OBSIDIAN HYDRATION ANALYSIS 79 OB3SIDIAN HYDRATICN DATING AND FIEL SITE TEMPERATURE Fred Trembour Irving Friedkan Introduction The study of obsidian hydration has led to one of the newest dating methods available to investigators in archaeology and the earth sciences. Following its announcement in 1960 (Friedman and Smith 1960; Evans and Meggars 1960), the technique quickly attracted other participants in both application and fur- ther development (Clark 1961; Michels 1967; Meighan, Foote and Aiello 1968; Johnson 1969; Layton 1972; Friedman et al. 1973). This dual pursuit continues to this day and there is still much work to be done to bring out the full potential of hydration analysis as an absolute dating tool (Friedman and Trembour 1978). This paper is devoted to a discussion of the important factor of field temperature as it relates to interpreting the age of hydratimng obsidian. Any consideration of the hydration topic must keep the fundamentals of the method in mind. Hydration is a form of geologic weathering that begins when the pristinc glass first encounters the environment and continues as a pro- gressive thickening of the hydrated layer. Under ambient conditions the hydration remains attached and integral with the artifact substrate for periods far longer than the span of archaeological time in the New World. The basic laboratory measurement made (usually in micrometers, u m) is the depth of the hydrated rind on a surface in question. It has been found that various obsidians possess specific hydration rates which are a function of their differing chemical compositions and some of these may vary by as much as a factor of 20 to 1 (Friedrian and Iong 1976). For any given obsidian and hydration temperature, however, the rind growth will proceed as the square root of time. The hydration rate for any particular obsidian rises exponentially with increasing temperature, following the Arrhenius diffusion equation (Figure 1). Thus, converting observed hydration depth of an obsidian sample into age terms requires 1) an estimate of the intrinsic hydration rate of the material (usually expressed as u m2/103 yrs.) and 2) an estimate of the effective hydration temperature in the thermal his- tory of the sample piece. In practice, all samples in close association at an archaeological site are assumed to have had the same thermal histories. As a simple case in relative dating, therefore, we can see that if two facets of the same artifact show different rind thicknesses, their relative ages are in direct ratio of the squares of these thickness measurements. Temperature and Hydration Rate According to the law that governs rate of hydration, increasing temperature causes a faster than linear rise in hydration rate, amounting to about a 10. increase in rate for each 1?C increase in ambient temperature. Hence at fluc- tuating (daily and seasonal) conditions the effective hydration temperature is 80 not its simple arithmetic mean but some higher integrated value (Norton and Friedman 1981). The more closely this value is estimated at any particular archaeological site, the more reliable will be the hydration age determination. Table 1 refers to the temperature dependent hydration properties of obsi- dian from a source outcrop near American Falls, Idaho which at 10.00C has a hydration rate of 2.4 um2/1000 years (Friedman and Long 1976). As the temper- ature is raised or lowered by small increments, the hydration rate changes by about 0.03 um2/1000 years per 0.10C. An error of 1?C in the hydration temper- ature of a 2000 year old specimen, for example, leads to age assignments between 1740 and 2200 years. The importance of obtaining accurate field temperature estimates may be better appreciated if temperature and time are considered as equal determinants in the formation of hydration rinds. In fact, we may calcu- late it the other way: if the "true" age of an obsidian artifact were assured by independent means (i.e. by a direct association between the artifact and a radiometrically dated sample), then the effective hydration temperature at this particular archaeological site could be derived fram the hydration depth. Table 2 is a selected listing of western North American obsidian sources for which hydration rates and temperature have been published. Some of these supplied the obsidian recovered from archaeological sites in the Great Basin (e.g. Hughes 1983). Tb estimate the natural ambient temperature level at archaeological sites it has become customary to resort to published long-term mean annual air temper- ature values recorded at suitable nearby national weather stations. By comparing data from two or more stations a correctional factor for altitude differentials may be calculated and applied for the elevation of the archaeological site. An additional temperature refinement can be applied for subsurface locations. It should be noted, however, that this approach usually ignores the set of other variables that may contribute to the important concept of microclimate variation: vegetation and tree cover, direction and