The Archaeology of the 16th And 17th Century Caddo in the Post Oak Savannah of Northeast Texas: The Tuinier Farm (41HP237), R. A. Watkins (41HP238), and Anglin (41HP240) Sites in the Stoots Creek Basin, Hopkins County, Texas

Cite this Record Perttula, Timothy K.; Dowd, Elsbeth; Green, Lee; Morgan, George; Nelson, Bo; Schniebs, LeeAnna; Schriever, Beau; Todd, Jesse; and Walters, Mark (2009) "The Archaeology of the 16th And 17th Century Caddo in the Post Oak Savannah of Northeast Texas: The Tuinier Farm (41HP237), R. A. Watkins (41HP238), and Anglin (41HP240) Sites in the Stoots Creek Basin, Hopkins County, Texas," Index of Texas Archaeology: Open Access Gray Literature from the Lone Star State: Vol. 2009, Article 32. https://doi.org/ 10.21112/.ita.2009.1.32 ISSN: 2475-9333 Available at: https://scholarworks.sfasu.edu/ita/vol2009/iss1/32


INTRODUCTION
The Tuinier Farm (41HP237), R. A. W<Jtkins ( 41 HP238 ), and Anglin ( 41HP240) sites are W 11 to 17 1 h century Caddo sites in Lhe mod(;rn-day Post Oak Savannah of Northeast Texas (Diggs et al. 2006:Figurc 2). All three of the sites arc located on Stouts Creek, in the eastern part of Hopkins County, Texas, a northward-Bowing Lrihutary to White Oak Creek in the Sulphu r River basin; the modern channel of White Oak Creek lies ca. 15 km north of these sites. The Culpepper site (41HPl ), a previously investigated mid-to late l7 1 h century Caddo hahitation and cemetery site (Scurlock 1962), is about 2 km downstream. Small areas of tall-grass prairie lie to the north between the Stools Creek sites and White Oak Creek, but the eastern extent of the larger While Oak and Sulphur prairies (sec Jordan 1981) is approx i malcly 15 krn to the west and northwest.
Al the time of the Caddo occupation of the Stouts Creek sites, the eli male was welter and warmer than today, with signific<Jnt mesic periods between A.D. 1477-1524, A.D. 1539-1572. and A.D. 1603-1670(Pertlula 2005 Table 2-3). After A.D. 1670, the years from A.D. 1671-1676 were relatively cool and dry. The more mesic periods had more equitable rainfall (adequate growing season rainfall) and this, combined with the warmer temperatures (see Perttula 2005, ed .. : Figure 2-3a), led to an increased net productivity and carrying capacity of plants and animals in the Post Oak Savannah and Pineywoods that were settled by Titus phase populations.
The Tuinier Farm site is the closest of the three sites to the headwatt:rs of Slouts Cn:ek. It is situated on a relatively Jlat and sandy upland ridge

HISTORY OF EXCAVATIONS AND CHARACTER OF THE SITES
The Tuinier Farm, R. A. Watkins, and Anglin sites were located and recorded by Lee Green between [2004][2005][2006][2007] during survey investigations of the Slouts Creek valley around the small community of Pine Forest (see Figure 1). All three arc Late Caddo period, Titus phase, habitation sites with midden deposits, either now in pasture, or in a recently cui-Livaled field in the case of the Tuinier Farm site. Sh'-lfer and Green (2008) report on the excavation of a Woodland period biface cache from a borrow pit area at the southern end of the Tuinier Farm site. Three Late Caddo burials were also encountered in the borrow pit area. At the Tuinier sile in 2007, in addition to obtaining general surface collections in recently plowed fields (Figure 2a The Archaeology uf the 16th And I 7th Century Caddo in the Post Oak Sava11nah ufNortheast Texas 5 40 x 40 ~m unit t:xcavatcd to obtain flotation and fine-screen samples from the South midden. Units I, 3, and 4, and ST I , 2, and 6 were ex~avated in the area of the South midden or Midden I. In the North midden (Midden 2), we excavated ST 3-5 and Unit 2 ( Figure 3). Midden deposits (very dark grayish-brown sandy loam) between 20-34 em in thickness were identified in ST I CSouth midden or Midden I), ST 2 (South midden or Midden l ), ST 4 (North midden or Midden 2), and ST 6 (South midden or Midden 1) at the Tuinier Farm site. ln the South midden, the midden an.:heologil.:al deposits ranged from the surface to 23-25 em hs in Unit 1 and 3 excavations. The North midden deposits extended to a maximum of 30 em bs in Unit 2. A yellowish-brown sandy loam E-horizon underlay both the South and North middens at the Tuinier Farm site.
The R. A. Watkins site is a Late Caddo midden site; the m idden mound is about 15 m in diameter. We conductcu no excavations here but studied a small surface collectio n of artifacts (n=20 l) with ceramic sherds, burned clay, daub, clay objects, and a few pieces of lithic debris.
Prior to our work at the Anglin site in Fehruary 2007, an area ca. 11 x ll m in size had been excavated over the past several years by Lee Green and associates in and around a well-preserved midden deposit about 10m in diameter on a sandy knoll ( Figure 4 ). These excavations were done in various sized units, sometimes with excavations by levels, but for our purposes here, the collections from those excavations are treated as a single provenience unit since they come from a small and discrete mi d den deposit. The 2007 arch aeological work focused on identifying remaining uncxcavatcd and undisturbed midden deposits at the site and on a smaller knoll ahout 25 m to the north. For this, we excavated two shovel tests (ST B and C) on the small northern knoll and ST 1-2 and Units l-3 (I x 1 m in size) along the northern, southern, and western margins of the midden deposits lFigure 4 ). Unit I was excavated to 20 em bs in I 0 ~m levels, but was terminated when no midden deposits were encountered . The situation was the sam~: in Unit 2, except it was excavated to 25 em bs in three arbitrary levels. Unit 3 did t:ncountn undisturbed midden deposits in the southern half of the unit from 0-35 em bs. These midden deposits arc a very dark brown ( 1 OYR 2/2) sandy loam, and lhey rest on a yellowish -hrown (I OYR 4/6) sandy loam E-horizon.

