the geologic sources for obsidian artifacts

53HUGHES -The GeoloGic SourceS for obSidian arTifacTS from minneSoTa

Introduction

Over ninety-five years ago, N.H. Winchell (1911: 415) brought obsidian to the attention of local pre-historians when he illustrated an obsidian projec-tile point in the Jacob Brower collection. Since that time, obsidian occurrences in Minnesota archaeo-logical sites have rarely been noted in professional literature. When Griffin (1965: 147) prepared his exhaustive list of obsidian finds in middle western archaeological sites there were "no known sites with obsidian west of eastern Iowa, or from south-ern Minnesota", although finds in neighboring states (Wisconsin, North and South Dakota) were on re-cord.

I began to re-analyze obsidian recovered from Ohio Hopewell mounds more than two decades ago and have examined more recently obsidian artifacts recovered from other middle western archaeologi-cal sites (e.g. Hughes 1995a, 2006a; Hughes and Fortier 1997). During this process, I became aware of the presence of obsidian in Minnesota archaeo-logical sites (see acknowledgments), although these sourcing results were not included in my broader Hopewellian study. While archaeological research in the Middle Western U.S. has accelerated at a diz-zying pace over the last three decades and many

more geochemical analyses of obsidian have been reported from adjacent areas (see, e.g. Anderson et al. 1986; Baugh and Nelson 1988; Davis 1995; Hughes and Lees 1991; Hughes and Nelson 1987; Hughes and Roper 1999; Stoltman and Hughes 2004), chemical data critical to understanding the broad outlines of prehistoric obsidian source use in Minnesota have been lacking. To help remedy this situation, this study, the first of its kind in the state, presents trace element data documenting the geo-logic sources of obsidian artifacts recovered from archaeological sites throughout Minnesota.

The Sites

Fifty-three obsidian artifacts from 28 Minnesota archaeological sites were subjected to non-destruc-tive energy dispersive x-ray fluorescence (EDXRF) analysis. 1 The general locations of these sites appear in Figure 1, and the site-specific details are listed in Table 1.

Three of the sites from which obsidian was analyzed (21-BL-8, 21-BL-11, and 21-BL-13) are located within the Leech Lake Indian Reservation of north-central Minnesota. The obsidian specimen from 21-BL-8 occurred in a Late Woodland context, dated ca. 800 –1400 A.D. on the basis of Blackduck and Sandy

THE GEOLOGIC SOURCES FOR OBSIDIAN ARTIFACTS FROM MINNESOTAARCHAEOLOGICAL SITES

Richard E. HughesGeochemical Research Laboratory20 Portola Green CirclePortola Valley, CA 94028www.geochemicalresearch.com

Fifty-three obsidian artifacts from 28 Minnesota archaeological sites were analyzed to determine their geologic source(s) of origin using non-destructive energy dispersive x-ray fluorescence analysis. Results indicate that the majority of samples, predominantly representing Woodland period times, were manufactured from volcanic glass of the Obsidian Cliff chemical type, located in Yellowstone National Park, Wyoming, and that even more distant obsidians from Idaho (Bear Gulch and Malad) occur at some sites in the state. Non-volcanic glass from the Powder River area of Montana occurred at one site. Obsidian from Southwestern U.S. sources was not identified in this sample.

54 THE MINNESOTA ARCHAEOLOGIST - VOLUME 66 - 2007

Lake pottery, but additional testing in 1986 revealed that this site also has a Laurel component (ca. 200 B.C. – 500 A.D.). Site 21-BL-11 is also multi-com-ponent, containing an historic Ojibwa cemetery, a possible Late/Middle Woodland component (ca. 400-500 A.D.), and a Late Woodland component, ca. 800 –1200 A.D., marked by the presence of Blackduck pottery. 21-BL-13 contained Blackduck pottery but may also have a Late/Middle Woodland component.

Because the obsidian flakes from all three of these sites came from shovel probes, their association with dated components is problematic. Three obsid-ian samples were analyzed from the Big Rice site (21-SL-163), located north of Virginia in northeast-ern Minnesota. Excavations were conducted at this site from 1983-1986. There is a Paleoindian compo-nent, but its major fluorit was during the Woodland period. A single Pelican Lake point was analyzed

Figure 1. General location of archaeological sites in Minnesota included in this study (area enlarged from Figure 2).


from 21-AN-2. The site also yielded Knife River Flint and Howard and Onamia ceramics, pointing to a Middle – Late Woodland temporal affiliation. 21-AN-108 yielded seven flakes for analysis, and also contained Knife River Flint, Tongue River Silica, and Howard and Onamia ceramics suggesting occu-pation/use during Middle-Late Woodland times. The two specimens analyzed from 21-AK-7 were recov-ered in the upper 20 cm. of deposit from excavation units, which may date to Middle-Late Woodland or Late Woodland times. The sole obsidian flake recov-

ered from 21-OT-36 also came from excavations in the upper 20 cm. of deposit, which may be associ-ated with the Late Woodland period. 21-DL-85 is a multi-component site containing Woodland and Ar-chaic components. The flake analyzed from 21-SL-785 (containing Laurel and Blackduck components) came from a shovel test 20-30 cm. below surface. Likewise, the specimen analyzed from 21-SW-14 came from a shovel test, 15-20 cm. below surface, and the presence of cord-roughened body sherds suggests a Woodland age for the site. Specimens

___________________________________________________________________________________________

Site n samples Artifact Type(s)

21-AN-2 1 Pelican Lake point21-AN-108 7 Flakes21-AK-7 2 Flake, Tool fragment (scraper # 942.39.1)21-BL-8 1 Flake21-BL-11 1 Flake21-BL-13 1 Flake21-BL-37 2 Flakes21-BW-24 1 Flake21-CA-586 1 Flake21-DL-2 1 Flake21-DL-85 1 Flake21-DL-101 1 Flake21-HE-129 1 Flake21-IA-26 1 Flake21-KH-36 1 Flake21-ME-1 2 Flakes21-ME-14 11 Flakes21-ML-1 3 Flake, point fragment (# 729.74)21-ML-11 1 Flake21-ML-40 1 Flake21-NI-44 1 Flake21-OT-36 1 Flake21-SL-1 1 Tool fragment (scraper)21-SL-163 3 Flakes21-SL-785 1 Flake21-SW-14 1 FlakeRoosevelt Lake Narrows 3 FlakesWindy Bead (05-373) 1 Flake ___ Σ= 53

