A proliferation of smaller arrow points, especially within the last approximately 1,500–2,000 years, is commonly regarded as representing the introduction of bow-and-arrow technology to atlatl-and- dart-using peoples throughout most of North America (Bettinger Reference Bettinger and Beck1999; Blitz Reference Blitz1988; Goodale Reference Goodale and Prentiss2019; Hughes Reference Hughes1998). Arrow and dart points are commonly distinguished by metric attributes (Hildebrandt and King Reference Hildebrandt and King2012; Shott Reference Shott1997; Thomas Reference Thomas1978), and although such attributions may have substantial validity, this binary classification requires a level of assumption that oversimplifies important variation. This is not a novel observation; more than 35 years ago, Blitz (Reference Blitz1988:125) asked, “How valid is the critical assumption that large, heavy projectile points were used to tip atlatl darts, while small, light points represent arrowheads?” Variation in point size and form is linked to variation in application (à la the “form follows function” maxim of architect Louis Sullivan [Reference Sullivan1896]), but relating these variables to a particular delivery system is not absolute. Hughes (Reference Hughes1998:346) notes that many researchers have demonstrated metric and morphological change through time, “but the meaning of these changes is uncertain.” Erlandson et alia (Reference Erlandson, Watts and Jew2014:162–163) pointedly observe that “archaeological inferences about how a particular point was used appear to be based on intellectual tradition.” In a discussion of arrow versus dart points, Thomas (Reference Thomas1978:461) acknowledges that archaeologists employ “a staggering array of suppositions.”
Atlatls, which presumably propelled darts with relatively larger projectile tips (with respect to arrows), are weapons with ancient roots that may have been used by the earliest people in the Americas, and they were used along the Oregon coast and lower Columbia River for sea-mammal hunting at least into the most recent millennium (Losey and Hull Reference Losey and Hull2019; Spaulding Reference Spaulding1953). However, there is currently no direct evidence or historic reference to their use in the nearby Willamette or other proximate interior valleys. Discussions regarding the timing of the first presence of bow-and-arrow technology and the persistence of dart-and-atlatl weaponry have been extensive, and the timing of the use of these weapon systems remains controversial. At the early end of the time spectrum, Maschner and Mason (Reference Maschner and Mason2013:133) make a case for the bow being present in coastal Alaska during the early Holocene. Erlandson et alia (Reference Erlandson, Watts and Jew2014) report traditionally “arrow-sized” projectile points on the California Channel Islands as early as 12,000 BP, speculating that they could have tipped harpoons. Ames et alia (Reference Ames, Fuld and Davis2010) have argued that the bow and arrow may have been present on the Columbia Plateau of western North America as early as 8,500 years ago and that they were “ubiquitous” by 4400 BP; they became predominant after approximately 1500 BP, and atlatls disappeared by 1000 BP. Chatters (Reference Chatters, William and Kuijt2004) argues that bows and arrows were present on the Columbia Plateau “earlier . . . than anywhere else in temperate North America,” but puts their entry just prior to 2400 cal BP. Castro et alia (Reference Castro, Yebra, Cortegoso, Marsh, Castillo, Rughini, Victoria Fernández, Garvey, Belardi, Bozzuto, Fernández, Moreno and Neme2021) suggest that the bow and arrow first appeared in the central Andes between 3500 and 3000 cal BP.
Arguments for the early presence of bows in the Americas are based largely on the presence of smaller projectile tips. And although Erlandson et alia (Reference Erlandson, Watts and Jew2014:168) acknowledge that they cannot determine with what weapon system small Channel Island points may have been used, they reasonably point out, “If the bow and arrow was present in northeast Asia near the end of the Pleistocene . . . archaeologists might ask why such an advanced technology was not transmitted earlier to the Americas.” One possibility for limited use was suggested by Maschner and Mason (Reference Maschner and Mason2013): the simple self bow, which was present in the early Holocene, may have lacked the power to reliably bring down larger prey.
Focusing on the physics of flight and weapon engineering, Hughes (Reference Hughes1998:346) found that dart flight stability can be achieved either by increasing forward mass (i.e., a larger projectile tip) or by increasing rear drag (the addition of fletching). Fletching, which she posits appeared after around 7,500 years ago—based on her analysis of points from Mummy Cave, Wyoming—allows for smaller and lighter tips, which also increases effective flight distance and accuracy. According to Shott (Reference Shott1993:428–429), “Point sizes and forms are specific compromises between conflicting needs for velocity and accuracy, thrust or kinetic energy, and other performance requirements.” A smaller tip can improve projectile speed and distance, but a larger tip can hit its target with greater destructive force and penetrating power, causing more tissue damage for a quicker kill (Grossman Reference Grossman2021; Wells Reference Wells2024; Wilkins et al. Reference Wilkins, Schoville and Brown2014). An optimal projectile tip depends on context: the type of prey, hunting method, and environment. Furthermore, context suggests that different projectile tips might be used for hunting and for warfare (Loendorf et al. Reference Loendorf, Simon, Dybowski, Woodson, Plumlee, Tiedens and Withrow2015, Reference Loendorf, Plumlee and Tiedens2017; Stevens Reference Stevens1870:564).
