Obsidian Source Characterization and Hydration Analysis at the Connley Caves (35LK50) in South-Central Oregon
 |
| Unit 6 at Connley Caves |
by Dennis L. Jenkins
Director, Northern Great Basin Field School
Museum of Natural and Cultural History
University of Oregon
The Connley Caves are located near Paulina Marsh in the Fort Rock Basin of Central Oregon. UO field school investigations at the Caves have identified seven primary strata. The upper 4 strata are primarily colluvium composed of angular pinkish-tan volcanic tuffs and dark gray andesites. Reworked Mazama tephra is liberally mixed in with these deposits to a depth of roughly 170 cm. Radiocarbon ages for these deposits range from ca. 5300 to 200 cal. BP. Below these strata, between 170 and 190 cm, is Stratum 5, a more massive quantity of Mazama tephra 10 to 20 cm thick which has generally been disturbed by rodent and anthropic activity, but occasionally preserved in nearly pure quantities. Stratum 6 is a dark, moist, sandy-silt colluvium deposit below the thick Mazama tephra. Radiocarbon ages for these deposits range from 7700 to 13,100 cal. BP.
At the bottom of the site is Stratum 7, a sandy-gravelly Pleistocene beach deposit overlying bedrock. The presence of rounded beach gravels which originated from Stratum 7 in the overlying strata is a measure of the amount of mixing that has occurred at the site. In other words, rounded beach gravels generally increase with depth at the site and there is some quantity of Stratum 7 gravels in all strata other than a few small pockets of pure Mazama tephra. The predominant cause for this mixing is the anthropic movement of cave sediments throughout the period of human occupations, though upward movement of cave sediments by animals has also certainly occurred.
Obsidian hydration dating depends in large measure on the comparison of associated radiocarbon ages with obsidian hydration rind measurements. To be accurately employed, this process requires the use of calibrated radiocarbon dates, which tend to be more linear than uncalibrated radiocarbon dates. All dates and ages cited in this paper are calibrated years BP.
 |
| Annalisa Romano, Unit 11, D, Cave 5 |
Radiocarbon dating of the Connley Cave deposits must be approached with extreme caution because their floors are littered with branches, twigs, bones, and pellets deposited by raptors, introducing rich accumulations of non-cultural debris to the deposits. However, recent developments in sampling techniques, more precise measurement of the C-14 half-life, and calibrating fluctuations in the long-term atmospheric C-14 curve, have significantly improved the accuracy and precision of radiocarbon dating. These developments allow Connley Cave C-14 dates to be calculated in actual years before the present, a necessary step before they can be compared to obsidian hydration ages.
Bedwell Stephen (1973) reported 21 radiocarbon dates for the Connley Caves. When sampling, he selected large charcoal pieces collected from 10 cm levels exhibiting high concentrations of cultural debris. In most cases, he was not directly involved in the collection of the samples, and there is no evidence that he even considered the possibility of either 'old carbon' contamination or that the charcoal he was dating might not have been associated with the cultural materials he thought he was dating. Thus, there is a high probability that carbon introduced into the site or relocated within the site by natural and cultural causes were included in some of his charcoal samples.
Bedwell identified four cultural units or periods in the Fort Rock Basin. He identified Period I - roughly 16,000 to 13,000 cal. BP, Period II - 13,000 to 7700 cal. BP, Period III - 7700 to 5300 cal. BP, and Period IV 5300 to 3000 cal. BP. Of these, the Connley Caves included the last three periods dating from 13,000 to 3000 BP, though Period III is a time when the caves were not occupied by humans. Three of Bedwell's radiocarbon dates, 3370 cal. BP, 5370 cal. BP and the oldest at 13,110 cal. BP were stratigraphically out of sequence, leaving 18 dates which may be applied to dating the cultural components of the site. Nine recently acquired UO field school AMS dates duplicate a portion of his record and extend the documentation of occupations forward from the end of Period IV at 3000 BP into the Historic period. These AMS dates were generally acquired from tiny samples of sagebrush charcoal human coprolites, and perishable artifacts, though one sample, dated at 200 cal. BP, was from a marine shell bead.
 |
| In-situ basketry and cordage fragments |
Bedwell identified two pre-Mazama components, the early pre-ash Period from 13,000 to 9000 cal. BP and what he termed the late pre-ash Transitional Period from 9000 to 7700 cal. BP. Our radiocarbon dates tend to verify the division of the Stratum 6 cultural deposits. The Upper pre-Mazama dating between 9000 and 8000 cal. BP and the Lower pre-Mazama dating between 10,000 and 11,000 cal. BP. While our Lower pre-Mazama component samples come from the base of the best deposits, they may not date the oldest occupations at the site because we were limited to working outside the caves where older carbon may not have survived well. Further dating may shed light on this topic and this is one area in which obsidian hydration may contribute substantially to dating the site deposites. To be reliable, obsidian hydration requires obsidian source characterization of each specimen to be dated since volcanic sources exhibit unique chemical compositions and each source generally hydrates at its own rate of speed. Thatcher reported obsidian source characterization of 168 artifacts collected by Bedwell at the Connley Caves. She identified 31 different obsidian sources in this sample. The Silver Lake/Sycan Marsh (SL/SM) and Cougar Mountain obsidian sources are located in the Fort Rock Basin closest to the site and predictably exhibit the largest samples, comprising 17 and 20% of the entire sample, respectively.