Table 1: Site-specific Distribution of Minnesota Obsidian Samples Analyzed via Energy Dispersive X-ray Fluorescence Spectrometry


from 21-ME-14 were recovered from unstratified deposits in this Transitional Middle-Late and Late Woodland habitation site. 21-ML-40 also revealed Transitional Middle-Late and Late Woodland period use, but the obsidian sample could not be attributed a more specific temporal affiliation. The flake ana-lyzed from 21-DL-2 came from excavation along with smooth-surfaced ceramic body sheds possibly indicating an early Woodland affiliation. The speci-men from 21-HE-129 was recovered from a shovel test in a single-component Middle-Late Transition period site, containing Onamia ceramics, Great Oa-sis Ware, and Knife River Flint. The sample ana-lyzed from 21-SL-1 came from disturbed deposits there, with no additional provenience information. Roosevelt Narrows specimens were recovered from depths of 20-45 cm. below surface, but the temporal period(s) of site use are unknown. The samples ana-lyzed from 21-BL-37 and 21-CA-586 also cannot be attributed to specific time period. The specimen

analyzed from the Windy Bead site (05-373), pre-dominantly an Initial Woodland period site marked by Laurel ceramics, was recovered from bioturbated soils in Unit 1. The samples analyzed from 21-BW-24, 21-DL-101, 21-IA-26, 21-KH-36, 21-ME-1, 21-ML-11, 21-NI-44 were recovered either from the surface, or surface plowzone.

X-ray Fluorescence Analysis and Results

The obsidian samples were subjected to trace ele-ment study using EDXRF analysis. Laboratory anal-yses were conduced by the author using a QuanX-EC™ (Thermo Electron Corporation) EDXRF spectrometer equipped with a silver (Ag) x-ray tube, a 50 kV x-ray generator, digital pulse proces-sor with automated energy calibration, and a Peltier cooled solid state detector with 145 eV resolution (FWHM) at 5.9 keV. The x-ray tube was operated at differing voltage and current settings to optimize

Figure 2. The study area in relation to geologic obsidian source deposits in the western United States. Dots show the location of obsidian and non-volcanic glass source material identified at Minnesota archaeological sites; open circles depict other regionally significant western North American obsidian sources, and source areas, not identified at any site in Minnesota.


Table 2. Energy Dispersive X-ray Fluorescence Data for Artifacts from Minnesota Archaeological Sites

Trace Element Concentrations RatioSite / Cat. Obsidian SourceNumber Rb Sr Y Zr Nb Ba Ti Mn Fe2O3T Fe/Mn (Chemical Type)

05-373, 168 49 44 297 53 771 nm nm nm 52 Bear Gulch, IDLot # 20 ±4 ±3 ±3 ±4 ±3 ±10

21-AN-2, 242 4 77 169 44 nm nm nm nm 73 Obsidian Cliff, WY19 ±4 ±3 ±3 ±4 ±3

21-AN-108, 188 21 70 228 52 181 nm nm nm 55 Powder River, WY1 ±4 ±3 ±3 ±4 ±3 ±10

21-AN-108, 190 16 71 236 53 213 nm nm nm 61 Powder River, WY5 ±4 ±3 ±3 ±4 ±3 ±10

21-AK-7, 160 46 42 291 54 738 nm nm nm 52 Bear Gulch, ID942.37.1 ±4 ±3 ±3 ±4 +3 ±10

21-BL-37, 224 5 69 157 10 nm 490 198 nm nm Obsidian Cliff, WY213.476.4 ±4 ±3 ±3 ±4 ±3 ±17 ±11

21-BL-37, 250 5 77 164 41 nm nm 245 1.49 nm Obsidian Cliff, WY387.84.3 ±4 ±3 ±3 ±4 ±3 ±11 ±.10

21-BW-24, 124 75 32 89 15 1628 nm nm nm nm Malad, ID157.21.1.1 ±4 ±3 ±3 ±4 ±3 ±12

21-CA-586, 249 7 76 163 39 nm nm 241 1.41 nm Obsidian Cliff, WYLL96.6.1.1 ±4 ±3 ±3 ±4 ±3 ±15 ±.11

21-DL-2, 118 72 33 84 14 1727 nm nm nm nm Malad, ID2004.138.203.1 ±4 ±3 ±3 ±4 ±3 ±15

21-DL-101, 245 6 79 175 43 nm nm nm nm 72 Obsidian Cliff, WY1990.276.4 ±4 ±3 ±3 ±4 ±3 21-ME-1, 231 6 79 168 44 nm nm nm nm 77 Obsidian Cliff, WY365.39.619 ±4 ±3 ±3 ±4 ±3

21-ML-1, 166 48 41 300 55 762 nm nm nm 48 Bear Gulch, ID729.74 ±4 ±3 ±3 ±4 ±3 ±10

21-ML-1, 218 6 74 160 40 nm nm nm nm 76 Obsidian Cliff, WY729.116 ±4 ±3 ±3 ±4 ±3

21-NI-44, 241 7 81 170 44 0 nm nm nm 70 Obsidian Cliff, WY081339 ±4 ±3 ±3 ±4 ±3 ±15

21-SL-1, 223 5 72 162 42 nm nm nm nm 68 Obsidian Cliff, WY233.173 ±4 +3 ±3 ±4 ±3

21-SL-785, 250 6 81 172 45 nm nm nm nm 66 Obsidian Cliff, WY1999.446.2.1 ±4 ±3 ±3 ±4 ±3

Roosevelt Lake 255 3 86 179 46 nm nm nm nm nm Obsidian Cliff, WYNarrows, # 1 ±3 ±3 ±2 ±3 ±2


excitation of the elements selected for analysis. In this case analyses were conducted on all specimens for the elements rubidium (Rb Kα), strontium (Sr Kα), yttrium (Y Kα), zirconium (Zr Kα), and nio-bium (Nb Kα) and certain artifacts were analyzed to determine concentrations of barium (Ba Kα), ti-tanium (Ti Kα), manganese (Mn Kα), and total iron (Fe2O3

T). Iron vs. manganese (Fe Kα/Mn Kα) ratios also were computed for some samples. In all cases, x-ray tube current was scaled to the physical size of each specimen. Other details involving calibration procedures, element-specific measurement resolu-tion, and comparative literature references appear elsewhere (Hughes 1988: 257-258; 1994a: 265-267; Hughes and Pavesic 2005: 249-250). The analyses were completely non-destructive; sample pretreat-ment was limited to a simple cleaning with distilled water to remove any possible surface contaminants.