We examine a large collection of projectile points (1,131 classifiable and 464 fragments) from a set of eight sites bordering Mill Creek in the mid–Willamette Valley of western Oregon. We recorded standard metrics, assuming that familiar patterns would follow, such as clear arrow/dart distinctions with temporally predictable occurrences, but our findings did not meet expectations. Perhaps the lack of temporal clarity in the projectile points should have been expected. Toepel (Reference Toepel1985:82–83) was the first to propose a projectile-point typology with explicitly chronological meaning in western Oregon’s Willamette Valley, based on the premise that projectile points “appear to decrease in size over time in the Willamette Valley as well as elsewhere.” She focused on size metrics, particularly neck width, base width, maximum width, and thickness. Ultimately, however, she conceded that “the temporal range of most of the types can be defined at this time in only very general terms” (Toepel Reference Toepel1985:113–114). This outcome was frustrating, in light of examples of well-developed temporal point typologies from adjacent areas of the Columbia Plateau and Great Basin (e.g., Leonhardy and Rice Reference Leonhardy and Rice1970; Lohse Reference Lohse and Campbell1985; Thomas Reference Thomas1981), and it has not significantly improved in subsequent decades. Roulette et alia (Reference Roulette, Wilson, Ellis and Chapman1996:12.20) similarly found that “the continued presence of dart points until the end of the Late Archaic indicates . . . that darts were not completely replaced by bows and arrows,” based on their work at multiple upper–Willamette Valley sites along a pipeline corridor. For western Oregon, one source of frustration was the technological conservatism and exceptional durability of styles over longer spans of time than was characteristic of neighboring regions (Connolly Reference Connolly1988; Connolly and Baxter Reference Connolly, Baxter and Kenneth1986).
With the Mill Creek data, we address two main points: first, although large and small points can be reliably sorted by metrics, they are not as clearly arranged temporally in our study area. Both large and small projectile tips were common in the Willamette Valley throughout the most recent millennia, bringing into question the link between point size and delivery system.
A second observation is that regardless of its initial presence, bow-and-arrow technology (for which smaller point tips were most often favored over smaller-diameter arrow shafts) accelerated throughout the Americas after about 2,000 years ago, and in the Willamette Valley after around 1,800 years ago (Pettigrew Reference Pettigrew1980, Reference Pettigrew1981). Three primary drivers can be considered: (1) Butler and Campbell (Reference Butler and Campbell2004) document an intensified focus on deer in the Pacific Northwest during recent millennia—a change that could affect point frequency; (2) Bettinger (Reference Bettinger and Beck1999:69) argues that the later increase in point frequency may relate to accelerated population growth in the western Great Basin and California, an explanation that may also extend to western Oregon; and (3) it may also be the case that the increase in projectile tips coincides with heightened levels of conflict and concern with boundary maintenance in western Oregon, as it was throughout the Pacific Northwest (Chatters Reference Chatters, William and Kuijt2004; Maschner Reference Maschner, Debra and David1997; Sampson Reference Sampson1985). We examine these factors as they relate to the Willamette Valley and note that point size likely varies with application and context, quite apart from the mode of delivery.
The Mill Creek Sites
The Mill Creek Archaeological Site Complex consists of 11 sites associated with present or former channels of Mill Creek, a first-order tributary of the Willamette River in the central Willamette Valley, Oregon (Figure 1). The Mill Creek sites are 4.8–8.0 km (3–5 mi.) above the Willamette confluence; the City of Salem has developed along lower Mill Creek, overprinting the Native settlement of Chemeketa, a principal village of the Santiam Kalapuya.
Location of the Mill Creek Complex sites in Oregon’s Willamette Valley. Image by Thomas J. Connolly.
Large-scale excavations were conducted at eight Mill Creek sites (35MA7, 35MA9, 35MA11, 35MA12, 35MA63, 35MA64, 35MA65, 35MA70) intermittently from 1997 to 2003 in an area now occupied by the cloverleaf interchange of Oregon Highway 22 and Interstate Highway 5. The North Santiam River formerly flowed north through Turner Gap between the Salem and Waldo Hills (located immediately south of the Mill Creek sites), then through the project area. About 7,640 years ago, the eruption of Mount Mazama deposited substantial ash in the valley (Baitis et al. Reference Baitis, Gregory J. and James2015). Some time prior to 6,000 years ago, the North Santiam passage through Turner Gap was blocked, and the much smaller Mill Creek now occupies its abandoned corridor (O’Connor et al. Reference O’Connor, Sarna-Wojcicki, Wozniak, Polette and Fleck2001). It is not known if these are related events. The cultural record from the project sites spans the last 6,000 years and is secured by 78 radiocarbon dates; a few represent multiple age determinations from a single feature, but the vast majority are associated with individual cultural features, primarily earth ovens (Figure 2). Sites 35MA12, 35MA63, and 35MA70 appear to have been localities primarily focused on processing camas (Camassia spp.) where early oven dates predominate. Sites 35MA64 and 35MA65 are late and are more residential occupations. Sites 35MA7 and 35MA9 have both early camas ovens and an overlay of later-occupation debris.