RADIOCARBON DATES
Two radiocarbon dates have been obtained from the Tuinier Farm site. The samples submitted for radiocarbon analysis arc charred Hickory (Cary a sp.) nutshells from Unit 4 flotation samples (10-20 ~m and 20-30 ern hs) in the South midden.
The calibrated intercepts suggest that the Caddo occupation at the Tuinier Farm (or at least that part of the South miuden occupation in the vicinity of the Unit 4 archaeological deposits) may have begun as early as the mid-15th c..:ntury A.D. and lasted until the mid-17th century A.D. At 2 sigma (95% probability), the two calibrated radiocarbon dntes overlap between AD 1520-1630 (Table 1), and this is considered the most likely chronological range of the domestic Caddo occupation at Tuinier Farm; the burials at this site may be younger than that based on the presence or a mid-17th century style Taylor Engraved inverted rim carinated bowl (sec below). The R. A. Watkins and Anglin sites appear to be contemporaneous with the Tuinier Farm Caddo occupation, based upon an examination of the range and styles of the decorated ceramic shcrds found at each site (see he low).
Four sherds from th..: Tuinier Farm site are to he submitted for thennoluminescencc (TL) dating, but the results are not expected to be in hand until mid-2009(Dr. James Feathers, September 2008. The TL dating of Caduo sherds is in its infancy, hut good results (i.e., the TL dates arc comparable to the calibrated ages received through radiocarbon dating or ~harred plant remains from the same archaeological deposits) have recently been obtained from the Lang Pasture site (4lAN38) in the upper Neches River has in in East Texas (Feathers 200R;Perltula 2008).

MATERIAL CULTURE REMAINS
The prehistoric and historic 1 material culture remains analyzed at the three sites (not including bone and shell artifacts uiscusscd below), 6766 artifacts in total, is a product of the prior work (excavations and surface collections) by Lee Green and associates comhined with the limited shovel testing and hand-controlled excavations at the Tuinier Farm and Anglin sites. Material remains are the three sites are abundant, particularly ceramic vessel sherds (Tahlc 2), as these account for at least 80% of all the analyzed artifacts from the Stouts    Reimer et al. (2004) and IntCal04. * a large sample of lithic debris was found at Anglin in the earlier excavations, but they have not been counted or analyLt:d since they come from unprovcnic:nced contexts within the midden there; **includes a perforated sheru (spindle whorl); ***cut naHs; +==glazed brick fragments; ++==docs not include the lithic debris from the earlier investigations Creek sites (83% at the Tuinicr Farm site, 91% at the R. A. Watkins site, and 81% at the Anglin site). Burned clay and daub is well r~presentL:d at the Anglin site, as arc day objects and ear spools. Elbow pipe sherds arc present at both the Tuinier Farm and Anglin sites ( Table 2).
Based on the limited amounls of lithic debris found in the 2007 investigntions, the knapping of stone tools was not an important activity at tbc Tuinier Farm site during Late Caddo times, as is often the case al other Titus phase sites (Perttula 1998: ~0), but may have been a more common task during the Late Caddo occupation at the Anglin site. This is not properly rellected in the small sample of lithic artifacts studied fur this analysis since at least 2000 pieces of lithic debris have heen previously colkcted from the Anglin site during earlier unprovenieneed excavations in the midden deposits here. The relative ahundance of chipped stone arrow points, and the residue of chipped stone tool manufacture, suggests that the Caddo peoples living here were taking and processing large amounts of hunted resources, perhaps even engaging in long-distance hunting in the nearby tall grass prairies to tht: west as well as focusing on game animals that favored the forested woodlands. The Anglin hunters may have taken advantage of the accessibility (compared to the Pineywoods Caddo) of good hunting areas in the upper Sulphur River basin and the possibility thatdue at least in part to changing and drier climatic conditions-small herds of very large game animals would have been available for procurement (see discussion in Perttula and Shennan 2008:303-304 ).