degree of terrain slope, snow cover and certain others. Because the hydration process is so clearly temperature-sensitive, a word of caution is important here. We have found that even short exposures to abnorally high heat can severely distort the outcome of the dating analysis. Thus, surface-collected specimens should always be distinguished from excavated ones. When obsidian is being recovered in the field, the archaeologist should carefully note and record a associated signs of burning at the findspot, such as hearths, house fires, field and forest fires, etc. On-Site Temperature Measurements Obviously, post-discovery temperature determinations at selected archaeo- logical sites could go far to exclude the imponderables that were referred to in the problem of microclimate assessment. What is needed is a means of obtaining an integrated record of ambient fluctuations at the locality over a sufficiently long time period, say a year or more. Preferably, these particular localities should then be compared with the records of a suitable weather station for the same period to permit adjustments if the test period has been abnormal with respect to the regional long-term mean. 81 Although recording instrumentation for this field purpose is available on the commercial market, its use poses problems: high initial expense, power supply, maintenance and protection against disturbance and natural damage. An early attempt to achieve simplicity and economy for field applications of this kind was published by Pallmann et al. (1940; see O'Brien 1971), who used the method in forestry tree growth research in Switzerland. They devised a small glass ampoule containing a sucrose-water solution at a controlled pH value which inverts to glucose and fructose at a rate dependent on temperature. The extent of inversion after a period of time is measured as optical rotation of a beam of light with a polarimeter. Laboratory calibration enables conversion of the rotation reading to a mean integrated temperature figure. The Pallman device was the first integrating ambient temperature sensor to be free of all field support needs. More recently, another principle, water diffusion through a permeable mem- brane or partition, was applied to the same purpose (Ambrose 1976) to date obsi- dian artifacts in the South Pacific Islands. Ambrose used a spherical shell (or cell) of epoxy resin, filled with a synthetic zeolite desiccant, surrounded by water. In this manner a 100%. H20 vapor pressure differential was maintained across the permeable cell wall, and the rate of diffusion and entrapment of water was a function only of temperature under specific conditions. With this assembly, weight change of the plastic desiccant container due to water uptake over a period of time becomes a measure of the integrated value of the fluctuating environmental temperature. Conversion of weight to a mean temperature figure is obtained by laboratory calibration of the cell at known temperatures. It should be noted that the sensor operation is entirely maintained by ambient energy, and that neither a support system nor service attention are required between deployment and retrieval in the field. Emplacement of the units in the open air, water or soil are equally feasible. The effective field life or capacity of a cell varies with design; runs of a year or more are commonly carried out. The assemblies are inexpensive -- in the ten dollar range -- small and inconspicuous. Ordinary tools and supplies serve for fastening the devices to trees or burying in soil. Our laboratory has worked with both the Pallmnn and Ambrose type integrating temperature cells for some years. Figure 2 depicts the parts of a Pallmann assembly, including the sealed glass ampoule and the protective polyvinyl chloride 3/4" pipe length for ground emplacement in a prebored hole. The plastic probe may be up to 2 meters (6.6 ft.) long and contain several sensors for operation at various depths. Figure 3 shows an original spherical Ambrose diffusion cell and our cylin- drical modifications of acrylic resin and stainless steel designed to fit into tubular ground probes like those described above. The sensor cells may, of course, be either desiccant-filled or water-filled, with the other component on the outside. We have found that the water-filled model is preferable for very long field runs and/or warm situations, and that the desiccant-filled kind is better suited for places that are subject to frequent freezing and thawing cycles. The cells with water and desiccant components are enclosed in a capped 82 polypropylene jacket tube (which is nearly impermeable to water) before loading into the protective ground probe. Figure 4 shows a short ground probe assembly for horizontal positioning close to the surface. To suppress abnormal vertical heat transfer in operation, the empty sections of a probe pipe are filled with insulation, and excess space in the bore hole is refilled with earth. As implied earlier, two weighings of the sensor on a chemical balance (just before and after the field exposure) suffice