Twenty artifacts were large enough to generate reli-able quantitative composition estimates. As Table 2 and Figure 3 show, 13 of these were manufactured from volcanic glass of the Obsidian Cliff chemical type in Yellowstone National Park, Wyoming (cf. Nelson 1984: Table 5, source # 49; Anderson et al. 1986: Table 4, source # 30 [also Baugh and Nelson 1988: Table 5; Hughes 1995b: Table 2]), three match the profile of Bear Gulch, Idaho, obsidian (Hughes and Nelson 1987: Table 1), two were fashioned from obsidian from the Malad, Idaho, source (Nelson 1984: Table 5, source # 31; Hughes 1984: Table 3),

and two others have the same trace element profile as Powder River non-volcanic glass (Frison et al. 1968: Tables 1 and 2; Hughes 2007b: Table A.4). Though other elements could be used to illustrate chemical contrasts among sources, yttrium (Y) and zirconium (Zr) concentration values unambiguously separate Obsidian Cliff glass from all other archaeo-logical samples in this study. 2

Most of the specimens analyzed here were too small (i.e. <9-10 mm diameter) and/or too thin (i.e. < ca. 1.5 mm thick) to generate reliable quantitative com-position estimates using non-destructive EDXRF. I have recently completed successful semi-quantita-tive analysis experiments on similarly small speci-mens from the Big Horn Basin of Wyoming, so I applied this same analysis protocol to these Minne-sota specimens (see Table 3). This approach utilizes Rb/Sr/Zr intensity ratios in combination with Zr/Y and Rb/Sr ratios to compare with geological source standards. In Hughes (2007b) comparative analysis was restricted to archaeologically significant obsid-ians in Wyoming, but for the present analysis I ex-panded the source universe to include all of the ma-jor obsidian chemical types identified to date in cen-tral Plains and Middle Western U.S. archaeological assemblages (see Figure 4). The Rb/Sr/Zr intensity ratios, as well as Zr/Y and Rb/Sr ratios, determined for these 37 obsidian flakes align with values de-termined for geologic samples from Obsidian Cliff,

Roosevelt Lake 260 4 80 175 50 nm nm nm nm nm Obsidian Cliff, WYNarrows, # 3 ±3 ±3 ±2 ±4 ±2

Roosevelt Lake 273 3 87 176 47 nm nm nm nm nm Obsidian Cliff, WYNarrows, # 4 ±3 ±3 ±2 ±3 ±2 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -RGM-1 143 105 23 214 8 804 1616 278 1.87 62 Glass Mtn., CA (measured) ±4 ±3 ±3 ±4 ±3 ±10 ±15 ±10 ±.02 RGM-1 149 108 25 219 9 807 1600 279 1.86 nr Glass Mtn., CA (recommended)

Note: Values in parts per million (ppm) except total iron [in weight %] and Fe/Mn intensity ratios; ± = expression of x-ray counting uncertainty and regression fitting error at 120-360 seconds livetime. nm= not measured. nr= not re-ported. Data for samples from 21-BL-37 and 21-CA-586 from Hughes (2002), Roosevelt Lake Narrows sample data from Hughes (1994b). Most samples have Minnesota Historical Society (MHS) catalogue numbers. Sample RGM-1 is a U.S. Geological Survey Reference standard.

Table 2. cont. Energy Dispersive X-ray Fluorescence Data for Artifacts from Minnesota Archaeological Sites


Wyoming (n= 23), Bear Gulch, Idaho (n= 7), and Powder River, Montana (n= 7).

Despite these correspondences, the inherent limits of non-destructive semi-quantitative analysis must be acknowledged (see Hughes 1998:106-107 for discussion) though I have attempted to minimize, if not nullify, their effects here by conducting quan-titative and semi-quantitative intralaboratory analy-ses on suitably large geological and archaeological specimens, which yield ratio-level equivalents for specimens of smaller physical size. Though samples should, whenever possible, be analyzed to gener-ate quantitative composition data, in circumstances where this is not possible (i.e. when tiny obsidian flakes are all that occur at a given site), semi-quanti-tative analysis- utilized with appropriate caution and coupled to a quantitative reference foundation- can

be used productively to generate reliable and repli-cable non-destructive results. Crosschecks between techniques, like to ones completed here, should rou-tinely be conducted to help evaluate the substantive utility of EDXRF lab results derived from different approaches.

Combining the results from quantitative and semi-quantitative analysis of these 53 Minnesota speci-mens (Tables 2 and 3), it was determined that 35 were made from Obsidian Cliff, Wyoming, glass, nine were fashioned from obsidian of the Bear Gulch, Idaho chemical type, seven have the same trace element profile as Powder River non-volcanic glass, and that two others were manufactured from obsidian of the Malad, Idaho, geochemical type (see Figure 2).

Figure 3. Y vs. Zr composition of obsidian artifacts from Minnesota archaeological sites. Data from Table 2 herein. Open triangles represent values for artifacts, while dashed lined demarcate the range of variation measured in geological obsidian reference samples. The numbers of plots do not correspond exactly to the numbers in table 2 because of convergence in data points at this scale.