Calibrated radiocarbon ages from the Mill Creek sites (median and 2σ error); those shown in gray are from features that are not unquestionably cultural, or that exhibit historic disturbance.
A notable aspect of the Mill Creek record was the common presence of earth ovens, most containing charred camas bulbs. Other frequently present botanicals included charred acorn and hazelnut shells and meats. Although badly deteriorated and fragmentary in most contexts, mammal bones were also relatively common; small and large mammals, fish, and fowl were identified, but deer were the most frequently identified, and nonidentifiable bone was dominated by deer-sized fragments. The locality was regularly used for 6,000 years to access resources of the Chemeketa Prairie (especially for camas), and for the last approximately 1,700 years, it was used as a seasonal—if not longer-term—residential place.
Throughout the project area, archaeological remains were limited to a single stratum deposited over a span of about 6,000 years that was, at most, around 50–60 cm thick. Due to the general lack of definable stratigraphy within this unit—coupled with the facts that much of the precontact cultural activity involved the digging of food-processing ovens and that some level of disturbance is attributable to bioturbation, agricultural practices, and highway construction—it was not always possible to reliably sort the assemblage into meaningful chronological components. The primary exceptions are two sites, 35MA64 and 35MA65, which date entirely (35MA65) or primarily (35MA64) within the last millennium. More than half of all recovered points are from 35MA65, and 74% of points are from these two sites, illustrating the dramatic increase in projectile points in very late contexts.
The Mill Creek Projectile Point Assemblage
The Mill Creek sites produced nearly 4,000 chipped and ground stone tools, and over 108,000 pieces of chipped stone debitage. Early deposits reflect short-term harvesting/processing sites dominated by camas oven features and only modest amounts of chipped and ground/pecked stone artifacts. Later long-term occupations reflect more diverse domestic and industrial activities. The two late (dating within the last ca. 1,700 years) mound sites—35MA64 and 35MA65—have high densities of chipped stone tools (projectile points, bifaces, unifaces, and drills/gravers, expedient flake tools, and flaked cobble choppers) and ground and pecked tools (stone bowls, pestles, manos, percussors, and abraders).
Projectile points constitute 44% of chipped stone tools recovered from the excavations, twice the number of edge-modified flakes (used flakes), a category of expedient tool that is often the most abundant. Such an abundance of projectile points illuminates their importance at the Mill Creek sites.
Projectile points are often categorized as arrow sized or dart sized, which are generally distinguished by neck width or blade width. Because notch position or stem configuration varies (for example, small side-notched points often have relatively broad neck widths, and stemless points have none), we used point width, weight, and thickness as distinguishing measures, following Fenenga (Reference Fenenga1953), Shott (Reference Shott1997), and Bradbury (Reference Bradbury1998) (see also Thomas Reference Thomas1978), whose discriminant analyses identified these as the most significant variables for separating arrow and dart points. This also increased our sample sizes, because broken proximal ends often make neck-width measurements impossible, and this metric cannot be measured on stemless points.
The Mill Creek projectile points had been previously sorted into 18 types based on morphology (stem and shoulder configuration) and metric attributes (e.g., neck width). Types 1–11 include points commonly identified as arrow tips, such as narrow-necked stemmed points, small side-notched points, and small foliates. Types 12–18 include projectile tips commonly identified as dart points, including broad-necked stemmed points, large side-notched points, and larger foliates (Figure 3).
Representative examples of the Mill Creek projectile-point types. (Color online)
With width, thickness, and weight measures, point types are sorted quite neatly into small (“arrow-sized”) and large (“dart-sized”) points (neck width is also bimodal, if side-notched points are excluded; Figure 4a). Foliate (stemless) points were initially sorted into “small” (Type 11) and “large” (Type 12) varieties based on width (Figure 4b). The small foliates grouped with the “dart-sized” points based on thickness and with the “arrow-sized” points based on width, but here they are considered with the smaller points—following Thomas (Reference Thomas1978) and Shott (Reference Shott1997), whose analyses found that width is the more reliable metric for distinguishing arrow and dart points. Small points (Types 1–11) typically have a width of less than 15 mm and a weight of under 2 g, whereas larger points are typically wider than 15 mm and weigh more than 2 g (Figures 5a, b). For thickness, small points generally measure 4.5 mm or less (Types 1–10), whereas more robust “dart-sized” points (Types 11–18) exceed this thickness (Figure 5c). Except for small foliate points (Type 11), all other types sort into the same large and small categories by all three measures.
Selected projectile point metrics: (a) neck widths for “arrow-sized” (Types 1–7) and “dart-sized” (Types 12–15 and 17–18), excluding side-notched (Types 9 and 16) and unstemmed (Types 8 and 10–12) points; (b) maximum width of small (Type 11) and large (Type 12) foliate points. (Color online)
Projectile-point metrics with mean and standard deviation by type: (a) maximum width, (b) weight, (c) thickness.
