Temporal Range

Our interest in the Rosegate chronology was sparked by excavations at Wolf Village, located at the south end of Utah Valley. This site has nine excavated structures and an unusually large assemblage of artifacts for the Fremont region (Johansson et al. 2014; Lambert et al. 2016). An early occupation (Wolf Village 1) most likely dates to the late AD 600s. A total of 61 projectile points were found associated with two activity areas dating to the early occupation (Lambert et al. 2016). Of the 61 projectile points found in these areas, 87% (n = 53) are Rosegate points; almost all the other points are either dart points or indeterminate points. The one exception is a Nawthis side-notched point that was found immediately below the modern surface, and probably dates to the later occupation of the site (Wolf Village 2). The lack of other Fremont projectile point types fits the expected chronology and matches the radiocarbon dates; however, 137 Rosegate projectile points were found in later contexts, comprising 19% of all projectile points from the later occupation. Radiocarbon and tree-ring dates place this occupation between the mid AD 1000s and early 1100s (Johansson et al. 2014:35; Lambert 2018:41). No early dates were obtained from these areas, and none of the structures were superimposed over substantially earlier occupations. The Rosegate points were mostly found associated with floors or the fill of the structures and were found in association with other point types. These Rosegate points were not associated with any evidence for an earlier occupation, and the associated assemblages all appear to date at least several decades, and possibly more than a century, after AD 1000. The initial analysis of projectile points was based on Holmer and Weder’s chronology, which suggested the presence of Rosegate points in the later assemblages was anomalous and led to the questions we address in this study.

In order to determine whether Rosegate points are commonly found after AD 1000 in other Fremont sites, we assembled a database of 21 sites containing 5,744 projectile points and 1,707 Rosegate points (Table 1; Figure 3 displays site locations). The sites included are a subset of excavated Fremont sites. We first looked for sites in Utah Valley, close to Wolf Village, and then expanded our search to other Fremont sites creating an opportunistic sample. We concentrated on sites with a large number of analyzed projectile points. We also primarily looked for reports that used Holmer and Weder’s typology, to better maintain consistency in the data. Because we are primarily interested in the relationship of Rosegate points to other points associated with the bow and arrow, we calculated percentages of Rosegate points to the total number of arrow points. We removed all point types usually associated with the Archaic period, or more accurately with the atlatl. Small numbers of these points were found at nearly every site and may have been used as knives (Weder 1980:44). Hildebrandt and King (2012; see also Weder 1980) outline a convenient method of discriminating between dart and arrow points by adding the neck width and maximum thickness. This method was used as a general, but not absolute, guideline for the Wolf Village points, as well as for points from recent excavations at the Hinckley Mounds site in Provo. When working with published data from other sites, we assigned points as dart points based on the reported type. There are likely a few errors in assigning points to arrows or atlatl darts, but this distinction is important in this study as some sites—particularly caves—had long occupations that often predate the arrival of the bow and arrow; thus, removing dart points provides a more accurate picture of how common Rosegate points are versus other arrow points in the assemblages. Percentages stated below are based on the number of Rosegate relative to the total number of arrow points.

Figure 3: Locations of sites in this study and sites used by Holmer and Weder (1980) in their Rose Spring analysis (sites mentioned in the text are labeled).

Table 1: Fremont Projectile Point Data from this Study

Site	Period	Total Projectile Points	Total Rosegate Points	Total Dart Points	References
American Fork Cave (42UT135)	Mixed	115	57	42	Janetski and Smith 2006
Backhoe Village (42SV662)	Mixed	51	46	0	Seddon 2001
Baker Village (26WP63)	Late	546	142	7	Wilde and Soper 1999
Block 49 (42SL98)	Late	35	12	3	Talbot et al 2004
Bull Creek\(^a\)	Mixed	76	6	7	Jennings and Sammons-Lohse 1981
East Fork Village (42MD974)	Mixed	38	31	1	Reed et al. 2005
Five Finger Ridge (42SV1686)	Late	199	73	8	Talbot et al. 2000
Bull Creek\(^b\)	Mixed	20	2	6	Mooney 2014
Hinckley Mounds 2 (42UT111)	Late	41	3	9	Unpublished data
Hunchback Shelter (42BE751)	Mixed	473	241	90	Reed et al. 2005
Kay’s Cabin (42UT813)	Late	314	26	14	Janetski 2016
Lost Ridge (42UT635)	Early	10	6	4	Janetski and Smith 2006
Mosquito Willie (42TO137)	Early	120	104	11	Young et al. 2008; Janetski 2006
Mud Springs Site (42IN218)	Late	10	3	3	Reed et al. 2005
Paragonah (42IN43)	Mixed	321	53	6	Woods 2009
Parowan (42IN2262)	Mixed	390	90	10	Woods 2009
Scorpio Site (42WS2434)	Mixed	45	12	3	Reed et al. 2005
South Temple (42SL285)	Late	42	19	1	Talbot et al 2004
Spotten Cave (42UT104)	Mixed	61	30	18	Woods 2004
Summit (42IN40)	Mixed	1790	500	72	Woods 2009
Wolf Village 1 (42UT273)	Early	61	53	0	Unpublished data
Wolf Village 2 (42UT273)	Late	727	137	10	Unpublished data
Woodard Mound (42UT102)	Late	255	61	13	Richens 1983
Note:
\(^a\)Bull Creek incorporates several site numbers: 42WN226, 42WN230, 42WN231, 42WN261, 42WN326, 42WN337, 42WN991, 42WN996.
\(^b\)This site is composed of several mounds that were excavated and reported at different times.

