Traditional methods for sourcing archaeological materials, such as X-ray Diffraction and X-ray Fluorescence, Particle-Induced X-ray Emission, and Neutron Activation Analysis all have some technical limitations or complicating factors such as accounting for the diluting effects of temper (Neff 1995; Larson et al. 2005:100). Where ceramics are concerned, another problem is that clay sources may be homogenized or thoroughly mixed over large areas, making precise sourcing difficult (e.g., Neff 2008; Steponaitis et al. 1996). Although these limitations may be addressed by focusing on surface slips or pigments rather than clay matrices (e.g., Neff 2003; Sall et al. 2005; Speakman 2005; Vaughn et al. 2005), the development of new methods and techniques for sourcing artifacts is desirable.
Aquatic mollusks intake chemicals in general proportion to what is in the environment (e.g., Lee and Wilson 1969), providing a theoretical base for employing shell in sourcing studies. This has been done for salt-water species (Claassen 1998:212-218), but prior to this NCPTT-funded study it had not been done with freshwater mussel shell. Because of modern pollutants and other changes in aquatic ecosystems, modern shells serve poorly in establishing comparative baseline data (Claassen and Sigmann 1993; Clayton et al. 2001; Das and Jana 2003; Jacomini et al. 2003; Markich et al. 2002; Miller 1980). However, enormous amounts of freshwater shell have been excavated from archaeological sites, and although shell artifacts and shell-tempered ceramics could easily have been imported, unmodified valves incorporated into site matrices usually represent the remains of food obtained from waterways near the sites of deposition (Peacock 2000, 2002). Thus, the material necessary for establishing baseline elemental data already exists in archaeological collection, data that would be especially pertinent to sourcing shell-tempered ceramics.
Methods and Materials
The method employed in this study was Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry, a method increasingly being employed in archaeometric work because of its high precision (ppm) and the fact that it is essentially non-destructive (Resano et al. 2009; Speakman and Neff 2005). Data were obtained from over a hundred specimens from six sites in three major drainages. These include site 15CL58, a Middle Archaic shell midden on the Ohio River in Kentucky, site 22SU531, a Mississippian site on the Sunflower River in Sunflower County, Mississippi, and four sites (22LO527, 22LO530, 22OK520, and 22OK578) in the Tombigbee River drainage of east Mississippi. The Tombigbee sites are particularly interesting in that they allow a detailed look into the precision of the LA-ICP-MS shell sourcing method. Two of the sites, 22LO527 and 22LO530, are very near to one another on the main stem of the central Tombigbee River and thus should be chemically very similar. The other two sites are on lower-order tributary streams. Site 22OK520 is located on Line Creek, a tributary of Tibbee Creek, itself a tributary of the Tombigbee River. Site 22OK578 is located near Hollis Creek, a tributary of the Noxubee River, a major Tombigbee tributary. If analysis shows these individual feeder streams to be chemically distinguishable, then it would suggest that the method is potentially even more precise than clay sourcing.
Shells of various species were obtained from each site. Samples were cleaned with deionized, demineralized water. Specimens were then cut with a diamond-bladed band saw. About two- thirds of the posterior portion of the shell was removed; this is cut in half lengthwise, with the bottom half being retained as a chemical voucher specimen. A 1 cm-wide slice was then removed from the top half for chemical analysis (see figures in interim report; also on CDs sent under separate cover). If samples broke or if there was some problem with the test, additional slices were removed as necessary. The saw was cleaned with deionized, demineralized water between samples.
Individual slices of shell were mounted in clay with a cut side facing up. The laser track was set in a raster pattern over this surface. A pre-ablation pass was made to remove any adhering surface materials. An ideal sample consisted of 20 valves of at least 10 different species, to account for inter-species variation in chemical uptake (see data tables on CDs sent under separate cover). Shells of different sizes were used to average out differences in chemical uptake related to faster shell growth in younger individuals. The laser was set to run over seasonal growth rings, as there can be seasonal variation in chemical uptake by mollusks. A total of 46 elements was scanned for, including Li 7, Na 23, Mg 24, Al 27, K 39, Ca 44, Sc 45, Ti 47, V 51, Mn 55, Fe 57, Co 59, Ni 60, Cu 65, Zn 66, As 75, Rb 85, Sr 88, Y, Zr 90, Nb 93, Sn 120, Sb 121, Cs 133, Ba 138, La 139, Ce 140, Pr 141, Nd 142, Sm 152, Eu 153, Gd 158, Tb 159, Dy 164, Ho 165, Er 166, Tm 169, Yb 174, Lu 175, Hf 180, Ta, 181, Pb 208, Th 232, U 238, and Si.