Recent UO obsidian studies include 86 tools and 96 flakes. The sampling design employed in the selection of these samples was intended to provide for the comparison of obsidian source and hydration distributions through time. Two to five specimens were selected from alternating levels in caves 5 and 6. Tools and flakes were selected from even numbered levels from the surface to the bottom of Cave 5 excavation units. In Cave 6, where mixing of the upper post-Mazama deposits was a serious concern, tools alone were selected from upper deposits, and debitage was sampled only from below 165 cm.
 |
| Front of Connley Cave #6 |
The obsidian source data is quite revealing of the varying patterns of material source use through time. Among tools, the two largest local obsidian sources, Cougar Mountain and Silver Lake/Sycan Marsh compose the majority of specimens, although non-local obsidian makes up about 44% of the sample. The debitage samples exhibit a much different pattern. Cougar Mountain composes 63% of the sample and the Silver Lake/Sycan Marsh source comprises the next largest sample at 14%.
The flake sample from Cave 5 shows a strong preference for Cougar Mountain obsidian, though the amount of this source steadily decreases with depth. From 0 to 100 cm Cougar Mountain composes 96% of the sample. Between 100 and 170 cm it composes 80%, from 175 to 235 cm 57%, and from 235 to 325 cm only 44% of the sample. In other words, there is a steadily strengthening obsidian procurement pattern through time, with Cougar Mountain obsidian arriving at the Connley Caves in much larger quantities during the Late and Middle Holocene than during the Early Holocene.
Obsidian hydration offers a method of directly addressing the age of individual obsidian specimens. The sample of obsidian hydration readings currently available offers good evidence that obsidian hydration dating will work very well at the Connley Caves. However, the acceptance of hydration ages must be tempered with the knowledge that the hydration rind thickness of any particular specimen is more than just a measure of its age.
Hydration rinds form in response to many factors, and temperature seems to be one of the most important. Hydration rinds form more quickly as ambient temperatures increase, and more slowly as they decrease. Thus, the varying post-depositional histories of individual artifacts affects the cumulative rate of hydration for each. In theory then, each hydration measurement represents the history of the individual specimen. Artifacts on or near the surface hydrate more quickly than deeply buried artifacts, and artifact classes that tend to be reused by later occupants may remain nearer the surface for longer periods of time. As an example, projectile points were commonly scavenged from archaeological site deposits and reused by later site occupants in Fort Rock. This had a significant affect on the individual hydration readings with the result that projectile points tend to exhibit thicker hydration rinds than the debitage they accompany. In the following analysis only the most congruent hydration readings will be accepted for each artifact class.
Calculating the proper rate of hydration for application at the Connley Caves was primarily a function of comparing hydration means to the two most recently and reliably radiocarbon dated pre-Mazama components in Cave 5. The most pertinent specimens come from levels between 175 and 210 cm for the Upper pre-Mazama component, and between 240 and 280 cm in the Lower pre-Mazama component. Upper pre-Mazama specimen measurements ranged from 4.1µ to 7.3µ. Recently obtained AMS dates for these deposits suggest that they date from about 8560 to 8980 cal. BP, and exhibit a mean age of about 8850 cal. BP. Employing the hydration rate of 3.3µ/1000 yrs. suggested by Pettigrew and Hodges indicates a range of 5.3µ to 5.5µ, with a mean reading of 5.4µ producing an age of 8840 BP. It would seem most congruent to consider all specimens with hydration rinds <5.1µ and >5.8µ to be out of stratigraphic sequence. This process results in a sample of seven specimens exhibiting a mean of 5.4µ. The most pertinent Lower pre-Mazama component sample is composed of specimens recovered from between 235 and 280 cm. These specimens exhibited hydration measurements ranging from 5.0µ to 7.7µ. Rejecting hydration readings <5.8µ and >6.6µ as out of sequence, leaves a sample of 11 specimens exhibiting a mean of 6.4µ. A recently obtained AMS date suggests that these deposits date from 10,940 to 11,070 cal. BP, with a mean age of 10,670 cal. BP. Employing the hydration rate noted above indicates that the range of 6.0µ to 6.6µ hydration readings for the Lower pre-Mazama sample dates these deposits to ages between 10,900 and 13,200 cal. BP. Bedwell's oldest radiocarbon date from this site exhibits a mean calibrated age of 13,110 cal. BP. However, this sample is stratigraphically out of sequence. His next oldest acceptable date exhibits a range of age from 11,960 to12,910 cal. BP. Once again, the obsidian hydration rate proposed by Pettigrew and Hodges (1995:2-16) for south-central Oregon sources seems to fit remarkably well.