To evaluate the chronology, we compiled a database of 218 radiocarbon dates from published sources and new dates (see supplemental table). We also compared the dates with the chronological estimates given by the authors of each report to ensure contextual information and other dating methods did not affect the analysis. Dates were calibrated using the intcal20 standard (Heaton et al. 2020) and converted to calendar years using the Bchron package (Haslett and Parnell 2008) in R (R Core Team 2020). Many older radiocarbon dates suffer from the well-known old-wood problem common in Fremont archaeology (Allison 2016); thus, where possible, we focused on more recently taken dates from annual plants (mostly maize), including a number of previously unpublished dates.

Figure 4: Histogram of radiocarbon dates used in this study.

Figure 5: This plot shows the sites in this study arranged by median date. Radiocarbon dates from annual plants (usually maize) are shown as circles, whereas all radiocarbon dates including annuals are shown as triangles (sites with only annual dates are only shown once).

Figure 4 shows the distribution of dates, which peaks at around AD 1100. Figure 5 shows the sites in this study arranged chronologically according to median radiocarbon date. Medians of the dates on annual plants are depicted separately from the median of all dates, although for most sites the annual plant dates are not significantly different—Backhoe Village being the primary exception. Most sites have median dates after AD 1000, but each site contains Rosegate projectile points. Figure 6 shows the median dates compared to the Rosegate percentage at each site. There is a clear decline over time, but the decline slows at around AD 1000, and the loess smoothing line (see Cleveland and Devlin 1988) never dips below 25%. There is also much greater variation after AD 1000. A linear regression of these data returns an adjusted r² of 0.31 (F = 9.7, df = 22, p = 0.01). Rosegate points do decline over time, but they cannot account for most of the variation in dating for these sites.

Figure 6: Rosegate percentages plotted against the median calibrated radiocarbon date for each site with a loess smoothing line.

To simplify the chronology, we have divided the sites into early and late periods for some analyses. Sites dating before approximately AD 1000 are defined as early, and sites dating after this period are defined as late. Several sites did not have accurate enough data to easily separate the assemblages, or the date ranges significantly overlap the early and late periods and cannot be confidently assigned to one or the other. A conservative approach was taken when assigning sites to periods and most sites were placed in the mixed category. Even in sites confidently assigned to the late period, Rosegate percentages ranged as high as 46% and were never lower than 9% (see Table 2 for percentages by period and Figure 7 for all site percentages).

Figure 7: Dot plot showing the percentage of Rosegate points to arrow points.

Table 2: Numerical Summary of Rosegate Projectile Points by Period

	Early	Late	Mixed
Number of sites	3	9	11
Total Rosegate points	163	476	1,068
Total Rosegate %	93	23	34
Mean site %	94	28	46
Median site %	95	26	29
Minimum site %	87	9	9
Maximum site %	100	46	90

Only three sites in the dataset are assigned to the early period: Mosquito Willies, Lost Ridge, and the early component at Wolf Village. Rosegate points make up 94% of the combined projectile point assemblages from these sites. Most of the other points are undetermined points that are either too fragmented to classify, or else represent uncommon or untyped variants. The high percentage of Rosegate points for these sites substantiates the current view that Rosegate points are the earliest arrow points in the Fremont region. The sample size in this dataset is small, but no other arrow points are known to be produced in this period.

Nine sites clearly date primarily to the late period. The percentage of Rosegate points drops to 23% in the combined late assemblages, but with considerable variation between sites. The South Temple site has the highest percentage of Rosegate points in this period at 46%, but the Kay’s Cabin site only has 9%. Notably, several late sites have large numbers of Rosegate points; Baker Village has the most with 142 out of 539 points (26%).

Most of the sites could not clearly be dated entirely after or before AD 1000. Rosegate points comprise 34% of points from the sites assigned to the mixed category. Long or multiple occupations, such as American Fork Cave, are one cause of dating problems, but in many cases, the radiocarbon error range is too large to rule out earlier dates. Allison’s (2016, 2019) analysis of radiocarbon dates indicates that Fremont populations rapidly increased starting at about AD 1000 and it is probable that most of the points in these sites date after the presumed end of Rosegate points. Regardless, the sites that clearly date after AD 1000 have too many Rosegate points to be attributed to the curation of old points.

Some sites have widely differing radiocarbon dates and one possible explanation for the occurrence of Rosegate points in apparently late assemblages at these sites is mixed stratigraphy. But Rosegate points have been found on structure floors postdating AD 1000 at Wolf Village, Woodard Mound (Richens 1983:72), Five Finger Ridge (Talbot et al. 2000:Table 6.24), Paragonah (Woods 2009:Table B.3), and other sites. While no provenience is necessarily conclusive, close association with a use-surface is usually a good indication that the artifact was used by the occupants of the structure. More Rosegate points are found in the fill of structures than floor contexts, but these points are often mixed with other point types, and we find it improbable that the fill of these structures is commonly composed of a thoroughly mixed combination of older and later artifacts. Rosegate points clearly continue to be used until the end of the Fremont period, although Rosegate points are often less common than other types in post-AD 1000 assemblages.