Marc L. Banks
Department of Anthropology
University of Connecticut

Introduction
For over a century, the occurrence of ground-slate semilunar knives at prehistoric sites has drawn the attention of Northeastern archaeologists. Despite this fascination, the origin these knives and the relationships between the prehistoric groups in New York and southern New England who used them are incompletely understood. The disappearance of ground-slate knives from the archaeological record in these areas after a relatively short span of time also remains a mystery. The sites that may hold the answers to these questions have yet to be found. In the meantime, more fundamental information related to the manufacture, function, and distribution of ground-slate knives may provide new perspectives for addressing the larger cultural issues. With this in mind, ethnographic data is here combined with the analysis of ground-slate knives from Connecticut. The knives examined include those recovered by the author at the Indian Hill Site in north central Connecticut and a large sample from the Norris Bull Collection.

Origins
The remarkable similarity of prehistoric ground-slate semilunar knives to the modern Eskimo ulu (ulo) or woman's knife was the basis for early speculation about the origin of these blades and possible relationships between northern cultures and prehistoric populations in New England (Abbott 1881; Beauchamp 1897; and Willoughby 1973[1935]). With the advent of radiocarbon dating, it became clear that New England and New York specimens actually predated those from more northern sites; earlier explanations based on either migrations or diffusion from the north were abandoned (Ritchie 1969). Discoveries in the Lake Champlain region and the St. Lawrence drainage resulted in Ritchie's proposing the Laurentian Tradition, which he envisioned as an adaptation to the resources of the conifer-hardwood forest that existed in portions of the Northeast during the Late Archaic Period (Ritchie 1969). The semilunar knife was identified as one of a number of diagnostic traits of the Laurentian. Because of this forest association, the Lake Forest Archaic has been suggested to account for the many local permutations of this adaptation, which existed throughout the Northeast (Tuck 1978; Snow 1980; McBride 1984). However, the distribution of semilunar knives clearly exceeds the boundaries of either the Laurentian Traditon or the Lake Forest Archaic. In Connecticut, where a mast forest was established by 6000 BP (Jacobson, Webb, Grimm 1987), semilunar knives excavated at the Indian Hill Site (11-2; 6HT121) and others from surface collections indicate that the distribution and use of this tool form were not limited to a particular environmental zone. The appropriateness of the term "Laurentian" for southern New England has been questioned because the projectile points upon which this tradition is based have not been associated with other core traits identified by Ritchie (Dincauze 1975). Sites in the lower Connecticut River Valley have been assigned to the Golet Phase based solely on the occurrence of Otter Creek, Vosburg, and Brewerton projectile points (McBride 1984). However, at the Indian Hill Site an additional Laurentian trait was also present, the ground slate knife.

Description Of Semilunar Knives
Prehistoric specimens from the Northeast are commonly referred to by the Eskimo name ulu. These implements are semicircular in shape and diminish in thickness approaching the blade edge. Most examples are made from slate, although some fine-grained sandstone knives have been found (Beauchamp 1897; Ritchie 1969). Both lithic materials lend themselves well to grinding and polishing. Other materials such as quartzite and chert were also used to make semilunar knives, but these knives were usually chipped rather than ground. The initial stages of manufacture of some ground specimens involved flaking as a means of thinning the blade prior to grinding (Willoughby 1973[1935]; Turnbaugh 1977). Willoughby described two basic forms of semilunar knives: 1)those with a raised ridge forming a handle, which he referred to as monolithic (Figure 1); and 2)those with a drilled (sometimes scribed) perforation(s) used in attaching a wooden handle (Willoughby 1973[1935])(Figure 2). To these can be added a plain blade form lacking both a raised ridge and hafting perforations (Turnbaugh 1977)(Figure 3). In some instances where a raised ridge was absent, an incised line separated the blade from the handle (Turnbaugh 1977)(Figure 4).

The ground-slate knife offered a number of advantages over chipped knife forms. Foremost, slates possess what is referred to as a "slaty cleavage" which is a tendency to split in thin parallel sheets (Gillen 1982). These cleavages penetrate the entire rock, making it possible to split the rock to desired thicknesses (Philpotts, personal communication). Slates also contain a number of silicate minerals (Gillen 1982) that would enhance their cutting ability. The smooth, uniform blade would have been equally suited for slitting fish (Stewart 1977) or dressing animal hides. The ground-slate knife could be resharpened many times, thus greatly extending its use-life (Hayden 1977; 1981). Osgood (1940) observed that the average life of ulus used by Ingalik women of Alaska was about three years. Grinding would result in a much slower reduction in the size of the knife than would flaking techniques, thus providing "more cutting edge per pound (Hayden 1977:81)." Hayden has hypothesized that this tool form was developed in response to changes in subsistence and the use of mass-capture technologies for resource such as fish following the Pleistocene (Hayden 1981). These new technologies (e.g., weirs) necessitated development of specialized tools to process these resources. The ground-slate knife was an economical solution to this need.

