ANALYSIS BY X-RAY FLUORESCENCE OF SOME AMERICAN OBSIDIANS J. R. Weaver and F. H. Stross1 Analysis of minor or trace components of minerals has shown that the relative concentrations of these components often are uniquely characteristic of the source of the mineral, while those of the major components, e.g. silica or alumina, are not. In the case that there exist relatively few sources of the mineral but relatively many sites on which artifacts made from this mineral are found, such analysis can provide significant information on the mode of distribution and related attributes of the object. Obsidian is an example of such a mineral; a recent analytical study of European, Asian, and African obsidians (Cann and Renfrew 1964) has evolved information of the type mentioned. Recently developed instrumental methods of analysis are highly sensitive and rapid, and can often be carried out without damage to the specimen. It is the object of this and the following paper to describe the appli- cation of a technique of this kind aimed at relating obsidian objects found in Middle America to putative sources of the mineral, and to distinguish the sources from each other. This, we hope, will provide insight into the mutual contact of the various cultures with each other in prehistoric times. Obsidian has also lent itself to other types of examination. A fresh surface of obsidian absorbs water from the ambient to form hydra- tion layers (Friedman and Smith 1963) that can be measured under the microscope. The hydration begins when the piece is chipped or flaked, and continues at a knowable rate. Measurement of the thickness of the hydration layer thus provides an estimate of the time elapsed since the surface was worked. The rate of hydration is affected by the average temperature of the site (corrections are easily applied), but the aver- age relative humidity seems to have no significant effect because there is always enough water in a natural ambient to saturate the outer surface with a molecular film of water. This technique, then, may be useful in direct dating of an artifact, but the analytical method to be described here is to be applied to correlations of a quite different kind. To attain our objective, we must find the distinguishing components that are similar in concentration within a source, but differ between 1 Shell Development Company, Emeryville, California. - 89 - sources. To ascertain the presence of such components a limited number of carefully chosen obsidian materials from the same, and from different, sources and a few artifacts found on different sites were subjected to analysis. In view of the promising and consistent results, a more detailed study has been planned. The successful correlation obtained by Cann and Renfrew (1964) was based on the analysis of obsidian by means of optical emission spectros- copy. The use of that method represented a real advance over conventional chemical analysis since it permits the detection and determination of about seventy elements in a mere fraction of the time required by the latter methods. This suggested to us that an additional saving in time might be effected by the use of the x-ray fluorescence technique. It has attained much favor in recent years because of the great rapidity with which many kinds of samples may be analyzed; all elements having atomic numbers greater than 11 (Na) can be detected with conventional apparatus. METHOD The x-ray fluorescent method is simple both in principle and practice. When atoms are irradiated with sufficiently energetic x-rays, each different kind of atom emits a characteristic spectrum of x-rays; these x-rays are sorted according to wavelength, by means of a crystal, analogous to the sorting of visible light by means of a diffraction grating, and they are detected by electronic devices. In practice, the sample, either as a solid or in solution, is placed in a cell and its spectrum recorded on a strip chart. This gives a sort of "finger print" analysis, and is the technique applied in this study. In the more elaborate quantitative mode of operation, careful intensity measurements are made at discrete wavelengths without recording the complete spectrum. The instrument used was a General Electric XRD-6 with a chromium- target tube operated at 50 kv and 60 ma. In order to record the entire region of interest, two scans were made on each sample. A lithium fluoride crystal (2d = 4.0267 