Despite less-than-ideal chronological controls for the artifact assemblage, some chronological patterning of typed projectile points can be tracked. Although six of the eight project sites showed intermittent use throughout most of the last 6,000 years, two sites date exclusively within the last 1,700 years, and both date primarily within the last 1,000 years. Eight of 10 radiocarbon ages from 35MA64 and all eight ages from 35MA65 fall within the last millennium. It is noteworthy that these two sites have significant proportions of dart-sized points; 41.8% at 35MA64 (102 of 244 points) and 25.5% at 35MA65 (150 of 588 points). These numbers suggest either that atlatls remained in common use alongside of bows throughout the last millennium or that the relationship between point size and function is more complex than is commonly assumed.
We also attempted to use obsidian hydration (OH) to evaluate the temporal order of projectile point types. Most obsidian at the Mill Creek sites was identified as being from the Inman Creek geochemical source, which is available in nearby river gravels. Obsidian Cliffs material is also present; its primary source is near the crest of the Cascade Mountains, which border the east margin of the Willamette Valley, but it can also be found in Willamette River gravels (Baxter et al. Reference Baxter, Connolly, Skinner, Terry and Ron2015). Hydration analysis was conducted on 271 projectile points from these sources (247 from Inman Creek and 24 from Obsidian Cliffs). Obsidian hydration rates for both sources have been estimated on the valley floor (Cowan Reference Cowan2019; Minor Reference Minor and Toepel1985; O’Neill et al. Reference O’Neill, Connolly and E. Freidel2004; Roulette et al. Reference Roulette, Wilson, Ellis and Chapman1996). We acknowledge the complexity of OH dating, but given that our primary goal for the current study is to evaluate relative ages of “dart” and “arrow” points rather than to assign specific ages, we have confidence that OH serves this objective.
Because arrow-sized points (Types 1–11) outnumber dart-sized points (Types 12–18) by a factor of 10, age estimates are presented as a percentage within each group (Figure 6). Surprisingly, the estimated age profiles of arrow versus dart points do not show the expected sequential separation. Instead they correspond rather closely, reaffirming that the dart-sized points at the Mill Creek sites overwhelmingly date within the last 1,500 years or so.
Estimated obsidian-hydration ages for “arrow” (Types 1–11) and “dart” (Types 12–18) points. (Color online)
Under the assumption that there is something of a chronological relationship between arrow and dart points, we ordered the Mill Creek sites by the relative proportion of arrow-sized and dart-sized projectile points (Figure 7). Surprisingly, site 35MA7 has the highest proportion of arrow points (80 of 101, or 80.2%), which would typically suggest substantial late use, even though eight of the 10 radiocarbon dates from the site (mostly associated with oven features) range in age between 4,500 and 2,000 years old. Site 35MA70—where 18 of 22 radiocarbon assays predate 2,500 years ago, and 14 exceed 4000 cal BP—also has a high proportion of arrow points (39 of 63, or 61.9%). Both sites also saw use during the last 1,200 years or so. At these sites, it may be the case that early use, which focused on camas processing, has been overprinted with substantial later use that was more focused on hunting, but the dominance of small points at these sites is at least suggestive that bow and arrow use was part of the weapon repertoire for millennia prior to 2000 BP.
Sites ordered by proportion of “arrow” versus “dart” points. (Color online)
Lithic material in the Mill Creek point assemblage (including unclassifiable fragments) includes cryptocrystalline silicates (CCS; 49.3%), obsidian (50.0%), and a small amount of fine-grained volcanics (FGV; 0.7%). Material type breaks down quite differently by type, however, with the small types (1–11) being 59% obsidian and 41% CCS (and <1% FGV); larger types (12–18) are 61% CCS and 37% obsidian (and 2% FGV). This may relate, in part, to the generally small size of obsidian nodules available in local stream gravels, making them unsuitable for the production of larger points (Pettigrew Reference Pettigrew1980:66).
Comparative Studies
Thomas (Reference Thomas1978) presented a discriminant statistical analysis using length, width, thickness, and neck width of curated, hafted, and complete projectile points determined to be arrows or darts based on the size of the hafted shaft. With metric data, he was able to correctly identify 87% of the 132 arrow tips and 70% (7 of 10) of dart tips. Following Thomas, Bradbury (Reference Bradbury1998) developed discriminant functions using only blade width and neck width to allow use of fragmented projectile points. That same year, Shott (Reference Shott1997) produced discriminant functions for each of the four variables Thomas had used, but in the end, he determined—as had Thomas—that width was the most significant variable in predicting the point function. Hughes (Reference Hughes1998) approached translating point morphology into point function from an engineering perspective. She presents two metrics based on width and thickness as determinative. Tip-Area (TCSA) is a measure of point sharpness to create an initial puncture, and Tip-Perimeter (TCSP) is a measure of the point’s ability to widen the initial puncture to limit shaft drag and improve depth of penetration.