Summary and Concluding Comments

The results of the current study document that ob-sidians from two major sources-- Obsidian Cliff in Yellowstone National Park, Wyoming, and Bear Gulch in the southern Centennial Mountains of Ida-ho-- were most frequently represented at archaeo-logical sites in Minnesota, with Obsidian Cliff by

far the most dominant numerically. This result was to be anticipated, based on knowledge from adjacent areas. Obsidian Cliff has long held renown as a ma-jor source of high-quality obsidian (Holmes 1879; Iddings 1888)—so much so that nearly a century ago Winchell (1911: 415) speculated that it was the source for the obsidian artifact from the Brower Col-lection (but see note 1 below). This obsidian was ex-

Element Intensities Intensity Ratios Obsidian SourceSite Cat. No. Rb Sr Zr ΣRb,Sr,Zr Rb% Sr% Zr% Zr/Y Rb/Sr (Chemical Type)

05-373 Lot # 20 2934 844 9153 12931 .227 .065 .708 8.7 2.9 Bear Gulch, ID21-AN-108 1 3276 233 7072 10581 .310 .022 .668 4.2 7.8 Powder River, MT21-AN-108 2 3847 145 6905 10897 .353 .013 .634 4.2 9.2 Powder River, MT21-AN-108 3 4716 138 8631 13485 .350 .010 .640 4.5 10.3 Powder River, MT21-AN-108 4 4278 226 8128 12632 .339 .018 .643 4.2 9.6 Powder River, MT21-AN-108 5 3944 168 7939 12051 .327 .014 .659 4.4 9.4 Powder River, MT21-AN-108 6 4105 99 7875 12079 .340 .008 .652 4.5 10.3 Powder River, MT21-AN-108 7 4697 190 7902 12789 .367 .015 .618 4.3 9.9 Powder River, MT21-AK-7 942.40.1 3244 1028 9764 14036 .231 .073 .696 9.0 2.9 Bear Gulch, ID21-BL-8 483 5259 0 5938 11197 .470 .000 .530 3.0 33.8 Obsidian Cliff, WY21-BL-11 362 5997 0 5995 11992 .500 .000 .500 2.7 40.4 Obsidian Cliff, WY21-BL-13 763 4880 0 5175 10055 .485 .000 .515 2.8 42.0 Obsidian Cliff, WY21-DL-85 186.1 4735 0 5419 10154 .466 .000 .534 2.7 30.2 Obsidian Cliff, WY21-HE-129 1993.151.17.1 5711 0 5551 11262 .507 .000 .493 2.7 41.9 Obsidian Cliff, WY21-IA-26 179.15.11.3 2156 0 1929 4085 .528 .000 .472 2.8 39.5 Obsidian Cliff, WY21-KH-36 183.25.36 4765 0 4952 9717 .490 .000 .510 2.6 34.6 Obsidian Cliff, WY21-ME-1 365.39.620 3363 901 10995 15259 .220 .059 .721 9.0 3.1 Bear Gulch, ID21-ME-14 1991.294.33.9 2336 0 2383 4719 .495 .000 .505 2.8 31.8 Obsidian Cliff, WY21-ME-14 1993.132.36.1 4161 0 3955 8116 .513 .000 .487 2.9 36.3 Obsidian Cliff, WY21-ME-14 1993.132.529.1 5687 0 5945 11632 .489 .000 .511 2.8 44.1 Obsidian Cliff, WY21-ME-14 1993.132.629.2 4861 0 4392 9253 .525 .000 .475 2.6 34.4 Obsidian Cliff, WY21-ME-14 1993.132.629.8 5584 0 5527 11111 .503 .000 .497 2.7 33.1 Obsidian Cliff, WY21-ME-14 1993.132.645.16 6285 0 6002 12287 .512 .000 .488 2.7 37.3 Obsidian Cliff, WY21-ME-14 1993.132.646.1 4848 0 5296 10144 .478 .000 .522 2.8 35.0 Obsidian Cliff, WY21-ME-14 1993.132.650.1 4849 0 4236 9085 .534 .000 .466 2.6 45.8 Obsidian Cliff, WY21-ME-14 1993.132.650.10 4543 0 4943 9486 .479 .000 .521 2.8 39.2 Obsidian Cliff, WY21-ME-14 1993.132.655.2 5842 0 5662 11504 .508 .000 .492 2.8 42.5 Obsidian Cliff, WY21-ME-14 1993.132.674.1 5958 0 5244 11202 .532 .000 .468 2.6 38.0 Obsidian Cliff, WY21-ML-1 729.107.1 3404 763 10146 14313 .238 .053 .709 8.8 3.3 Bear Gulch, ID21-ML-11 578.286 5185 0 5154 10339 .502 .000 .498 2.7 31.0 Obsidian Cliff, WY21-ML-40 1988.368.7.5 5076 0 5153 10229 .496 .000 .504 2.8 37.2 Obsidian Cliff, WY21-OT-36 667.28 5513 0 5606 11119 .496 .000 .504 2.7 32.7 Obsidian Cliff, WY21-SL-163 L298, 95/94 2820 834 9043 12697 .222 .066 .712 9.1 3.3 Bear Gulch, ID21-SL-163 L329, 85/73 2750 861 8658 12269 .224 .070 .706 9.1 3.2 Bear Gulch, ID21-SL-163 L377, 95/93 2605 829 7998 11432 .228 .073 .699 8.8 3.1 Bear Gulch, ID21-SL-785 1999.446.2.1 4768 0 5291 10059 .474 .000 .526 2.8 32.8 Obsidian Cliff, WY21-SW-14 60.1 6026 0 5383 11409 .528 .000 .472 2.7 35.3 Obsidian Cliff, WY

Table 3. Semi-quantitative Element Data for Obsidian Samples from Minnesota Archaeological Sites

Note: Element intensities (counts-per-second, peak above background) generated at 120 seconds livetime. Note: speci-men no. 2.1 from 21-SL-785, Lot # 20 from Windy Bead (FS site 05-373), and # 1 and # 5 from 21-AN-108 also subjected to quantitative analysis (see Table 2).


tensively exploited for thousands of years in the im-mediate vicinity of Yellowstone National Park (see Davis 1995; Cannon and Hughes 1995) and it has been recovered in Woodland, Late Prehistoric, and Protohistoric contexts throughout the middle West (e.g., Griffin et al. 1969; Anderson et al. 1986; Lo-gan et al. 2001; Stoltman and Hughes 2004; Hughes 2006a) and in western Ontario, Canada (Godfrey-Smith and Haywood 1984). Bear Gulch obsidian also was important prehistorically in Montana and vicinity (see Cannon and Hughes 1993, 1994; Will-ingham 1995), and it has been identified archaeo-logically in North Dakota (Baugh and Nelson 1988), South Dakota (Hughes 2007a) and in Middle Wood-

land contexts in Iowa (Anderson et al. 1986: Table 6 [Unknown A]; cf. Hughes and Nelson 1987; Logan et al. 2001), Illinois (Hughes and Fortier 1997), In-diana, Ohio and Wisconsin (Hughes 2006a).