A) and a dual flow proportional counter - scintilla- tion counter were used to cover the range 5? to 1450 (2 i); EDT (ethylene- diamine tartrate) crystal (2d = 8.8030) was used with the flow proportional counter and a helium path to record the range 600 to 1450 (2 i). Both scans were made at the rate of 20 (2 9) per minute. Fragments of the obsidian were ground to a fine powder in an alumina-lined vial. The vial, containing the chips and an alumina ball, was shaken by means of a Spex Industries Mixer/Mill (Model 8000) for ten 91 minutes. A shallow container was filled level full with powdered sample (about 2.5 g), inserted into the instrument, and the fluorescent peaks were recorded on the strip chart. Every recorded peak was identified and its approximate intensity in terms of counts per seconds was read from the chart. Results Nine specimens of obsidian were analyzed; the results are shown in Table 2. It is most important to understand that the units are observed counts per second for specific lines (KS in every case except Ba, where lal was used), and that a given concentration of one element will not neces- sarily produce the same count rate as the same concentration of a different element. In other words, it is possible, from these data, to deduce differences between samples, but not differences between elements. DISCUSSION Study of Table 2 reveals some interesting differences in composition of obsidians from different sources, and some rather striking similarities between the artifacts believed to be from the Pachuca source (Numbers 6 and 7) and the Pachuca mineral specimen (Number 3). It is also encouraging to note the chemical similarities between the Papalhuapa specimens (Numbers 4 and 8) despite the gross difference in physical appearance (black and red). The data are suggestive but by no means conclusive. The usefulness of the technique could only be demonstrated by analyzing numerous samples from each of the various sources of obsidian. One hopes that a distinctive and consistent composition will be found within each source, but even if this is not so (and it is unlikely), a single element may serve as the distinguishing characteristic. Since the x-ray fluorescent procedure is so rapid and is capable of detecting many elements even at low concentrations, it seems an ideal method for examining many samples in the essential pro- cess of finding which elements provide valid correlations. Although the instrument time was approximately two hours per sample, the time required of the operator during the recording process is trivial; with a little practice the analyst can probably identify and measure the significant peaks in about fifteen minutes per sample. 92 Table 2. X-RAY FLUORESCENCE ANALYSES OF OBSIDIAN Units: Counts/second above background 1 2 5 4 5 6 3 = 7 8 9 "Glass Mt., Napa Co., Calif." (Mineral) "Site Sol-2, Solano Co., Calif." (Mineral) "Pachuca, Hidalgo, Mexico' (Mineral) "Papalhuopa, Jutiapa, Guatemala" (Mineral) "Copan" (Artifact) "Teotihuacan, green obsidian presumed to be from Pachuca source" (Artifact) "La Venta green obsidian (from Pachuca source?) (Artifact) "Red obsidian frcm Papalhuapa" (Mineral) "El Chayal obsidian - Guatemalan (second Guatemalan source)" (Mineral) MOST.,) DUM prODaDly not aLJ the Cr target x-ray tube. may still be significant. or tne L.T intensity is irom The difference between samples 1 2 3 4 5 6 7 8 9 Zr 210 20 720 t f w 120 7 Io50 Nb 30 85 60 70 Rb 140 165 140 80 65 . 160 150 80 120 Sr 110 90 20 140 110 Cu 135 145 135 150 135 135 160 135 145 Zn 25 32 78 15 85 75 20 25 Ni 18 23 25 15 25 25 20 25 Fe 1130 1140 1880 1080 1060 1900 1940 1180 780 Co Trace Trace Trace 15 15 Ca) 840 820 820 750 760 810 840 740 740 Mn 9o 80 180 .170 110 180 200 120 35 Ba 32 25 Trace 65 60 65 50 Ti 75 60 140 170 150 145 150 160 120 Ca 85 70 30 230 208 35 35 245 130 K 34o 370 305 345 325 360 34 5 370 330 Cl 100 150 200 90 go 180 150 40 60 Si 1480 1410 1360 144o 1260 1380 1290 1460 1520 A1 510 420 .210 510 500 480 4 430 400, ,% I I r + N 4. %-ft ,-1 - I le , - _4 -' 'I A% _U- __== _ -.L _- a) 93 REFERENCES CITED Cann, J. R. and Colin Renfrew 1964 The Characterization of Obsidian and Its Application to the Mediterranean Region. Proceedings of the Prehistoric Society, Vol. 30:111. Cambridge, England. Friedman, Irving and R. L. Smith 1963 Obsidian Dating. In Science and Archaeology, Don Brotherwell and Eric Higgs, eds. Thames and Hudson, London.