Having demonstrated that obsidian hydration works in the relatively undisturbed pre-Mazama deposits we may now continue the process for the potentially more disturbed post-Mazama deposits. Specimens recovered from 0 to 100 cm produced readings ranging from 1.4µ to 4.5µ. The mean of these measurements is 3.4µ, suggesting an average age for these deposits of about 3500 BP, consistent with the latest Middle Holocene occupation indicated by Bedwell's radiocarbon dates. However, it is clear that there are significant clusters of hydration readings at 1.4µ, 2.3µ, 2.8µ, 3.5µ and 4.4µ suggesting individual events at about 600, 1600, 2300, 3500, and 5350 BP within this group of 32 specimens.
Between 100 and 170 cm readings range from 3.4µ to 7.7µ. Disregarding samples <4.5µ because they should generally be associated with deposits above 100 cm, and samples >5.2µ generally associated with deposits below 170 cm, leaves only three readings with a mean of 5.1µ suggesting a date of 7770 cal. BP for these post-Mazama deposits. Thus, it appears as if all of the samples recovered between 100 and 170 cm originated with either the upper or lower deposits and that the obsidian hydration results support abandonment or extremely light occupation of the Connley Caves between, as indicated by Bedwell . It is somewhat tantalizing then to consider the final set of hydration readings from specimens recovered between 280 and 330 cm. This batch of 10 pieces of debitage was drawn from a very small total assemblage. Hydration readings range from 5.5µ to 7.7µ. Disregarding samples <7.3µ, because these measurements are generally associated with cultural deposits located above 280 cm, leaves three specimens with a mean of 7.6µ suggesting an age of 17,500 BP.
In conclusion, assessing the results of obsidian hydration analyses at the Connley Caves strongly indicates the prehistoric mixing of deposits attested to by the presence of rounded beach gravels scattered throughout the deposits. I would contend that this is the nature of deposits in caves because they are frequently altered by site occupants, both human and animal. Artifacts belonging with older occupations are moved up through cave sediments by later excavations while younger materials are introduced into older deposits. However, after all else is considered, it is clear that obsidian hydration dating holds promise for dating at the Connley Caves. Younger artifacts are found at the top of the deposits and older artifacts tend to be found at greater depth, an unusual situation among my past experiences.
Bibliography
| Bedwell, Stephen F. | |
| 1970 | Prehistory and Environment of the Pluvial Fort Rock Lake Area of South Central Oregon. Ph.D dissertation, Department of Anthropology, University of Oregon. |
| Largaespada, Leah L. | |
| 2001 | From Sand and Sea: Marine Shell Artifacts from Archaeological Sites in the Fort Rock Basin, Northern Great Basin. Master's Thesis, Department of Anthropology, University of Oregon, Eugene. |
| Pettigrew, Richard M. and Charles M. Hodges | |
| 1995 | Prehistoric Hunter-Gatherer Land-Use Systems: Pacific Northwest. In Archaeological Investigations PGT-PG&E Pipeline Expansion Project Idaho, Washington, Oregon, and California, Michael J. Moratto, General Editor. Volume IV: Synthesis of Findings, Randall F. Schalk, Volume Editor, pp. 2-1 to 2-70. Pacific Gas Transmission Company, Portland, Oregon. |
| Puseman, Kathryn | |
| 2000 | Identification of Charcoal from Connley Caves, 35LK50, Oregon. Paleo Research Labs Technical Report 00-73, Paleo Research Laboratories, Golden, Colorado. |
| Ridings, Rosanna | |
| 1991 | Obsidian Hydration Dating: The Effects of Mean Exponenial Ground Temperature and Depth of Artifact Recovery. Journal of Field Archaeology 18:77-85. |
| 1996 | Where in the World does Obsidian Hydration Dating Work? American Antiquity 61(1):136-148. |
| Skinner, Craig E. and Jennifer J. Thatcher | |
| 2001 | Tables A-1 and B-1. Results of XRF Studies, and Obsidian Hydration Results and Sample Provenience: 35-LK-50, 35-LK-2758, and 35-LK-2737, Lake County, Oregon. |
| 2002 | Tables A-1 and B-1. Results of XRF Studies, and Obsidian Hydration Results and Sample Provenience: Connley Caves (35-LK-50), Lake County, Oregon. |
| Stuiver, Minze and Paula J. Reimer | |
| 1993 | Extended 14C Data Base and Revised Calib 3.0 14C Age Calibration Program. Radiocarbon 35(1): 215-230. |
| Thatcher, Jennifer J. | |
| 2000 | The Distribution of Geologic and Artifact Obsidian from the Silver Lake/Sycan Marsh Geochemical Source Group, South-Central Oregon. Master's Thesis, Departments of Anthropology and Geography, Oregon State University, Corvallis. |