.3GM1622.8532.49400.160.421.638.115.5113.378.621.484.151.8GM2479.4628.310398.859.126.235.20.0124.182.223.484.855.1GM3533.5500.89176.754.521.433.115.9112.476.119.575.049.1700°CV21055.8323.48158.434.819.432.613.4130.936.316.8123.816.1V3994.1341.38527.334.316.723.215.9139.437.118.2129.023.7BC1675.3457.99064.842.619.541.828.9344.44.869.596.443.6BC2581.2466.59395.554.723.941.340.8348.12.969.696.748.6BC3618.9455.29004.246.820.042.741.1333.13.366.690.042.2AW11714.0946.522531.4173.925.441.634.8354.85.6136.51287.295.0AW21974.0942.122596.9188.222.650.049.3313.42.5118.81152.889.8CC11315.9530.49672.446.919.425.418.4143.1117.925.2137.419.3CC21124.1486.29312.549.016.323.921.0143.4111.224.1133.617.2CC31354.0594.410286.651.620.123.615.0156.5122.730.2139.719.7GM1542.3535.09455.355.321.636.318.7117.877.816.781.857.6GM2485.6628.510367.661.425.040.822.9122.784.020.981.150.8GM3495.2561.69711.854.722.837.120.4117.378.123.582.954.4800°CV21168.2350.18541.347.223.527.720.5140.037.816.8125.315.6V31010.4329.28365.238.619.828.419.8144.133.220.0129.820.7BC1542.0498.79245.953.322.443.038.5364.14.169.698.745.4BC2628.1465.89004.449.323.041.038.2345.43.066.993.644.4BC3551.9452.08922.236.418.238.840.1343.42.669.792.440.9AW11807.11032.424632.8183.624.145.946.1379.14.8137.31333.3101.2AW21794.8864.520596.9151.324.644.643.3308.55.0117.51149.895.2CC11097.8439.98930.848.214.921.217.3141.0109.624.6128.222.1CC21177.3495.59517.772.122.828.420.0151.6110.622.6134.517.6CC3965.8464.49140.844.919.922.621.0145.2108.724.9129.518.6GM1532.1568.39636.053.923.034.120.5119.981.222.984.553.9GM2582.2671.510551.870.622.839.720.6123.585.023.682.759.8GM3524.4570.39599.656.621.334.121.3118.777.921.181.354.0 TEMPSAMPLETiMnFeZnGaPbThRbSrYZrNb 940°CV21069.2295.08476.040.317.531.723.8140.837.518.4130.113.8 V31074.0342.68508.837.217.728.214.2142.436.919.5127.720.7BC1636.5495.39160.548.022.342.840.1350.33.170.5101.748.6BC2573.9487.29113.840.921.338.135.4348.32.371.098.642.5BC3580.7467.69155.953.421.738.829.4352.02.166.898.541.4AW11652.61034.023297.7177.724.747.852.9367.34.2136.51306.299.1AW22201.91046.422446.3165.424.342.536.5317.24.9120.51192.397.6CC11303.9640.910596.657.921.629.419.9164.5121.823.6143.319.9CC21151.9542.59470.246.617.925.412.9146.1114.528.1136.122.0CC31282.4541.09972.753.920.122.219.7158.1121.328.1141.218.1GM1563.3526.09333.059.622.435.320.0114.681.920.077.558.2GM20.0 613.210645.766.524.938.519.2123.385.524.283.857.9GM3529.1553.49583.460.819.236.817.7115.875.621.079.951.91010°CV2?1014.3344.68415.034.618.059.831.6150.837.919.4135.922.01639.3 476.210662.462.527.257.023.3172.9 45.723.5146.425.1BC1?550.1515.69673.544.324.559.939.1358.03.572.197.143.7BC2?511.2400.09006.549.122.343.529.9324.73.065.696.142.8BC3624.4571.410325.956.623.760.139.4383.75.274.8102.741.7AW11654.81006.422707.6173.326.743.840.6339.55.1125.81240.995.7AW21735.5651.117325.5136.814.939.024.0253.1 6.5105.8986.7 80.4CC11165.3520.99620.396.823.843.715.0147.3111.825.2131.816.6CC2?1191.6502.39528.146.218.141.821.0144.9113.428.0138.119.3CC3?1142.5506.49597.751.418.032.323.6151.3108.525.8123.621.0GM1496.3537.49397.254.320.550.914.2109.675.721.882.851.5GM2?0.0 478.19163.556.020.740.816.1108.874.723.375.746.2GM3473.7491.79204.256.920.952.215.9109.376.820.279.452.8 Those samples marked with a “?” are samples that deformed too much to determine which sample of the source group that particulsample belonged.Table 4a. X-ray fluorescence concentrations for archaeological samples from Rooms 15 and 16, Upper Ruin,Tonto National Monument (from Shackley 1998a). All measurements in parts per million (ppm).SAMPLETiMnFeRbSrYZrNbSource 1401243.5208.38365.7101.730.017.9100.61.2Superior* 156771.3301.46285.375.812.916.554.016.1Superior*294597.826.84116.14.814.20.07.04.1burned* These are source probabilities based on best linear fit of the calibration utility (Shackley 1995). Those samples markedwith "*" can only tentatively be assigned to source due to a less than adequate fit with the available source standards.These samples appear to be burned and/or chemically weathered such that the elemental chemistry may be altered.Table 4b. Superior (Picketpost Mountain), Arizona source standard mean and central tendency data (Shackley 1995). Ba 243.7 5.57 237.0 254.6 13 19 Figure 1. Sources of archaeological obsidian in the greater North American Southwest. 20 Figure 2. Experimental samples before heating. 21 Figure 3. Samples after heating to 1080ºC. Off-white material is the broken ceramic base plate incorporated into glasswhile heating to this temperature (see text). 22 1401400 100 1201200 200 1001000 400 80800 60600 400 Figure 4. Rb, Sr, Zr ambient and 1080ºC concentrations for experimental samples. AMB=ambient measurements;+1000=1080ºC measurements. 23 1400120010008006004002000Rb ppm4003002000 SAMPLEAW2AW1Antelope Wells(source standards) Figure 5. Rb versus Zr biplot of elemental concentrations for Antelope Wells samples and source standards after heatingto 1080ºC. 24 Figure 6. Top: Selected Snaketown Serrated points from Snaketown. Approximately 40% are burned to some degree.All could be assigned to source (from Shackley 2000); Severely burned and physically modified projectile points fromSnaketown. Note incorporation of matrix on center and right specimens (from Gladwin et al. 1938, plate XXXVIIbottom). 25 Figure 7. Burned obsidian sample with surface accumulation of matrix (Sample 294, Room 16, Upper Ruin;courtesy WACC/NPS). SOURCESuperior? 40120 0 20 100 3080 160 2040 20 Figure 8. Rb, Sr, Zr three-dimensional plot of three artifacts from the Upper Ruin, Tonto National Monument, andSuperior (Picketpost Mtn) source standards.