By Michael J. O’Brien, Ph.D. and R. Lee Lyman, Ph.D.
In Dr. Leslie Pfeiffer’s excellent article on the East Wenatchee Clovis site (also known as Richie–Roberts) in central Washington, he states that among the most important finds at the site were 13 beveled bone rods made from mammoth or mastodon limb bones.
(Pfeiffer 2008). (There actually were 14 rods, one of which was extremely fragmented.)
He points out that “While there is no doubt that the Clovis people used mammoth bone in many ways, including bone points, the best hypothesis is that these bone rods were lidely foreshafts to mount stone points to wood shafts” (Pfeiffer 2008:76).
This is a widely held proposition in American archaeology, going back to at least the work of Luther Cressman (1942) in the Great Basin and that of John Cotter (1954), one of the original excavators of Blackwater Draw, the Clovis type site. The proposition later received formal expression from Larry Lahren and Rob Bonnichsen (1974) based on their analysis of the Anzick Clovis “burial-cache” materials from Montana (Fig. 1) and even later from Dennis Stanford (1996), who proposed a compound arrangement of beveled rods to create a hunting shaft. After considerable analysis of bone and ivory rods from across the western United States, coupled with engineering-design analysis of the rods from East Wenatchee (Lyman and O’Brien 2000; Lyman et al. 1998), we offer a different proposition—that many of the rods served as levered hafting wedges.
Most of the functions posited for bone and ivory rods—foreshafts, points, sled shoes, and so forth—have been based on suspected analogous specimens, especially those from the ethnographic and late prehistoric records. The problem with such an approach to explaining the archaeological record lies in the requisite assumption that the past is no different from the present—that the uniquely historical development of technology is denied as we force our archaeological observations into some ethnographically documented category ofphenomena. Usually what happens is, the more we start investigating details of construction, the more the arguments start to fall apart. Thus, we began our research without reference to ethnographically documented uses of rods and merely attempted to build a tool that functioned efficiently and simultaneously accounted for numerous features evident in the archaeological record. In short, we employed mechanical inference based on experimental evidence.
Figure 1. Lahren and
models of beveled bone
rods functioning as
foreshafts: (left) single-
shaft, and projectile
point; (right) bi-beveled
specimen, shaft, and
One thing that struck us as we compiled data on Clovis points from well-known caches in the western United States was their large size compared to that of points found in association with mammoths or bison at sites such as Black Water Draw Locality No. 1 (New Mexico), Lehner and Naco (Arizona), Lubbock Lake (Texas), and Sheaman (Wyoming) (Fig. 2). In fact, six of the 14 Clovis points from East Wenatchee originally reported by Michael Gramly (1991, 1993, 1996) were larger than any of the other 64 cache or non-cache points in our survey (Lyman et al. 1998). The exceedingly large size of the Clovis points from East Wenatchee suggests they were intended to be used as butchering tools—specifically, saws—rather than as weapons designed to pierce hide. Gramly (1991, 1993), in fact, refers to them as “knives.”
Other attributes of these particular specimens appear to validate such a categorization. For example, they all have either parallel or slightly convex blade edges, rather sinuous edges, and nearly straight to noticeably concave bases, the latter forming either a shallow V or a (sometimes deep) U. We suspect all these features are functionally and mechanically related. The specimens with convex edges could have been resharpened more times than a parallel-sided or distally converging-sided specimen. Further, specimens with convex edges, if used as weapons, would have cut holes through hide that would have been larger than holes made by points with parallel or distally converging blade edges, the former allowing penetration of a foreshaft or shaft.
We believe the association of rods with Clovis points in settings where animals were butchered—such as at Blackwater Draw—and in caches—such as at East Wenatchee—is not simply fortuitous. The fact that 14 fluted projectile points and evidence of 14 bone rods were found in the East Wenatchee cache suggests that a ratio of 1:1 of these two kinds of implements might be significant. Given these considerations, our experimental goal was to build a butchering tool—not a projectile—that employed both the large stone points and the rods. Using the specimens from East Wenatchee ,as a model, we focused on two things: building an efficient butchering tool—one that required minimal maintenance effort during use—and determining the function of the rods.
One critical issue in manufacturing an efficient butchering tool is how the blade is hafted to the handle. Our experimental work indicates that the amount of sinew required to haft a biface depends on both the size of the shaft to which the biface is hafted and the size of the biface. A 6- to 8-cm-long Clovis point that is to be used as a projectile point can be hafted with a single strip of sinew 0.5 cm wide and 30-40 cm long. Seating the point, wrapping the sinew, and allowing the sinew to dry sufficiently to tighten takes about 30 minutes if the air is dry and warm, plus additional time to apply mastic. Hafting larger bifaces—such as those from East Wenatchee—that are to be used as saws requires as many as 20 strips of sinew 0.5 cm wide and 40 cm long. The greater amount of binding is required because the handle has a larger diameter than a dart shaft, and the force applied to the biface during use as a butchering saw is different than that applied to one used as a projectile point. Because more sinew is required, and the haft comprises multiple layers of sinew, the entire process of seating the biface, wrapping, drying, and applying mastic to the haft has, in our experiments, taken almost two hours.