The geochemical identification of Powder River non-volcanic glass at 21-AN-108 was a major finding. This material, first identified as poor-quality obsidi-an (Frison et al. 1968: 214-215), was later described as a natural (non-volcanic) glass derived from the Powder River Basin of Montana and Wyoming (Fri-son 1974: 61; Frison and Wilson 1975: 29). Because of the excellent flaking properties of smaller nodules (Frison 1974: 61), this raw material was particular-

Figure 4. Bivariate plot of Rb/Sr vs. Zr/Y intensity ratios for artifacts from Minnesota archaeological sites. Dashed lines represent range of variation measured in geologic obsidian source samples. Dots represent plots for samples from Minnesota sites from data in Table 3. Number of artifact plots do not correspond exactly to the tabulation in Table 3 because of overlapping values at this scale.


ly significant during very late prehistoric times in the Big Horn Basin of northern Wyoming (Hughes 2007b) and the northwestern Plains (Frison et al. 1968: 216), but little is known of its earlier use. To my knowledge, this is the first geochemical identifi-cation of Powder River glass in the Middle Western United States. In addition, the presence of Malad, Idaho, obsidian at two sites in this study (21-BW-24 and 21-DL-2) is notable; this obsidian is rare even in North Dakota and South Dakota sites, though it has a much wider distribution in the central and south-ern Great Plains and has recently been identified in Illinois (Hughes 2006a: Table 20.2, 370) and Arkan-sas (Hughes et al. 2002).

As Table 1 shows, the vast majority of artifacts in this study were flakes. Twenty of the 28 sites exam-ined here are represented by single obsidian flakes, while six other sites had 2-3 specimens, and only two sites (21-AN-108 and 21-ME-14) yielded seven or more. Given this, caution is clearly in order when attempting to discern potential patterning in these data. A single Pelican Lake point fragment from 21-AN-2, an untypable point fragment from 21-ML-7, and two specimens identified as scrapers (one each from 21-AK-7 and 21-SL-1) were the sole formed tools analyzed so it was not possible to combine time-sensitive artifact form with geochemistry to ar-rive at more precise estimates of the time period(s) during which obsidian was conveyed in to most of these sites. Though the overwhelming majority of sites in this sample are of Woodland period age based on time-sensitive ceramics, absent convincing stratigraphic or direct feature associations, the time period(s) represented by pottery may, or may not, apply to the episode(s) of obsidian deposition. Only two sites in this study (21-ME-1 and 21-ML-1) had obsidian of more than one chemical type; all other sites contain exclusively Obsidian Cliff, Bear Gulch, Powder River, or Malad. But because the vast ma-jority of sites produced only single specimens, this “pattern” is very likely to be merely an artifact of sample size. With this geochemical baseline study complete, it might be productive to employ obsidian hydration analysis to discern possible temporal dif-ferences in obsidian source use.

In a previous study (Hughes 2006a), I noted a difference in the areal distribution of Bear Gulch obsidian in several Middle Western Hopewellian sites. Although only a single Indiana site (the Mann site) was represented therein, it contained significantly greater relative proportions of Bear Gulch obsidian than any Hopewellian site in Illinois or Ohio. These distributional data suggested the possibility that Bear Gulch obsidian may have been introduced into Indiana (the Mann site) either earlier or later than the obsidian analyzed from Illinois and Ohio, which featured dominant use of Obsidian Cliff glass, and that it may have been obtained or conveyed via different routes (or networks). However, because it was not possible to assign specific dates (or very tight age ranges) to the artifacts analyzed here, the potential Obsidian Cliff vs. Bear Gulch age/distribution/functional contrasts, anticipated years ago for Hopewell by Struever and Houart (1972; also Flannery 1972: 131-132), could not be addressed meaningfully with these Minnesota data. Furthermore, no strong geographic trends (i.e., north/south, east/west) are apparent in the results of this study, so examination of issues of broader regional significance must await better dating and increased sample size. Bear Gulch glass has been identified in Wisconsin sites, but its occurrences there have been limited to formal tools (Hughes 2006a: Table 20.2). While this may be fortuitous, in that the archaeological recovery techniques employed by Thomas (1894) and McKern (1931) were not optimized for flake capture, it also may reflect temporal variability in obsidian source use similar to that hypothesized for source-use differences between Indiana and adjacent Hopewell expressions (Hughes 2006a).

In addition to detailing what was found by this laboratory study, what was not found also is wor-thy of note. Although obsidian artifacts made from volcanic glass originating in the Jemez Mountains of northern New Mexico have been recovered in neighboring states (i.e., at 39PO61 in South Dakota [Hughes 1999, 2006b] and at 25LP8 in Nebraska [Hughes and Roper 1999]), no obsidian from this southwestern U.S. source area was identified in this sample from Minnesota, despite the fact that


the Jemez Mountains are slightly closer to south-ern Minnesota than is Malad, Idaho (see Figure 2). While Baugh and Nelson (1987) showed that Jemez Mountains glasses dominate the obsidian fraction identified in many southern Great Plains assem-blages, and the importance of these obsidians has been documented during Protohistoric times in the central Plains (Hughes and Lees 1991; Hughes and Roper 1999), northern New Mexico obsidians have not- to my knowledge- been geochemically identi-fied to date in Iowa, North Dakota, or Wisconsin.

Finally, it is worth reiterating that the goal of this pi-lot study was to provide an overview of the source-specific occurrences of obsidian in the state. There are undoubtedly obsidian samples from many more sites in existing institutional repositories that should be analyzed to amplify and refine what was learned here. It is hardly necessary to underscore the im-portance to non-destructive laboratory-based scien-tific studies of well-maintained museum collections; without access to the Minnesota Historical Society holdings, research of this statewide scope literally could not have been completed. Particularly impor-tant in future provenance studies will be a focus on refined dating which will allow us to better specify the time period(s) during which obsidians of differ-ent chemical types came to, and were used by, the prehistoric peoples of Minnesota.