to establish the mean diffusion rate for the temperatures prevailing at the site. This value is usually expressed as milligrams per day. The cells of a given design have been labora- tory calibrated under controlled conditions beforehand, from which a conversion graph is prepared (Figure 5). From the appropriate weight change on the ordinate axis of this plot, the "Ambrose temperature" can be read at the abscissa. The temperature-diffusion rate continuity of the cell operation is virtually unaf- fected in crossing the water freezing point, and cell contact with either liquid H20 or saturated vapor serves eqlually well as a water source at any temperature. The sensitivity of the temperature measurement of the Ambrose cell is between 0.10 - 0.20C, depending on the temperature concerned. It will be observed from Figure 5 that the cell's weight gain rate rises exponentially with temperature, much like the behavior of the rise in hydration rate of obsidian with temperature. Thus an adjustment from "Ambrose temperature" to the linear "effective temperature" scale is required. This adjustment must be made based on the annual temperature fluctuation range. For ambient air temperature at a site, this range can be estimated adequately using published maxima and minima from a relevant weather station. Some weather stations also record soil temperatures which vary in narrower regular annual cycles. In the absence of such data for an archaeological site, tmw instantaneous temperature readings made at the expected semiannual occurrences of soil maximum and minimum can provide useful range figures. Norton and Friedman (1981) have published a series of conversion graphs to derive "effective temperature" for both the Pallmann and Ambrose integrated means and for various temperature levels and ranges of fluctuation. The calculated relationships are based on measured activation energies of the reactions and assumed sinusoidal temperature vari- ation within the ranges. An example is given in Figure 6; here it can be observed that for a mean of 140C and a range of 17.50C, the effective temperature is 150C for an Ambrose reading and 16.60C for a Pallmann reading. Conclusions This treatment of long-term integrated field temperature determination has focused on its application for obsidian hydration dating, particularly for archaeological purposes. Of the methods considered, the thermal diffusion cell of Ambrose is favored because of its comparative simplicity, compactness, ruggedness and economy. These advantages make it attractive for liberal use at particular archaeological sites, and in numbers for more thorough exploration of extensive areas. It should be as useful in aiding the temperature-sensitive anino acid racemization dating technique as for obsidian hydration. Many other environmental study applications can be predicted for the device, such as in biology, pedology and climatology. 83 The Arrhenius equation relating diffusion rate to temperature: k = Ae -E/R where k is the obsidian hydration rate (micrometers squared per 103 yrs.) A is a constant E is the activation energy of the hydration process (calories/mole) R is the gas constant (calories per degree per mole) T is the absolute temperature in degrees Kelvin Figure1. The Arrhenius equation and the definition of its terms. PVC cm. 0 1 2 Figure 2. A Pallmann sugar inversion ampoule and ground probe elements for integrated temperature measurements. I I q ?9 A. cm. 0 1 2 Figure 3. An original spherical Ambrose water diffusion cell and cylindrical nndifications apolied for integrated ground temperature determinations over long time periods. PVC - cm 0 2 4 A short ground probe of PVC pipe with fittings that houses an activated water diffusion cell encased in protective polypropylene tubing that contains zeolite desiccant. 84 Figure 4. I I I i I i . a I ",NI: I I I I If I I I I e-11- 85 2.00 . , 1. 75 1.50 1.25 1.00 .~ ~~~~~~ // 4) 0.75// 0.50 - -, 0.25 p 3 3 3 * I i I * I * I a I I * I 0 5 10 15 20 25 Temperature, 0C Figure 5. Typical time-weight change curves for water diffusion (Ambrose) cells of acrylic resin at normal temperatures and for two 110 partial pressure differentials: 100% - 50% 86 21 O S w X0 U, 19~~~~~~~~~~~~ 1S~~~~~~~~~~~~~~~1 ~~H5 ___~~ 2 2 __ Teerature range, in degrees Celsius Arithmetic Mean = 140C Figure 6. Effective temperature versus fluctuation range for arithmetic mean of 140C for three different temperature integrating systeme. 87 Table 1. Interrelation of time and temperature effects on the hydration of a specific obsidian frcm American Falls, Idaho. Effect of hydration temperature on age conversion for 4.8 (u m)2 of hydration rind on a specific obsidian from American Falls, Idaho. (Adapted from data in Friedman and Long, 1976) Effective Hydration Temperature 0C Hydr. Rate, (u m)2/1000 yrs. 9.0 9.2 9.4 9.6 9.8 10.0 10.2 10.4 10.6 10.8 11.0 2.1 2.2 2.2 2.3 2.3 2.4 2.4 2.5 2.6 2.6 2.7 Age conversion for 4.8(u m)2 of hydration Yrs. B.P.