Caddoan Archaeology in the Little Cypress Creek Valley: Recent Investigation at the Griffin Mound Site (41UR142), Upshur County, Texas

Repository Citation Nelson, Bo; Perttula, Timothy K.; and Turner, Mike (1994) "Caddoan Archaeology in the Little Cypress Creek Valley: Recent Investigation at the Griffin Mound Site (41UR142), Upshur County, Texas," Index of Texas Archaeology: Open Access Gray Literature from the Lone Star State: Vol. 1994 , Article 8. https://doi.org/10.21112/.ita.1994.1.8 ISSN: 2475-9333 Available at: https://scholarworks.sfasu.edu/ita/vol1994/iss1/8


INTRODUCTION
As part of the long-term study of the prehistoric archaeology of the Caddo peoples in Northeast Texas, we are currently focusing our investiga~ions on the Little Cypress Creek valley in Upshur County.
Although poorly known archaeologically (Thurmond 1985(Thurmond , 1990, background research conducted to date, discussions with landowners, and selected surveyl imited testing efforts over the last few years indicates that there are extensive Archaic and Caddoan archaeological remains preserved in the Little Cypress Creek valley (Nelson 1993;Nelson andPerttula 1993a, 1993b;Hori zon Environmental Services, Inc. 1993, 1994 . Caddoan period archaeological sites (ca. A.O. 800-1600) are particularly common . The investigations of one of the more significant Caddoan sites found to date in the valley, the Griff in Mound site {41UR142), is the subject of this paper.

SITE SETTING
The Griffin Mound site is a natural mound situated at the base of a steeply sloping upl and landform i n the Caney Creek valley. Caney Creek, about 220 m from the site itself, flows southeast about 10 km to its confluence with Little Cypress Creek north of the community of Enon.
The mound is about 1-1.5 min height and covers approximately 900 m 2 (ca . 0.2 acr es) . Currently, the site is in a pasture. The site surface is pocked with gopher mounds, and a feeder station has disturbed a small area . Based on excavation profiles, t he natural mound is composed of 80 cm of a dark brown sandy loam midden deposit overlying a yellowish-brown sandy loam/loam that extends to at least 130 cm below the present ground surface.

INVESTIGATIONS
The Griffin Mound was recorded by Bo Nelson in the summer of 1992. He noted the presence of cultural materials on the surface of the natural mound, primarily exposed on gopher mounds. The disturbed soil in the gopher mounds was stained black, which he thought was indicative of subsurface midden deposits. Collected from the site's surface were 260 6 artifacts : 27 ceramic sherds (inc luding two engraved rim sherds and six decorated body sherds), 205 pieces of lithic debris, two quartzite cobbles, one core, one arrowpoint fragment, a dart point, one piece of daub, and 22 animal bones or bone fragments. Bob D. Skiles, then of the U.S. Forest Service, Lufkin, Texas, and Joshua S. Nelson) excavated four shovel tests and a single 1 x 1 m unit at the site to better define the site's vertical extent, and to determine the integrity and content of midden deposits apparent on the natural mound ( Figure 1). Shovel tests actually documented that the midden is distributed over most of the natural mound, with these cultural deposits ranging from ca. 38-92 cm in thickness; the shallowest deposits are present along the eastern side of the mound nearest the base of the uplands .
Shovel test 4 had a noticeably higher density of artifacts, as well as large sherds and numerous faunal remains, than the other shovel tests, and most of the artifacts occurred deep in the deposits. This suggested that the shovel test had encountered a cultural feature and/or artifact concentration. Accordingly, a lxl m unit was laid out immediately adjacent to the shovel test to investigate the potential cultural feature.
The lxl m unit, Test Unit 5, was excavated in 10 cm levels within recogni zable cultural strata (i.e . , plow zone and midden) using shovel and trowel; all soil matrix was screened through 1/4-inch mesh hardware cloth.
The excavations first documented a 20 cm thick plow zone (Zone l) of brown sandy loam; artifact density in the plow zone was 135/m 3 (primarily lithic debris and pottery sherds). Zone 2 is a 60 cm thick midden of dark brown sandy loam . Artifact density in the midden increased to 300/m 3 , with lithic debris, faunal remains, pottery sherds, and charred nutshells particularly common. As the excavations progressed to greater depths, the soil became increasingly water-saturated. This made excavating and screening the test unit difficult and tedious, but did help to heighten the contrast between the midden and the underlying yellowish-brown sediments when the unit reached approximately 80 cm below surface (bs). At that depth, the edges of a large pit (Feature 1) were visi-ble in the floor of the unit (Figure 2) . The pit fill (a black sandy loam) proved to have abundant cultural materials (600 artifacts/m3), principally charred nutshells, pottery sherds, lithic debris, and faunal remains. A radiocarbon sample of charred nutshells was secured between 110-130 cm in the feature, while a sample of 20 liters of feature fill was collected between 80-90 cm bs.
Feature 1 was excavated by sectioning it along an east-west line, and removing the cultural deposits in only the northern half of Test Unit 5; excavations at 80 cm bs in the southern half of the unit, and at 130 cm bs in feature fill in the northern half. The total depth of Feature 1 was not determined because the feature fill below 130 cm bs was extremely saturated and virtually impossible to properly excavate. Rather than risk damaging the lower feature deposits, excavations in Test Unit 5 were terminated at 130 cm bs. The unit was then backfilled, and the ground surface returned to its original contour.