Acknowledgments. Christy Hohman-Caine, former Min-nesota State Archaeologist, supported this project in its early days by sending samples and detailed comments on context and dating. Gordon Peters (formerly Superior National Forest Archaeologist), and John Hunn (former Superior National Forest archaeological technician) pro-vided specimens and contextual information, as did Grant Goltz (Soil Scientist). Jodie O’Gorman (formerly Manka-to State University) contributed the specimen from 21-NI-44 along with contextual information. William Clay-ton (Superior National Forest Archaeologist) provided specimens for analysis and documentary assistance. Deb-orah Schoenholz and Kent Bakken were particularly en-thusiastic and helpful supporters of this project and were instrumental in helping me secure artifact loans. I extend special thanks to Patricia Emerson (Head of Archaeol-

ogy, Minnesota Historical Society) and Rose Kubiato-wicz (Assistant Registrar, Minnesota Historical Society) for approving and facilitating the loan of obsidian from the Minnesota Historical Society collections. Pat gener-ously provided detailed notes on the contexts of the finds. Without Pat and Rose’s help, this study would never have been completed on time. Finally, I very much appreciate Tammara Norton’s drafting of Figures 1 and 2.

References Cited

Anderson, D.C., J.A. Tiffany, and F.W. Nelson1986 Recent Research on Obsidian from Iowa

Archaeological Sites. American Antiquity 51: 837-852.

Baugh, T. G., and F. W. Nelson1987 New Mexico Obsidian Sources and Ex-

change on the Southern Plains. Journal of Field Archaeology 14: 313-329.

1988 Archaeological Obsidian Recovered from

Selected North Dakota Sites and Its Rela-tionship to Changing Exchange Systems in the Plains. Journal of the North Dakota Ar-chaeological Association 3: 74-94.

Bennyhoff, J.A.1994 Central California Augustine: Implications

for Northern California Archaeology. In R. E. Hughes (ed.), Toward a New Taxonomic Framework for Central California Archae-ology: Essays by James A. Bennyhoff and David A. Fredrickson, pp. 65-74. Con-tributions of the University of California Archaeological Research Facility No. 52. Berkeley.

Bowman, H.R., F. Asaro, and I. Perlman 1973 On the Uniformity of Composition in Ob-

sidians and Evidence for Magmatic Mixing. Journal of Geology 81: 312-327.

Cannon, K.P., and R.E. Hughes1993 Obsidian Source Characterization of Paleo-

indian Projectile Points from Yellowstone


National Park, Wyoming. Current Research in the Pleistocene 10:54-56.

1994 Emerging Patterns of Obsidian Utilization in Yellowstone National Park. Paper presented at the 59th Annual Meeting of the Society for American Archaeology, Anaheim, Cali-fornia.

1995 Its Not Just Obsidian Cliff: Diversity of Ob-sidian Source Use in Yellowstone National Park. Paper presented at the 2nd Biennial Rocky Mountain Anthropological Confer-ence, Steamboat Springs, Colorado.

Davis, L.B.1995 Cultural Utilization and Redistribution of

Obsidian Cliff Plateau Obsidian. In The Obsidian Cliff Plateau Prehistoric Lithic Source, Yellowstone National Park, Wyo-ming, by L.B. Davis, S.A. Aaberg, J.G. Schmitt, and A.M. Johnson, pp. 39-58. Selections from the Division of Cultural Resources No. 6. National Park Service, Rocky Mountain Region.

Flannery, K.V.1972 Summary Comments: Evolutionary Trends

in Social Exchange and Interaction. In E. N. Wilmsen (ed.), Social Exchange and Inter-action, pp. 129-135. Museum of Anthro-pology, University of Michigan, Anthropo-logical Papers No. 46. Ann Arbor.

Frison, G.C.1974 The Application of Volcanic and Non-Vol-

canic Natural Glass Studies to Archaeology in Wyoming. In M. Wilson (ed.), Applied Geology and Archaeology: The Holocene History of Wyoming. Geological Survey of Wyoming Report of Investigations 10: 61-64.

Frison, G.C., G.A. Wright, J.B. Griffin, and A.A. Gordus

1968 Neutron Activation Analysis of Obsidian: An Example of Its Relevance to Northwest-ern Plains Archaeology. Plains Anthropolo-gist 13: 209-217.

Frison, G.C., and M. Wilson1975 An Introduction to Bighorn Basin Archeol-

ogy. In F. A. Exum and G. R. George (eds.), Geology and Mineral Resources of the Big-horn Basin, pp. 19-35. Wyoming Geologi-cal Association 27th Annual Field Confer-ence Guidebook for 1975.

Godfrey-Smith, D.I., and N. Haywood1984 Obsidian Sources in Ontario Prehistory. Ontario Archaeology 41: 29-35.

Griffin, J.B.1965 Hopewell and the Dark Black Glass. In J.

A. Fitting (ed.), Papers in Honor of Emer-son E. Greenman, pp. 115-155. Michigan Archaeologist 11 (3-4). Ann Arbor.

Griffin, J.B., A.A. Gordus, and G.A. Wright1969 Identification of the Sources of Hopewellian

Obsidian in the Middle West. American An-tiquity 34:1-14.

Holmes, W.H.1879 Notes on an Extensive Deposit of Obsidian

in the Yellowstone National Park. Ameri-can Naturalist 13:247-250.

Hughes, R.E.1984 Obsidian Sourcing Studies in the Great Ba-

sin: Problems and Prospects. In R. E. Hughes (ed.), Obsidian Studies in the Great Basin, pp. 1-19. Contributions of the University of California Archaeological Research Facility No. 45. Berkeley.

1988 The Coso Volcanic Field Reexamined: Im-plications for Obsidian Sourcing and Hy-dration Dating Research. Geoarchaeology 3:253-265.


1994a Intrasource Chemical Variability of Arte-fact-Quality Obsidians from the Casa Dia-blo Area, California. Journal of Archaeo-logical Science 21:263-271.

1994b X-ray Fluorescence Analysis of Obsidian from the Roosevelt Narrows Site, Outing, Minnesota. Geochemical Research Labo-ratory Letter Report 94-82. Portola Valley, CA.