We have used all five methods to compare the complete points in the Mill Creek assemblage (n = 334; Table 1). The purpose of this exercise is to emphasize that different metric-based classifications can produce different results. Although constructed classificatory systems such as these can be useful (or essential) heuristic devices, we are mindful that they are analytical constructs that may approximate functional reality to varying degrees. The 238 Type 1–11 (arrow-sized) points are all rated as arrow tips by Hughes’s method; other methods suggest that some are dart points (Shott 2%, Thomas 6%, and Bradbury 21%). These same studies assign many of the 96 presumed Mill Creek Type 12–18 (“dart”) points to the arrow category; proportions range from 90% (Hughes’s TCSA measure) to 44% (Bradbury). The considerable range of results shown by these comparisons amplifies the fact that if projectile point morphology is dependent on functional requirements that vary by environment, functional context, and cultural tradition, then diagnostic metrics will likely vary over both space and time. Hildebrandt and King (Reference Hildebrandt and King2012) devised what they called a dart-arrow index (DAI) for the Sierra/Cascade–Great Basin interface region, but Walde (Reference Walde2014) found that this index did not work well for the southern Canadian plains. Corliss (Reference Corliss1972) observed different arrow/dart neck-width modes for regions occupied by Sahaptins compared to those occupied by Numic speakers. Bettinger and Eerkens (Reference Bettinger and Eerkens1999:231–232) observed “regional differences in morphology and attribute correlation in generally similar point types” in the western Great Basin, resulting in conflicting type assignments using different typological systems. One clear outcome of this exercise is to confirm that an arrow point/dart point dichotomy is, well, imprecise.
Mill Creek Projectile Points (Small Types 1–11; Large Types 12–18) Classified by Discriminant Function Analyses Reported by Other Researchers.
1 Unnotched types (8, 10–12) were not included in calculations requiring neck-width measurements.
2 Tip Cross Section Area (TCSA); Tip Cross Section Perimeter (TCSP).
The essence of the issue is aptly summarized by Whittaker (Reference Whittaker2013:109):
It is critical to remember that all bows are not the same, all atlatls are not the same, and the conditions of use and the skills of individuals are enormously variable. Some atlatls were light and propelled light projectiles, others were much heavier; the same is true of bows, and the different forms were often used (sometimes side-by-side) for different game or under different conditions of cover and terrain.
Nelson (Reference Nelson and Knecht1997:371) points out that projectile points have been useful temporal, spatial, and functional markers, but that as archaeological focus has shifted “from the material to the behavioral realm, more emphasis is placed on the organization and context of technology.” The point (so to speak) is that simply assigning projectile tips to arrow and dart functional categories based on morphometric attributes masks practical and social realities that are more nuanced.
The Mill Creek Assemblage in Context
The increase in projectile point frequency, particularly during the last approximately 1,500 years, has been previously noted in the Willamette Valley (e.g., Miller Reference Miller and Aikens1975; Pettigrew Reference Pettigrew1980). O’Neill et alia (Reference O’Neill, Connolly and E. Freidel2004) observed that extensive excavations in middle-Holocene components (ca. 5,000–4,000 years ago) at the Long Tom and Chalker sites in the upper Willamette Valley produced extensive evidence for camas processing—comparable to the Mill Creek sites—but only about half a dozen projectile points. Excavation of a roughly comparable volume of late-Holocene (<2,000 years ago) deposits at these sites produced approximately 60 projectile points, a tenfold increase. At the Mill Creek sites, 74% of all points are from the two sites (35MA64 and 35MA65) dating entirely within the last approximately 1,700 years, representing just 28% of the occupation span of the Mill Creek sites. There could be multiple reasons for the proliferation of projectile points—particularly small points—in recent millennia. Here, we consider multiple possible factors and their implications for point form and size.
A Hunting Explanation for Increased Point Frequency
An increasing emphasis on hunting has been cited as a possible reason for the greater frequency of projectile points in late Willamette Valley sites (Miller Reference Miller and Aikens1975; Pettigrew Reference Pettigrew1980). This may be one explanation. Butler and Campbell (Reference Butler and Campbell2004) document an increase in use of artiodactyls over time on the Northwest Coast and Columbia Plateau, which they correlate to anthropogenic or other environmental changes that improved forage quality and habitat extent. In the Willamette Valley, anthropogenic fire management is well documented (Boyd Reference Boyd and Boyd1999; Coughlan et al. Reference Coughlan, Johnston, Derr, Lewis and Johnson2024; Walsh et al. Reference Walsh, Duke and Haydon2018). Numerous early accounts note the use of fire as it relates to hunting deer; for example, in 1825, David Douglas (Reference Douglas1914:66) commented on the scarcity of deer in one area of the valley “in consequence of the plains being burned by the Indians to compel these animals to seek food in certain parts more convenient for hunting.”
For the Mill Creek sites, faunal preservation was very poor, with nearly all recovered bone (98%) being unidentifiable or identifiable only to animal size class. Identifiable bone was mostly deer, and most size-classed bone was deer size, but due to very poor bone preservation in the valley’s acidic soils, noncalcined bone was almost exclusively from very late contexts.