Historic Uses Of The Ulu
Ulus have been used by historic Inuit and Indian groups across North America. Interpretations of similar prehistoric knives are, to a large degree derived by analogy from the way in which ulus have been used by these groups. Several ethnographic examples illustrate how for many Native American groups, the ulu's basic functions was that of a fish knife; this was especially true in regions where people relied heavily on seasonal fish runs. Swan (1870) described the use of a Halibut chopper by the Makah of the Northwest Coast. This ulu-like knife was used to slice fish in preparation for drying. The Ingalik of interior Alaska also used the ulu for this purpose (Osgood 1940). Tlingit women used a similar crescent knife to slit and clean salmon (Krause 1956).

The ulus were not reserved exclusively for cleaning and preparing fish. Birket-Smith cited the skill with which Greenland Eskimos and Aleuts butchered seals using ulos (Birket-Smith 1959). More recent, Oakes has described the present day use of moon- and triangular-shaped ulus by Inuits in the Artic Bay region to prepare seal skins to make Kamiks (boots) and other items of clothing (Oakes 1987; 1991). These ulus are used for both cutting and scraping operations. The Ingalik used small ulus to perform fine skin work (Osgood 1940). Athabascan Indians in Northern Alaska even used the woman's knife to prepare fish skins to make mukluks (Rooth 1971).

The ulus or woman's knives of the Ingalik consisted of stone blades ground to the desired size and hafted to a wooden handle, the most common form being a semilunar blade glued into a slot made in a wooden handle (Osgood 1940). Osgood noted that a rocking motion was employed with the heel of the blade doing the actual cutting. Although men generally made these knives, they were used almost exclusively by women.

Materials other than slate have also been used to manufacture women's knives. Along the Northwest Coast, large mussel shells were ground to form such knives (Drucker 1955). In fact, Drucker noted that slate was used where it was available, but that by historic times, the shell forms were much more common than those of slate (Drucker 1955; 1965). Among the Cooper Inuit, ulu blades were among tools hammered from native copper (Birket-Smith 1959; Morrison 1992). As new metals became available, some Nootka groups living on the Northwest Coast eventually made iron and steel copies of earlier ground-slate knives (Stewart 1977). Likewise, some Eskimo groups frequently used saw steel to make ulus (Anderson and Eells 1935). Some of these were hafted to ivory or wooden handles.

The Indian Hill Site
The Indian Hill Site (11-2) in Bloomfield, Connecticut is located on a relatively flat terrace several hundred meters from the Farmington River, a major tributary of the Connecticut River. This location is particularly significant because of its proximity to a constriction in the Farmington River (Figure 5) that would have provided easy access to anadromous fish during spring runs. During excavations which the author began in 1989, over 80 sq. meters were excavated. Within the subsoil were artifacts generally associated with the Middle and Late Archaic Periods. These artifacts included Otter Creek, Brewerton, and Vosburg projectile points- all diagnostic of Ritchie's Laurentian Tradition as well as the basis of McBride's Golet Phase in the lower Connecticut River Valley (Figure 6). In addition to these points, fragments of five ground-slate semilunar knives were recovered (Figure 7). This element of the Laurentian was missing from Golet sites reported by McBride (McBride 1984). Charcoal from a hearth was radiocarbon dated at 4870+80 years BP, which falls well within the temporal limits for the Laurentian as defined by Ritchie (1969) and Funk (1976) and Golet sites in Connecticut (McBride 1984)(Figure 8). The archaeological evidence suggests that Indian Hill functioned as a base camp during the spring and summer. The frequency of artifacts and features provide evidence for a number of different activities across the site. Hearths at various depths and, in one instance, superimposed upon one another convey the idea that aboriginal groups returned to this site seasonally. The faunal assemblage included a large number of snake and turtle elements and four calcined sturgeon- plate fragments, further supporting the notion of spring-summer occupations. If indeed it was the fishing opportunity that drew human groups to this location, it is reasonable to suggest that the semilunar knife may have been used to process fish.

The Norris Bull Collection
During the 1930's and 1940's, Norris Bull assembled Connecticut's largest collection of Indian artifacts. The many thousands of artifacts included items which were surface-collected, dug from shellheaps, and purchased from other collectors. The collection, which contains virtually every type of artifact known from southern New England, was donated to The University of Connecticut after Bull's death. The collection includes fifty-seven ground-slate semilunar knives, a relatively large number considering the rarity of such finds (see Turnbaugh 1977). Although most of these artifacts are identified only by the particular town or county from which they were collected, they provide a great deal of information as to the manufacturing techniques, the size and forms that were made, the lithic material used, and the wear that resulted from resharpening and use.

Distribution of Ground-Slate Knives in Connecticut
The distribution of ground-slate knives examined in this study is shown on a map of Connecticut containing the major river drainages within the state (Figure 9). Interpretations of this map are limited by the vagueness of some of the proveniences and by the unsystematic manner in which many of the knives were collected. Some parts of the state are certainly over-represented. Yet the map does clearly indicate that semilunar knives were dispersed widely across the state. This is significant since the slates used to make these knives had to be obtained outside Connecticut. The closest sources were within the Hudson Valley, just west of the Connecticut border (Calogero; Philpotts, personal communications). These formations are among slate belts running from the Appalachians to Canada.