1995a Source Identification of Obsidian from the Trowbridge Site (14WY1), A Hopewellian Site in Kansas. Midcontinental Journal of Archaeology 20:105-113.

1995b X-ray Fluorescence Analysis of Geologic Obsidian Samples from Yellowstone and Grand Teton National Parks, Wyoming. Geochemical Research Laboratory Letter Report 95-8, Portola Valley, CA.

1998 On Reliability, Validity, and Scale in Ob-sidian Sourcing Research. In A. F. Rame-nofsky and A. Steffen (eds.), Unit Issues in Archaeology: Measuring Time, Space, and Material, pp. 103-114. University of Utah Press, Salt Lake City.

1999 X-ray Fluorescence Analysis of Obsidian Flakes from Archaeological Site 39PO61,

Located Along the Missouri River in North-Central South Dakota. Geochemical Re-search Laboratory Letter Report 99-94. Portola Valley, CA.

2002 X-ray Fluorescence Analysis of Obsidian from Two Archaeological Sites (21-BL-37 and 21-CA-586) in Northern Minnesota. Geochemical Research Laboratory Letter Report 2002-39. Portola Valley, CA.

2006a The Sources of Hopewell Obsidian: Forty Years After Griffin. In D. K. Charles and J. E. Buikstra (eds.), Recreating Hopewell,

pp. 361-375. University Press of Florida, Gainseville.

2006b Energy-Dispersive X-ray Fluorescence Analysis of Obsidian from Various Archae-ological Sites and Localities in South Dako-ta. Geochemical Research Laboratory Let-ter Report 2006-103. Portola Valley, CA.

2007a Energy-Dispersive X-ray Fluorescence Analysis of Obsidian from 39PO79, South Dakota. Geochemical Research Laboratory Letter Report 2007-32. Portola Valley, CA.

2007b Provenance Analysis of Obsidian. In G. C. Frison and D. N. Walker (eds.), Medicine Lodge Creek: Holocene Archaeology of the Eastern Big Horn Basin, Wyoming, pp. 231-244. Clovis Press, Avondale, CO.

Hughes, R.E., and A.C. Fortier1997 Identification of the Geologic Sources

of Obsidian Artifacts from Three Middle Woodland Sites in the American Bottom, Il-linois. Illinois Archaeology 9: 79-92.

Hughes, R.E., M. Kay, and T.J. Green2002 Geochemical and Microwear Analysis of an

Obsidian Artifact from the Brown Bluff Site (3WA10), Arkansas. Plains Anthropologist 47:73-76.

Hughes, R.E. and W.B. Lees1991 Provenance Analysis of Obsidian from Two

Late Prehistoric Archaeological Sites in Kansas. Transactions of the Kansas Acad-emy of Science 94: 38-45.

Hughes, R.E., and F.W. Nelson1987 New Findings on Obsidian Source Utiliza-

tion in Iowa. Plains Anthropologist 32:313-316.

Hughes, R.E. and M.G. Pavesic2005 Proximity and Provenance: A Lesson from

the Sterling Cache, Idaho. Journal of Cali-


fornia and Great Basin Anthropology 25: 221-225.

Hughes, R.E., and D.C. Roper1999 Source Area Analysis of Obsidian Flakes

from a Lower Loup Phase Site in Nebraska. Plains Anthropologist 44:77-82.

Iddings, J. P.1888 Obsidian Cliff, Yellowstone National Park.

In Seventh Annual Report of the U.S. Geo-logical Survey for 1885-1886, pp. 249-295. Washington, D.C.

Jack, R.N.1976 Prehistoric Obsidian in California I: Geo-

chemical Aspects. In R.E. Taylor (ed.), Ad-vances in Obsidian Glass Studies: Archaeo-logical and Geochemical Perspectives, pp. 183-217. Noyes Press, Park Ridge, NJ.

Jackson, T.L.1989 Late Prehistoric Obsidian Production and

Exchange in the North Coast Ranges, Cali-fornia. In R. E. Hughes (ed.), Current Di-rections in California Obsidian Studies, pp. 79-94. Contributions of the University of California Archaeological Research Facility No. 48. Berkeley.

Logan, B., R.E. Hughes, and D.R. Henning2001 Western Oneota Obsidian: Sources and Im-

plications. Plains Anthropologist 46:55-64.

Macdonald, R., R.L. Smith, and J.E. Thomas1992 Chemistry of the Subalkaline Silicic Obsid-

ians. U.S. Geological Survey Professional Paper 1523.

McKern, W.C.1931 A Wisconsin Variant of the Hopewell Cul-

ture. Bulletin of the Public Museum of the City of Milwaukee 10: 185-328.

Nelson F.W., Jr.1984 X-ray Fluorescence Analysis of Some

Western North American Obsidians. In R.

E. Hughes (ed.), Obsidian Studies in the Great Basin, pp. 27-62. Contributions of the University of California Archaeological Research Facility No. 45. Berkeley.

Shackley, M.S.2005 Obsidian: Geology and Archaeology in the

North American Southwest. University of Arizona Press, Tucson.

Stoltman, J.B., and R.E. Hughes2004 Obsidian in Early Woodland Contexts in the

Upper Mississippi Valley. American Antiq-uity 69: 751-759.

Stross, F.H., T.R. Hester, R.F. Heizer, and R.N. Jack

1976 Chemical and Archaeological Studies of Mesoamerican Obsidians. In R.E. Taylor (ed.), Advances in Obsidian Glass Stud-ies: Archaeological and Geochemical Per-spectives, pp. 240-258. Noyes Press, Park Ridge, NJ.

Struever, S., and G.L. Houart1972 An Analysis of the Hopewell Interaction

Sphere. In E. N. Wilmsen (ed.), Social Ex-change and Interaction, pp. 47-79. Muse-um of Anthropology, University of Michi-gan, Anthropological Papers No. 46. Ann Arbor.

Thomas, C.1894 Report on the Mound Explorations of the

Bureau of American Ethnology. Twelfth Annual Report of the Bureau of American Ethnology for 1890-’91. Smithsonian Insti-tution, Washington, D.C.

White, G.G., and E. Allison2002 Projectile Points. In G. G. White, Cultural

Diversity and Cultural Change in Prehis-toric Clear Lake Basin: Final Report of the Anderson Flat Project, pp. 221-260. Center for Archaeological Research at Davis, Pub-lication No. 13. Davis, CA.