FEATURES
A large pit was identified in Test Unit 5 at a depth of 80 cm; this is at the base of the midden. The pit fill was a black sandy loam (l0YR 2/1) with occasional yellow-brown mottles. Based on the plan of the feature as exposed in the unit's floor, the pit extent is about 2 min diameter. Feature 1 is at least 50 cm deep (80-130 cm bs) ; its total depth was not ascertained because the pit fill below 130 cm was too saturated to excavate.
We speculate that Feature 1 represents a storage facility, simi-8 lar to those seen on Caddoan sites from the Sabine River to the Red River. In fact, large unlined pit features have been noted to occur in proximity to Caddoan structures throughout the r e gion.
There is no evidence o f heat i ng, or in situ firing activities within the feature, and there are no disc rete lenses of trash in the fill. Although the feature is filled with homogeneous midden deposits, we would not expect that trash dumping was i ts primary function because it would be a considerable expenditure o f effo rt -... to excavate a large pit for trash disposal when the rest of the extensive midden debris on the site was clearly simply dumped on the ground surface and then built up to an appreciable thickness over Volume 5. Number 3 time . That the feature fill is comprised of midden sediments probably indicates that upon abandonment of the feature, it became filled with surrounding midden deposits.

ARTIFACT ASSEMBLAGE
The surface collection, shovel testing, and 1 x 1 m unit excavations recovered 946 artifacts and 680 pieces of charred nutshells and faunal remains. The greatest amount of artifacts and subsistence remains come from the contents of Feature 1, the large pit with abundant amounts of charred nutshells, fauna, and burned clay/daub pieces (Table 1) . To date, the nutshells and fauna l remains have not been identified or analyzed by the appropriate specialists .

Lithic debris.
The largest category of lithic artifacts from Griffin Mound is the lithic debris (n=477), the flakes and chips from tool manufacture and resharpen ing activities . The flotation results indicate that small (< 1 cmi in size) pieces of lithic debris are particurarly common in the _Feature 1 fill (Table 1) . Cortical pieces comprise about 18% o f the lithic debris.
The vast majority of the debris ( 88%) is on locally obtained quartzite, petrified wood, hematite, ferruginous sandstone, and cherts .
The local cherts and quartzites (ca. 85% of the debris) are reddish-brown, light tan, gray, rust red, and yellow in color, and they are similar in appearance to lithic raw materials obtained in terrace a n d stream gravels . Hematite (0 . 4%), ferruginous sandstone (1 . 7%), and petrified wood (2 . 1%) occur in low numbers in the lithic debris, 9 although these types of raw material are usually well represented in Late Archaic occupations in the region (Perttula et al. 1986) . The few pieces of firecracked rock are on coarse-grained quartzites.
The possible nonlocal lithic debris includes small flakes of some Red River gravel cherts (5 . 7%) and grayish-white, orange , and reddishbrown novaculite (5 . 7%). Most of the nonlocal debris was collected from the surface of the site; in e x cavated contexts from Test Unit 5, these materials represent only about 5% of the lithic debris sample . The low frequency of nonlocal materials i n the Griffin Mound lithic debris is consistent with Middle-Late Caddoan assemblages in the Lake Fork and Little Cypress Creek basins (Perttula et al. 1993;Horizon Environmental Services, Inc . 1994) .
Cores. The four cores are from small cobbles. Three (ST l; TU 5, 40-50 cm and 60-70 cm) are of local quartzites, and have only a few flake removals with large amounts of cortex remaining on the pieces. The other core, of a light gray chert that may have its source in Red River gravels (Banks 1990;Bruseth and Perttula 1981), is bipolar with multiple flake removals and no cortex remaining.
The bipolar technique of core reduction works well in removing flakes from small pieces of raw material, and has been noted in several Early and Middle Caddoan assemblages in Northeast Texas.