However, small calcined bone fragments were identified in flotation samples from 48 directly dated features, mostly earth ovens, spanning the entire occupation sequence. Although an age bias due to poorer preservation in older features may remain a factor, normalized counts of calcined bone fragments by number of features in each millennial time span (Figure 8) suggests that artiodactyl hunting may have increased through time, consistent with the findings of Butler and Campbell (Reference Butler and Campbell2004).
Normalized counts of calcined bone fragments in flotation samples from directly dated features, by millennial increments.
A Population Explanation for Increased Point Frequency
Population increase is another possible reason for a dramatic increase in projectile points. Based on the relative frequency of cultural radiocarbon dates, Cheatham (Reference Cheatham1988) argued that the Willamette Valley’s population began a dramatic increase by about 2,000 years ago. Although the use of radiocarbon dates as a population proxy is problematic (e.g., Crema Reference Crema2022), Walsh et alia (Reference Walsh, Pearl, Whitlock, Bartlein and Worona2010) note a peak in anthropogenic burning in the valley from around 1500 to 500 cal BP, which suggests intensified food management for a growing population.
Bettinger (Reference Bettinger2015) notes that California and Great Basin populations explode after about AD 450 (within the last ca. 1,500 years), coinciding with subsistence intensification. Rather than intensification being driven by population growth (Boserup Reference Boserup1981; Keeley Reference Keeley1988), he argues the opposite: “energy maximizing” fueled growth (Bettinger Reference Bettinger2015:40). Bettinger makes a convincing case that this change was at least partially due to the introduction of the bow and arrow, which altered hunting modes and permitted reduced residential mobility—factors more conducive to intensification of plant foods. This, in turn, promoted the privatization of stored plant foods as well as the physical places from which they were harvested.
One tool of resource intensification is anthropogenic burning, which was done for a variety of reasons but mostly to enhance the range, density, and productivity of critical foods (Boyd Reference Boyd and Boyd1999; Carney and Connolly Reference Carney and Connolly2024; Walsh et al. Reference Walsh, Duke and Haydon2018). Greater time and energy investments in landscape management led to increasing proprietary interest in lands and resources and a heightened interest in boundary maintenance. Zenk (Reference Zenk1976:46) reports that each Kalapuya village held exclusive rights of access to community resources—in particular, that managed fields of seed-producing tarweed were owned by each village group, and plots within these allotments were, in turn, individually owned (Zenk Reference Zenk1976:17, 58). In the Willamette Valley, home territories were well defined and defended, as is made clear by this translation by Jacobs:
The Tualatins hunted half way in the mountains. . . . Perhaps if they [the Tualatin] crossed to the Yamhill country a man who hunted (there) might get killed. (Beyond) half the mountain at pa´naxDin [the northernmost Tualatin village] if (the people of that village) should cross over (the mountain) to Clatskanie country, perhaps (the villager) would be killed. If a Clatskanie should cross over, possibly the Clatskanie would be killed (by a Tualatin) [Reference Jacobs1945:187–188; brackets added for clarity].
The record is clear that access to another’s territory was not restricted for legitimate purposes, such as trade, which was conducted extensively throughout the region, and access to certain resources might be allowed with proper permissions, particularly among parties connected by marriage, but some proprietary resources were always off limits to outsiders. For example, the Tualatin sometimes went to the Columbia River to hunt seals (probably in the territory of the Multnomah Chinook), but they never went there to fish; they were allowed to collect lampreys at Willamette Falls—where they traded for salmon with the resident Clackamas Chinook—but were not permitted to fish with dip nets (Zenk Reference Zenk1976:49–50).
A Conflict Explanation for an Increase in Point Frequency
Resource intensification, increased population, and sedentism resulted in consolidated stored resources and property ownership requiring increased concern for defense and warfare. Chatters (Reference Chatters, William and Kuijt2004) has convincingly argued that in the Pacific Northwest, the emergence of village communities (in contrast to earlier, widely scattered households), concealed food storage, village location in defensible settings, and the more frequent occurrence of arrow wounds on skeletal remains coincides with the appearance of bow-and-arrow technology. At the Nightfire Island site in the Klamath Basin, near the Oregon/California border, Sampson (Reference Sampson1985:514–515) reports that large dart points are mostly replaced by small arrow points about 1,700 years ago, accompanied by cemeteries with abundant evidence of violent death, as well as evidence for enhanced long-distance contacts. He speculates that evidence for a broadening of long-distance trade was accompanied by “copious hints of raiding and reprisal-seeking counter raids” (Sampson Reference Sampson1985:515). Others have noted that increasing sedentism and social complexity accompany the appearance of bow-and-arrow technology throughout North America (Bingham et al. Reference Bingham, Souza and Blitz2013; Kennett et al. Reference Kennett, Lambert, Johnson and Culleton2013; Maschner and Mason Reference Maschner and Mason2013; Reed and Geib Reference Reed and Geib2013). Lambert (Reference Lambert, Richard and Mendoza2007) argues that an escalation in violence becomes an attractive option when populations become sedentary because mobility or other strategies for coping with local shortages have diminished.