White, G.G., D.A. Fredrickson, and J. Rosenthal2002 Archaeology. In G. G. White, Cultural Di-

versity and Cultural Change in Prehistoric Clear Lake Basin: Final Report of the An-derson Flat Project, pp. 41-52. Center for Archaeological Research at Davis, Publica-tion No. 13. Davis, CA.

Willingham, C.G.1995 Big Table Mountain: An Obsidian Source in

the Centennial Mountains of Eastern Idaho. Idaho Archaeologist 18: 3-7.

Winchell, N.H.1911 The Aborigines of Minnesota: A Report

Based on the Collections of Jacob V. Brow-er, and on the Field Surveys and Notes of Alfred J. Hill and Theodore H. Lewis. Min-nesota Historical Society, St. Paul. The Pio-neer Company.

Notes

1. Among the materials housed at the Minnesota His-torical Society (MHS) are obsidian artifacts donated from various private collections. Perhaps the best known, from the standpoint of this study, is the ob-sidian specimen illustrated by Winchell (1911:Plate 1, no. 29; MHS cat. no. 28.77.1). Three specimens from the Harry Ayer collection from the Mille Lacs Trading Post (MHS cat. nos. 1988.120.38.51, 52, and 53) and a biface fragment from the collection of avocational archaeologist Dr. A. N.”Tony” Romano (MHS cat. no. Romano-1) also were subjected to edxrf analysis. Though all five of these specimens were analyzed and elemental data are on file at Geo-chemical Research Laboratory, the results are not included here because there are reasons to believe that the specimens very likely represent artifacts ob-tained during historic times by contemporary artifact collectors/traders. For example, trace element data show that the artifact illustrated by Winchell (1911:Plate 1, no. 29) from the Jacob Brower collection was manufactured from obsidian of the Napa Valley

chemical type, erupted in the North Coast Range of northern California (comparative chemical data in Bowman et al. 1973: Table 1; Jack 1976: Table 11.3; Stross et al. 1976: Table 13.2; and Jackson 1989: Table 2), as was a long, slender lanceolate biface (MHS cat. no. 1984.41.22.8) from the same col-lection. Likewise, a small serrated corner-notched projectile point (MHS cat. no. 1988.120.10.4) from the Ayer collection (Mille Lacs Trading Post) also was manufactured from Napa Valley obsidian. In all three cases, typological affinities suggest prehistoric Californian origin for the specimens. The projectile point illustrated by Winchell (1911:Plate 1, no. 29) from the Brower Collection, though upside down, is a clear example of a shouldered lanceolate (Ex-celsior) point (cf. White and Allison 2002: Figures 88-94), diagnostic of the Houx Aspect of the Berke-ley Pattern, dated to between 7,000 – 1,500 B.P. (White et al. 2002: 51) in California’s North Coast Range, while the small serrated corner-notched pro-jectile point from the Ayer Collection (MHS cat. no. 1988.120.10.4) is an example of a form diagnostic of Middle/Late Phase 1 times (ca. 1,100 – 1,500 A.D.) in central California (Bennyhoff 1994: Figure 6.2, no. 48). In both cases there are, as far as I am aware, no known typological counterparts from these time periods in Minnesota. Equally revealing is the fact that, despite its dominant use in prehistoric central California, no obsidian of the Napa Valley chemi-cal type has yet been documented from a bona fide prehistoric archaeological context more than a few kilometers east of the Sierra Nevada Mountains.

These artifacts are not the first to be identified as be-ing made from Napa Valley obsidian. In reviewing obsidian sourcing results from Wisconsin, Griffin (et al. 1969:13) wrote:

An obsidian knife (1330) given to the Na-tional Museum was identified as a surface find from Honey Creek, 5 to 6 mi. west of Prairie du Sac, Sauk County. This specimen is made from material from the Napa Valley, California, source, and Griffin doubts that it was used, or even seen, by Indians in Wis-consin.


Griffin obviously was convinced that the Napa Val-ley obsidian find in Wisconsin was not the byprod-uct of a genuine prehistoric trade/exchange event, but rather the result of activity, or activities, by non-prehistoric person(s) who, during Historic times, conveyed the artifact from its original archaeologi-cal context to (in this case) an Historic collection in Wisconsin. Geochemical, typological, historical, and archaeological distribution data at hand suggest that the Napa Valley obsidian artifacts identified in these Minnesota private collections also were ob-tained by various means during the Historic period and that they do not represent authentic prehistoric occurrences in the state.

Geochemistry, typology, and collection history also cast doubt on the local authenticity of other speci-mens from the Ayer Collection (MHS cat. nos. 1988.120.38.51, 52, and 53). In this case these three specimens (a polyhedral core [MHS cat. nos. 1988.120.38.51] and two prismatic obsidian blades each ca. 12 cm. long [MHS cat. nos. 1988.120.38.52 and 53]) were manufactured from Mesoamerican obsidians. The morphological (i.e. typological) af-finities of these specimens are unmistakably Me-soamerican, as are their obsidian sources of origin. Since “Ayer is known to have made collecting trips to the American Southwest” (Patricia Emerson, per-sonal communication) local prehistoric significance for these specimens is highly questionable.

2. Although Y/Zr ppm measurements effectively separate Obsidian Cliff glass from all others identi-fied in this study, they overlap slightly with values for Obsidian Ridge (a.k.a. Cerro Toledo Rhyolite) located in the Jemez Mountains of northern New Mexico (Shackley 2005: Table A.5). However, Nb values for Obsidian Ridge material are typically 90–100 ppm, while those for Obsidian Cliff range from ca. 40-45 ppm. Likewise, even though Obsid-ian Cliff is superficially similar to another obsidian erupted in the Jemez Mountains (Cerro del Medio, a.k.a. Valles Rhyolite) on the basis of Y/Zr (Shack-ley 2005: Table A.4), the two are easily distin-guished on the basis of Rb ppm contrasts. Obsidian Cliff contains ca. 100 ppm more Rb than does Cerro del Medio (cf. Macdonald et al. 1992: Appendix I, pp. 148-149).

the geologic sources for obsidian artifacts

Documents