Projectile points.
Projectile points, especially arrowpoints, are apparently abundant at the Griffin Mound site, based on the limited investigations conducted to date. Eleven arrowpoints (eight from Unit 5, six of which were in Feature 1) and two dart points (including one in Feature 1) were   Table 1).

The arrowpoints from Griffin
Mound are all rather small in size (range of 12-27 mm in length and 8 -1 7 mm in width) , are on both local (67%) and non-local (33%) raw materials {Ouachita Mountains cherts from Red River gravels), and usually have pronounced resharpened blades with prominent barbs. Typologically identifiable forms include one Alba (ST 4; Figure 3b), one Colbert (Unit 5, 30-40 cm; Figure 3d), and three Catahoula arrowpoints (two from . 70-80 cm in Unit 5 and one from 110-130 cm in Feature 1; Figure  3c,e,g; Turner and Hester 1993). Five arrowpoint fragments (one proximal, one proximal-medial, one distal, two tip fragments) and one complete arrowpoint are not identifiable to type, although the complete specimen (Unit 5, 40-50 cm) resembles the Bonham type ( Figure 3a). Bonham, Catahoula, and Alba arrowpoint types have been found in association with Middle Caddoan ceramics at several components in the Cypress Creek Basin (Thurmond 1990:227 There are two rather wide, thick Gary dart points in the artifact hssemblage. The one from the surf ace (Figure 3 i) is broader and thicker, whereas the point from Unit 5 (90-100 cm in Feature 1) has been extensively resharpened along the blade. Both Gary points are made on local heat-treated quartzites.
The occurrence of Gary points on Northeast Texas sites is usually indicative of Late Archaic and Early Ceramic period occupation. The landform on which the site is found is sufficiently old that Alba, 3b; Colbert , 3d; Catahoula, 3c, e , g ,· unidentified arrowpoints 3f,h; Gary,3i,j. there could well have been a sparse Late Archaic and/or Early Ceramic period occupation preceding the Middle Caddoan settlement at Griffin Mound. Nevertheless, the recovery of a Gary point in the pit fill of Feature 1, apparently filled with surrounding Middle Caddoan midden deposits, suggests that this resharpened Gary may we l l have been reused as a knife or cutting tool by the Caddoan inhabitants of the site; Figure 4. Plain a nd decorated rim sherds: Plain rims,4a,.b,c;incised rime,4d,e,g,h; punctated rims, 4£.
i t is doubtful they manufactured the dart point, however.

CERAMICS
Not incl uding the burned c l ay and daub, 155 ceramic sherds were recovered in the limited investigations at the Griffin Mound s i te .
The sherds are from well-made and well-fired bowls and jars wh ich had been formed by coili ng. Vessels were occasional ly polis hed and/or burnished as a form of surface treatment , and a few appear to be plain . However, in general, it appears that most of the vessels at the site were probably decorated, at least along the rim. One large sherd from a plain bowl was recovered in Feature 1 (Figure 4a). About 14% of the sherds (n=22) have decorations on their exterior 12 surfaces. The s he r ds are relatively small in size, which creates difficulties in discerning the type of decorative element (engraving, incising, etc .) as well as the stylistic mot if (crosshatched lines, pendant triangles, horizontal lines, etc. ) present in the assemblage, but the decorative elements include incising (n=l2, 55% of the decorated sherds ) , engraving (n=9, 41%), and punctating (n=l, 4.5% ) .
Both diagonal and horizontal motifs are ident i f i ed in the incised sherds (Figure 4 ) . One large diagonally incised rim sherd was recovered from ST 4 (Figure 4e), whereas horizontally incised rim and body sherds were found in ST 1 and in Unit 5 (0 -20 cm and 110-130 cm bs) . Other incised sherds without clear stylistic motifs were collected from the surface (n=4 ) ., as we 11 as lower leve 1 s of the midden in Unit 5 . The incised sherds from Griffin Mound probably represent Canton Incised and Davis Incised types (Suhm and Jelks 1962). Vessels in these types are usually large bowls, although Canton Incised jars have been recovered in Caddoan assemblages in Northeast Texas. The rims of the incised vessels are straight with rounded or flat lips and 6-7 mm in thickness.
The single punctated sherd is from ST 4 (Figure 4f). Decorations consist of parallel rows of broad fingernail punctations on the body of the vessel (probably a jar).
The nine engraved sherds are represented by several stylistic motifs: 1) thin horizontal lines on the rim, 2) diagonal lines beginning below the lip, and 3) opposing lines that extend down from the l ip of carinated bowls to the carination po int ( Figure 5). The latter two motifs are common