If the sinew binding absorbs moisture (which it does quickly if it is completely dry prior to use of the tool), such as from body fluids of an animal being butchered, the binding expands and becomes loose. Coating the sinew with mastic (e.g., tree resin) tends to waterproof it and extend the use-life of the haft, but a haft will nonetheless loosen through use and moisture absorption, even if coated with resin, which will wear off or flake off if it becomes hard and dry. Because it takes a rather long time to rehaft (particularly) large bifaces—the resin must be removed, the sinew unwound, the point reseated, the sinew rewrapped and allowed to dry, resin applied—the mechanical problem is to prolong the use life of the initial hafting. This can be accomplished by tightening the sinew via a wedge inserted between the sinew and the shaft to which the biface is hafted. Engineering evidence indicates that the beveled rods from East Wenatchee served this binding-tightening wedge function.
In the barest of terms (see Lyman et al.  for details), here is how the wedge works vis-à-vis the complete composite tool. During manufacture of the handle that eventually will hold the stone tool, a groove is cut in the wooden shaft or handle and extended onto the (tapered) nock tang (Fig. 3). This groove, used for seating the rod, should be at least as long and wide as the rod and about half to two thirds as deep as the rod is thick. Sinew is bound onto the shaft or handle as tightly as possible prior to inserting the rod. Before the sinew completely dries, the rod is set in the groove cut for it and is shoved up under the binding (Fig. 4). After the sinew has dried, the rod is levered down into the groove and the proximal end held in place with leather or sinew binding. The fulcrum should be distal to the center point of the rod length. The levering down serves as the final tightening of the sinew. It is this final levering down into the groove that mechanically explains why the face opposite the bevel must be convex rather than straight; were it not, the rod would lie flat in the groove and no lever-enhanced tightening of the sinew would be possible.
Grooves, usually in the form of crosshatching, cut into the bevels of the rod help keep the sinew and rod from slipping during use. As the sinew binding dries, it shrinks and sets down into the grooves cut in the bevel of the rod, making a mechanically sound haft (Fig. 2). (The photograph made by Pete Bostrum of the longest bone rod from East Wenatchee [Fig. 5], is an excellent close-up of the crosshatching found on the beveled portions of many Clovis rods.) Without crosshatching on the bevel, the rod has a tendency to slip out from under the sinew binding when it is levered down. The fewer, generally less deep grooves sometimes evident on the convex face opposite the bevel help secure the rod to the fulcrum area (Fig. 4) of the groove in the handle when a bit of hide is used to raise the fulcrum.
The hafting-wedge function of the rods readily accounts for why they were beveled on both ends. Should the beveled end being used as a binding wedge fracture, one has but to merely turn the rod around 180 degrees, insert the intact edge under the haft binding, and lever the rod down to maintain a tight binding. Beveling of the proximal end-toward the handle-also allows the binding holding it down to be more easily slipped on and off the levered-down end of the rod. The basically cylindrical cross section of the rods can be accounted for by noting that we have found it to be relatively easy to carve a groove for the rod that is curved in cross section using a scraper with a convex bit. Finally, the thick cross section of the East Wenatchee rods would have made for a larger cross section under the bevel, where the most force was concentrated when the rod was being used to tighten the haft binding.
In summary, we believe we discovered a way to haft large fluted Clovis bifaces that produces a mechanically efficient tool, which we have used to butcher large game and cut frozen meat easily. It would be a very efficient tool to use if faced with a proboscidian carcass. Our engineering work takes into consideration a suite of attributes of both the fluted stone points-sinuous convex edges, flutes, large size, concave base-and the bone rods-bevels, scoring of bevels, convex face opposite the bevel-in the East Wenatchee cache. These sets of attributes are functionally and mechanically interrelated in the final tool, which we believe accounts for (1) the co-occurrence of the large points and the rods in caches and in butchering contexts, (2) the breakage and wear evident on some rods, and (3) the varied sizes of both points and rods.
As a final note, we point out that the idea for our work came from Virgil Hayes, who also carried out all the replicative work. He joined us as co-author on the original paper, which was published in Journal of Archaeological Science (Lyman et al. 1998). Hayes never had any formal training in archaeology, but he possesses keen insights into ancient technology and an innate ability to reverse engineer what he sees. His experimental work is yet another excellent example of the significant breakthroughs and discoveries routinely made by nonprofessional archaeologists.