Sapir (Reference Sapir1907:252, 272) notes that some western Oregon groups “were accustomed to make raids” on their neighbors “to procure supplies of food and other valuables,” and that the “principal weapon of offensive warfare was of course the bow and arrow.” One incentive for raiding in the Willamette Valley was to secure slaves, who were often taken to trading centers on the Columbia River (Gatschet et al. Reference Gatschet, Frachtenberg and Jacobs1945:191, 193, 349; Jacobs Reference Jacobs1945:41). Zenk (Reference Zenk1976:5–6) reports that the Tualatin brokered in slave trading, and “at least occasionally conducted slave-raiding expeditions in such areas as the southern Willamette Valley and the central Oregon coast.” Work (Reference Work1923:264) noted in his journal that the Marys River (“Lautaude”) Kalapuya were on alert for a retaliatory raid from the Umpqua, in response to their prior raid. He continued, “They also inform us that 4 men of Lautaude Indians have been killed and 3 children taken slaves a short time since, as they suppose by a party of Faladin [Tualatin] or Yamhill Indians.”
Slavery was established on the Northwest Coast by 3500 years BP (Ames and Maschner Reference Ames and Maschner1999:187), but it was facilitated by the introduction of the horse and amplified during the Fur Trade Era of the early nineteenth century. The Chinook and their Tualatin Kalapuya neighbors held slaves but were better known as slave traders who preyed on the Central and Southern Kalapuyans (Jetté Reference Jetté2015:17). Stern (Reference Stern1996:236) reports that horse-mounted Klamath “had become accustomed to traveling down the middle fork of the Willamette to trade with the Clackamas and other Chinookans at the Falls, occasionally raiding the Yoncalla Kalapuya enroute for slaves to bring along.” Slave raiding became so intense that horse-mounted Modoc, Paiute, and Klamath raiders created severe population attrition among the Californian Atsugewi (Garth Reference Garth and Robert1978:238).
Conflict and Implications for Projectile Point Form
We can return to specific Willamette Valley projectile-point forms when considering warfare, as opposed to hunting. The two deeply serrated point types—7 (contracting stem) and 8 (stemless)—are a common form in the Willamette Valley (familiarly referred to as Kalapuya or Christmas tree types). All but one (97%) are obsidian. Edge serration maximizes internal hemorrhaging, but the deep serrations that characterize these types also facilitate breakage on impact. Fragments of a point left in a victim increase the lethality effect: “It did not matter if an enemy died later from infection rather than immediately after being shot, as the desired result was the same” (Engelbrecht Reference Engelbrecht2015:764; see Nelson Reference Nelson and Knecht1997:378; Wilkins et al. Reference Wilkins, Schoville and Brown2014:3). These may have been designed and have served specifically as “war points.”
Keeley (Reference Keeley1996:52) reports that several California tribes used tightly hafted notched points on hunting arrows that could be potentially recovered and reused. However, for war, loosely bound stemmed points were used so that when the shaft was extracted the point, or part of it, would remain in the wound. Dr. Elliott Coues (Reference Coues1866) treated arrow wounds in Arizona Territory in 1864, where he observed the following about Apache war points: “The head is apparently a small and trifling affair . . . so frail is the connection between the head and shaft, that the Indians themselves are obliged to carry their arrows with great care” (Coues Reference Coues1866:351), and due to “the extreme friability of the head . . . the chances of its shivering into bits vastly preponderate over the probability of its becoming fixed or glancing” (Coues Reference Coues1866:353). One observer noted of Yurok weaponry that obsidian arrow tips “are the most dangerous. . . . If they penetrate a human body, the glass generally splinters on the bones, the wound promptly festering with fatal results” (Heizer and Mills Reference Heizer, Mills, Heizer and Mills1952:124).
Body target areas also differ for hunting and warfare. For a quick, clean hunting kill, heart and lungs are the primary target. Abdominal areas are to be avoided, given that abdominal contents can contaminate and infect meat, and modern medical advice asserts that meat contacted by intestinal fluids is unfit for eating (American Veterinary Medical Association 2024; Rinella Reference Rinella2017). In contrast, US Army surgeon Joseph H. Bill (Reference Bill and Ashhurst1882) observed, “Wounds which penetrate the abdominal cavity, and injure its vessels or viscera, are the most fatal. Knowing this, the Indian, if he has time to deliberate, points his arrow and lance at the abdomen” (Bill Reference Bill and Ashhurst1882:107). He noted that for abdominal arrow wounds, a few bled to death, but most died from peritonitis—infection due to breach of the intestinal cavity (Bill Reference Bill1862:385).