CADDOAN ARCHEOLOGY NEWSLETTER
grog -bone, 3) bone-grit, 4) groggrit, and 5) grit (Table 2). In general, the ceramics from the site have a sandy paste with grit (small pieces of rock) inclusions, which apparently was added as a temper. The grog-bone-grit temper combination was most frequently chosen (62 . 4% of the sherds), particularly for the plain and incised vessels, followed by bonegrit (13 .4% ) principally among the engraved sherds, grit (12.1%) among plain and decorated sherds, grog-grit (8.7%) for plain vessels, and grog -bone (3 .4%) in the plain and incised vessels.
The thinner sherds are mostly engraved carinated bowls.
There are no clear changes across the site or within the midden in the use of the different ceramic temper combinations (Table  3), as grog-bone-grit is common across the site. While probably related to sample size problems, and the unique character of the Feature 1 archaeological deposits, it is interesting that about 91% of the sherds in Feature 1 are tempered with grog-bone-grit, compared to between 53 and 55% in the midden and plow zone, respectively. Many sherds from Feature 1 seem to be from one large, plain, grog-bone-grit tempered bowl, whereas the midden and plow zone sherds represent smal 1 parts of several different vessels. :4 One particularly notable characteristic of the Griffin Mound ceramics is the very high frequency of bone temper in the assemblage. About 79% of the sherds has some amount of burned bone added as a temper to the paste (Tables 2  and 3) . Among Caddoan sit:es at Lake Fork Reservoir, about 40 km to the west in the upper Sabine Table 3

RADIOCARBON DATING
A sampl e of 54 charred hickory nutshells from 110-130 cm bs in Feature 1 was submitted to Beta Analytic Inc.
for radiocarbon dating analysis .

INTERPRETATIONS AND CONCLUSIONS
The limited archaeological i nvestigations at the Griffin Mound site have demonstrated that it contains we 11 -preserved midden deposits of probable Middle Caddoan age with a large, intact pit feature. The depth and e xtent of the midden, along with the abundance of stone tools and ceramic vessels found at the site, attest to a relatively intensive Caddoan occupation at Griffin Mound, although it is probable the occupation did not last much more than one or two generations (cf. Perttu l a et al. 1986:55). As is seen e lsewhere in Northeast Texas at this time, habitation sites like Griffin Mound seem to represent small hamlets and farmsteads. They have structural remains, 15 features for cooking and storage, and midden deposits, and are common along many of the major and minor streams in the region .
Caddoan sites dating to ca. A.O. 1200-1400 in the Cypress Creek Basin are relatively rare, however, and seem to be concentrated in the "upper reaches of Big Cypress Creek" (Thurmond 1990 :227 -228, Table 63). Thurmond (1990) specul ates that the apparent clustering of Middle Caddoan sites in the basin may represent part of a discrete population group associated with the Keith (41TT11) and Hale (41TT12) mound centers near Mt. Pleasant, about 40 km north of the s i te. Perhaps the people who lived at the Griffin Mound site were part of a similar and contemporaneous Caddoan population which lived in the upper part of Littl e Cypress Creek valley. To furt her e xamine socio-political and settlement patterning questions, it is critical that a concerted program of identification, investigation, and dating of other Middle Caddoan period sites be completed in this part of the Cypress Creek Basin.
Because of likely changes in land control, and the activities of cows and wild hogs, it is doubtful that the Griffin Mound site can be preserved much longer .
We intend to complete additiona l investigations at the site in the near future, particularly by (1) expanding the excavation of the pit feature, (2) locating additional associated features, (3) acquiring more charcoal and nutshell samples in good context to refine the radiocarbon dating of the site, and (4) recovering larger and more representative samples of faunal and floral remains. The information that can be obtained fro1n these data sets will contribute towards a better understanding of Caddoan lifeways in the Little Cypress Creek valley of Northeast Texas .