Although weathering and the grinding and polishing involved in preparing the knives have obscured rock characteristics needed to make more positive identifications, the majority of the knives were macroscopically identified as slate. One of the Indian Hill specimens appears to be hornfels, a local material. A few other exceptions will be noted later. Microscopic identifications were not employed since they require procedures such as thin-sectioning, which are destructive to the artifacts. Incongruities in X-ray fluorescence readings and the absence of information on known slate sources eliminated trace elements as a useful avenue of study (Calogero 1991).

Semilunar Knife Forms
The semilunar knives were placed into four general categories. These included: plain; perforated; comb back; and chipped. The plain variety included those knives which exhibited the semilunar shape and a relatively uniform thickness. The perforated variety had one or more holes through the upper portion of the knife, evidently used for attaching a handle to the blade. Beauchamp reported almost identical forms being used by Eskimo groups (Figure 10). Those placed in the comb back category feature a raised ridge along the knife's back. In some cases, the ridge is further emphasized by scribed lines between the ridge and the blade. Willoughby (1973[1935]) had noted that these were the most common variety of semilunar knives found in southern New England. While chipped semilunar knives are not the topic of this paper the Bull Collection contained five such knives. All of these were made of materials other than slate (quartzite, chert, basalt). Some elements of different forms were combined on individual knives. For example, drilled holes were also present on some of the comb back knives. Table 2 shows the number of each type identified in the Bull Collection. Four of the five knives from Indian Hill were of the comb back variety.

Thirty-eight complete semilunar knives from the Bull Collection were selected for metric analysis. Three basic dimensions were taken: overall width; height; and maximum thichness (Figure 11). Table 3 shows the statistical data from these measurements. No significant correlation was found between these variables. While width and height dimensions were in part a reflection of knife function and the degree of sharpening, the maximum thickness of the knives seems to have been little affected by either. The thickness of the knives remained quite uniform regardless of the width or height dimensions (Figure 12). Since slate, as discussed above, can be split to a desired thickness, it may be that the mean thickness of 10mm for the Bull specimens resulted as a compromise between strength and the thinness necessary to create a usable blade (Philpotts, personal communication). The mean thickness for four of the Indian Hill fragments was 7.25mm.

Criteria for assessing blade resharpening
A ratio of the height to width was calculated to determine the degree to which the semilunar knives had been resharpened. The validity of this ratio is based on the assumption that knives were semicircular in shape initially (Figure 13). Subsequent resharpenings would have resulted in a increasingly diminished height without substantial reduction in width. The bottom row of Table 2 lists the statistics for the height/width ratios. Those knives with higher ratio values would be the ones which exhibited the fewest signs of resharpening. Thus, a ratio of of .24 would indicate a greater number of resharpenings than would a ratio of .63. Figure 14 graphically displays the the ratios calculated for these knives.

Other Blade Modifications
A few of the ground-slate knives studied exhibited pronounced beveling along the blade edge. This degree of beveling was absent on the majority of the knives; it is unclear whether this beveling is related to the intended function- possibly cutting rather than scraping- or represent a freshly sharpened knife. Much more common than beveling was the presence of edge damage. This usewear consisted of flake scars of varying size, suggesting that the blade had made contact with hard material, possibly bone. This type of wear would not be expected on knives used as scrapers.

Other modifications suggest these knives were often curated. Besides frequent resharpening, many knives were reworked following breakage. This modification could be seen in changes in knife morphology, ground edges, and differences in the heights of the combs from one side of the knives to the other. One of the ground-slate knives from Indian Hill displays both a high frequency of resharpening and modification (Figure 15). One side of this blade had become quite angular (45 degrees) and the length of the blade below the comb had been reduced to 18 mm at its maximum point. This knife also shows evidence of having been broken and then reground on one side. More extreme modification can be seen in an example from the Bull Collection where a small knife fragment was reworked into a scraper (Figure 16).

Conclusions
Ground-slate knives represent an additional element of the Laurentian Tradition that is present in Connecticut. While this tool form does serve as a useful chronological marker, it is still uncertain whether this and other Laurentian traits can be attributed to a particular group of people. The cultural significance of the the different semilunar knife forms is equally uncertain. We clearly need more archaeological data before we can evaluate the validity of constructs such as the Laurentian and understand the cultural links that existed in the Northeast.

Contact between prehistoric groups living in the Northeast is indicated not only by the appearance of these knives and particular projectile point types but also by the movement of slate to areas substantial distances from sources. These encounters undoubtedly explain the spread of tool forms such as the ground-slate knife among peoples sharing similar adaptations. The ethnographic data and the evidence from the Indian Hill Site support the notion that these adaptations involved aquatic resources such as anadromous fish. Assuming that the blades of prehistoric semilunar knives lasted as long as the ulus described by Osgood, the number of knives recovered at Indian Hill seems particularly significant. The degree of resharpening and modification these knives further attest to the heavy use these knives saw. The disappearance of this knife form during the Late Archaic Period is more likely a result of alternative materials, perhaps shell and bone, being substituted for slate than a consequence of adaptive changes. Hopefully, future excavations will provide the evidence to test the inferences which have been made and we can begin to fill in the sketchy picture our present data provides.