Implications of Hunting Methods and Context for Projectile Point Form
Although a case can be made that certain points (particularly Types 7 and 8, and possibly stemless Type 10) might be more suitable as “war” points than for hunting gear, we are left with considering the common presence of larger “dart-sized” points in very late contexts. One explanation for this might be found in the journal of fur trapper Alexander Henry (Reference Henry and Barry1992:661), who in 1814 recorded a “rather singular” method of deer hunting in the Willamette Valley: a Native man used a deer-head decoy to lure animals “near him.” This technique was also reported by Louis Labonte (son of an employee of Astor’s Pacific Fur Company and his wife, the daughter of a Clatsop chief), using a deer head to mimic the movements of a deer “until the victim was near” (Lyman Reference Lyman1900:172). In 1826, botanist David Douglas also recorded this decoy technique by the Kalapuyans, which lures the “unsuspecting animal . . . within a few yards” (Audubon and Bachman Reference Audubon, John and Victor1967:280). Another method used by the Santiam Kalapuya, reported by pioneer John Minto (Reference Minto1900:307), is a circle hunt that drove deer to a point that allowed “the bowmen to get close to the quarry.” Given these hunting strategies, which provided close proximity to prey, it may be that larger arrow tips were favored, because they had the advantage of maximizing injury over the need for greater flight distance achieved with a smaller tip (Grossman Reference Grossman2021; Hughes Reference Hughes1998; Wells Reference Wells2024). This argument is enhanced by the frequency of deer (or deer-sized animals) in the Mill Creek faunal assemblage.
Conclusions
The arrow/dart dichotomy reported throughout most of the Americas is less distinctly demarcated chronologically in the Willamette Valley study area. If we accept that bow-and-arrow technology—for which smaller point tips were generally favored for smaller-diameter arrow shafts—accelerated throughout the Americas after about 2,000 years ago, we are left with the possibilities that (1) atlatls continued to be used regularly as a companion weapon in the Willamette Valley (although there is no direct evidence or historic references to atlatls); (2) both large and small projectile tips were affixed to arrows, depending on the target (large or small game, close range or distant range, warfare/defense); or (3) some combination of these factors was the case.
It is clear that the frequency of small stone projectile tips dramatically increased in the Willamette Valley in recent millennia, as is the case throughout North America. It is equally clear that the use of larger projectile tips continued and accompanied the proliferation of smaller points. We have noted multiple explanations for increasing numbers of points (population growth, increased hunting, and escalating conflict), but beyond actual numbers, we acknowledge that varying objectives (e.g., durable points for hunting vs. fragile points for warfare) as well as use contexts (large vs. small prey, open landscape vs. close-quarter hunting conditions) also influenced point design and use.
It is also notable that small points are relatively common in contexts that produced middle-Holocene radiocarbon dates at the Mill Creek sites, although the association between dated features and proximate nonfeature sediments can be questioned. Nonetheless, the apparent association raises the question of earlier bow-and-arrow presence in the valley. Or, as some experimental researchers have found, small points can be used effectively with atlatl darts (Browne Reference Browne1938, Reference Browne1940; Evans Reference Evans1957; Fenenga Reference Fenenga1953:319; Thomas Reference Thomas1978:466). Experimental and ethnographic data confirm that the separation of arrow and dart tips strictly by size is not clear cut; Hughes (Reference Hughes1998:374) reports that unfletched dart tips typically weigh 9 g and more, whereas fletched dart tips are 8 g or less, and the range of arrow tip weights (up to 11 g) overlaps both dart ranges. Projectile points can be considered within two fundamental frameworks. One is based on formal attributes such as style and metrics. The other is point function, or the contexts in which points were used as they relate to hunting, conflict, and compatibility with delivery systems. Although it is clear that stylistic/metric patterns have meaning, it is all too easy to conflate formal and functional frameworks, leading to assumptions that arrow/dart-point classifications constructed by archaeologists have functional correlates that are more fixed than real-world applications are.
In the final analysis, we find that separating stone projectile tips into small- and large-sized categories is a relatively easy task, but whether these size categories directly relate to arrow or dart projectiles—or have reliable chronological meaning—is not at all clear. Optimum projectile-tip size and form are driven by a more nuanced set of factors than the delivery system alone.
Acknowledgments
Fieldwork for this study was conducted in 1996–1997 and 2002–2003, in coordination with the Oregon Department of Transportation, the Confederated Tribes of Siletz, and the Confederated Tribes of Grand Ronde. Work was performed under permits AP-234 (1996–1997) and AP-493 (2002–2003), issued by the Oregon State Historic Preservation Office. We thank Guy Tasa—who directed fieldwork and cataloging—and Dennis Jenkins, David Lewis, Richie Rosencrance, Geoff Smith, Paul Salimano, David Hurst Thomas, and two anonymous reviewers for their input.
Funding Statement
Initial work was supported by the Oregon Department of Transportation; subsequent analyses were partially supported by ODOT and the University of Oregon Museum of Natural and Cultural History.
Data Availability Statement
Associated materials and records are curated at the University of Oregon Museum of Natural and Cultural History (the Oregon State Museum of Anthropology) under accession numbers 1008 and 1539.
Competing Interests
The authors declare none.