* 2225 2175 2125 2100 2050 2000 1950 1900 1850 1800 1725 * Calculated age values have been rounded to the nearest 25 yr. multiple. Table 2. Selected source-specific hydration rates for obsidian sources that may have been used at some Great Basin archaeological sites. Obsidian Source Hydration Rate ( mm)2/103 yrs. EHr* oC Reference Coso Hot Springs, CA Clear Lake, CA Medicine Lake, CA Panum Dome, CA Obsidian Cliff, WY Teton Pass, WY Timber-Squaw Butte, ID Hawkins-Malad, ID Big Southern Butte, ID OwyheeBrown' s Castle, ID Annadel Farms, CA Bodie Hills, CA Napa Glass Alountain, CA Casa Diablo, CA Mount Hicks, NV Pine Grove Hills, NV 4.2 0.65 0.75 2.25 5.1 0.82 2.40 2.07 1.10 2.62 1.63 3.25 4.16 3.51 0.84 0.59 10.0 10.0 10.0 10.0 10.0 8.2 14.8 12.8 12.1 15.7 15.6 15.6 15.6 11.8 12.0 12.0 Friedman and Long Friedman and Long Friedman and Long Friedman and Long Friedman and Long Michels 1981a Michels 1981b Michels 1981c Michels 1982b Michels 1982c Michels 1982d Michels 1982e Michels 1982f Michels 1982a Michels 1983a Michels 1983b * EHr = Effective Hydration Temperature 1976 1976 1976 1976 1976 88 References Ambrose, W. 1976 Clark, D. 1961 Evans, C. 1960 Intrinsic hydration rate dating of obsidian. In: Advances in Obsidian GZass Studies: Archaeological and GeochemicaZ Perspec- tives, edited by R.E. Taylor. Pp. 81-105. Noyes Press, Park Ridge, New Jersey. The application of the obsidian dating method to the archaeology of central California. Ph.D. dissertation, Department of Anthropology, Stanford University. and B.J. Meggers A new dating method using obsidian, part II. American Antiquity 25: 523-537. Friedman, I. and W. Long 1976 Hydration rate of obsidian. Science 191: 347-352. Friedman, 1960 I. and R.L. Snith A new dating method using obsidian, part I. American Antiquity 25: 476-493. Friedman, I. and F.W. Trembour 1978 Obsidian: the dating stone. American Scientist 66(1): 44-51. Friedman, I., K.L. Pierce, J.D. Obradovich and W.D. Long 1973 Obsidian hydration dates glacial loading? Science 180: 733-734. Hughes, R. E. 1983 Obsidian source use at Hidden Cave, Churchill County, Nevada. In: The Archaeology of Hidden Cave, Nevada, edited by D. H. Thamas. AnthropoZogicaZ Papers of the American Musewn of NaturaZ History. In press. Johnson, L.R., Jr. 1969 Obsidian hydration rate for the Klamath Basin of California and Oregon. Science 165: 1354-1356. Layton, T.N. 1972 A 12,000 year obsidian hydration record of occupation, abandonment, and lithic change from the northwestern Great Basin. Tebiwa 15(2): 22-28. Meighan, C.W., L.J. Foote and P.V. Aiello 1968 Obsidian dating in west Mexican archaeology. 1075. Science 160: 1069- Michels, J.W. 1967 Archaeology and dating by hydration of obsidian. Science 158: 211- 214. 89 Michels, J.W. (continued) 1981a The hydration rate for Teton Pass obsidian at archaeological sites in Grand Teton National Park, Wyoming. MOHLAB TechnicaZ Report No. 3. State College, Pennsylvania. 1981b The hydration rate for Timber-Squaw Butte obsidian at archaeological sites on the eastern margin of the Columbian Plateau. MOHLAB TechnicaZ Report No. 4. State College, Pennsylvania. 1981c The hydration rate for Hawkins-Malad obsidian at archaeological sites in the uplands of the northeastern Great Basin. MOHLAB TechnicaZ Report No. 5. State College, Pennsylvania. 1982a The hydration rate for Casa Diablo obsidian at archaeological sites in the Manmoth Junction area of Mono County, California. MOHLAB TechnicaZ Report No. 6. State College, Pennsylvania. 1982b The hydration rate for Big Southern Butte obsidian at archaeological sites in the uplands of the northeastern Great Basin. MOHLAB TechnicaZ Report No. 9. State College, Pennsylvania. 1982c The hydration rate for Owyhee-Brown's-Castle obsidian at archaeo- logical sites in the uplands of the northeastern Great Basin. MOHLAB TechnicaZ Report No. 10. State College, Pennsylvania. 1982d The hydration rate for Annadel Fanms obsidian at archaeological sites i the Oakland area of California. MOHLAB TechnicaZ Report No. 12. State College, Pennsylvania. 1982e The hydration rate for Bodie Hills obsidian at archaeological sites in the Oakland area of California. MOHLAB TechnicaZ Report No. 13. State College, Pennsylvania. 1982f The hydration rate for Napa Glass Mountain obsidian at archaeological sites in the Oakland area of California. MOHLAB Technical Report No. 14. State College, Pennsylvania. 1983a The hydration rate for Mt. Hicks obsidian at archaeological sites in the Pine Valley area of Nevada. MOHLAB TechnicaZ Report No. 21. State College, Pennsylvania. 1983b The hydration rate for Pine Grove Hills obsidian at archaeological sites in the Pine Valley area of Nevada. MOHLAB Technical Report No. 22. State College, Pennsylvania. Norton, D.R. and I. Friedman 1981 Ground temperature measurenents, part I. U.S. GeoZogicaZ Survey ProfessionaZ Paper 1203: 7-11. 90 O'Brien, P.J. 1971 Pallmann method for mass sampling of soil, water or air temperatures. Geological Society of America BulZetin 82: 2927. Pallmann, H. 1940 Eine neue Methode der Temperaturmessung bei oekologischen oder bodenkundlichen Untersuchungen. Berichte der schweizerischen Botanischen GeseZlschaft 50: 337-362.