Ceramic Vessels
A total of 15 vessels and partial vessels have been recovered hy Let: Green and associates from three Caddo hurials discovered in the disturbed borrow pit area at the Tuinier Farm site; no information is available on which vessels were round together in the three hurials. These vessels include: two Taylor Engraved carinated bowls (5.2 and 8 em in orifice diameter) with direct rims; a late (mid-17'h century) style inverted rim Taylor Engraved carinated bowl (ef. Pemula 2007) with red pigment rubbed in the engraved lines (21 em in orifke diameter) (figure Sa); two Simms Engraved carinated bowls ( 11.2 em and I 2.3 em in orifice diamett:r) ( A large section of an cverted rim LaRue Neck Banded jar had also been found in Midden 1 or the South midden at the Tuinier Farm site. This grogtempered jar has 10 rows of neck banding with appliqued chevrons and slash punctates on the vessel body.

Odell Site (41HP239)
The Odell site is a contemporaneous Late Caddo site on Stouts Creek, located a few miles upstream from the Tuinier Farm site. Several whole vessels were documented from the site (presumably the grave goods from a single hurial) during the course of our investigation of the Stouts Creek Caddo sites. These include a fragmentary LaRue Neck Banded everted rim jar with four small strap handles ( Figure  6a), a Ripley Engraved carinated howl (23 em in orifice diameter) with a scroll and diamond motif repeated four times on the rim panel (Figure 6h), and a large grog and shell-tempered Taylor Engraved olla with a slight spool neck (Figure 6c).

Ceramic Shcrds
There are about 5530 ceramic vessel sherds in total from the Tuinicr Farm (n= 743 ), R. A. Watkins (n=l83), and Anglin (n=4606) sites (see Table 2). Between the three sites, the plain sherds (rims, body, and hase) comprise approximatdy 70% of the ceramic sherds (n=3860). There are 1679 decorated rim and body shcrds in the collections, 81% from the Anglin si tc. The plain to decorated sherd ratios (P/DR) at the three sites range from 1.62 (Tuinicr Farm) to 3.36 (R. A. Watkins), with a P/DR of 2.42 at the Anglin silt:. As these ratios suggest, plain ware vessels and/or vessels with   (Table 3). The proportions of utility ware and fine ware rims are quite consistent among the three sites, suggesting that the ceramic sherd assemblages from them provide a reasonably robust sample of the character of the domestic Late Caddo ceramics in this localit y. The decorated ceramic sherds from the Stouts Creek sites are dominated by engraved and redslipped fine wares and neck handed and appli4ued utility wares (Table 4). The number of rims of each decorated ware suggest that fine wares are at least twice as common as utility wares in these domestic assemblages. Among all the sherds, many of the fme wares, especially at the Anglin site, apparently have a hematite-rich red slip on both interior and exterior vessel surfaces (Table 4). However, the absence of red-slipped rim sherds in the Stouts Creek sites indicates that, unlike a numher of Titus phase assemblages in the Big Cypress Crct:k basin (Perttula 2005;Nelson and Perttula 2003), there are no plain red-slipped vessels in the former sites, only engraved vessels (usually carinated bowls, but also bottles) that occasionally have red-slipped surfaces. Other fine wares include a few trailed sherds and hurnishcd and/or red-slipped vessel rim shcrds with diagonal lip notching.
The decorated utility ware sherds from the Stouts Creek sites can be readily dividcd into five broad classes: appliqued, neck banded, corncob impressed, brushed, and incised/punctated (see Tahle 4). The appli4ued sherds arc primarily from McKinney Plain and Harleton Applique<.! jars while the neck banded shcrds an: from LaRue Neck Banded vessels. These two classes of utility ware pottery together comprise between 51-74"k of all the utility wares at the three Stouts Creek sites.
BrushcJ, corncob impressed (Anglin Impressed, a newly defined Caddo pottery type), anJ incised/ punctated poltery arc decidedly seconJary Jecorated utility wares, nowhere accounting for more than 20% of the utility wares at any one site. Brushed pottery comprises between 8.5% (Anglin) and 20% (Tuinier Farm) of the utility wares. Sherds with either punctaled, incised. or incised-punctatcd decorations account for only ll-17c>hl of the utility wares at the Stouts Creek sites.
The corncob impressed sherds are present only aL the Tuinier Farm and Anglin sites (7.5-15% of the utility wares). Corncob impressed pottery had been previously identified only from the Spoonbill site (41WDI09) in the Lake Fork Creek basin (Bruseth and Perttula 1981 : Table 5-8 and 82), where it was dubbed ··corn Cobb Incised." The temporal and cu It ural connotations of this class of pottery at the Spoonbill site were not explored in Bruseth and Perttula (I 981 ), hut its recovery at the Stouts Creek sites in 16th and 17th century contexts, and at Spoonbill where materia l of simi-Jar age is known, is consistent with the fact that there is a late Titus phase occupation at Spoonbill (Walters 2007).

Tuinier Farm
Engraved and red slipped fine ware vessel sherds account for 60.8% or all the decorated sherds at the Tuinicr Farm site (see Table 4). Other fine wares include a lip notched rim and a single body shcrd wirh a curvilinear trailed line (Keno Trailed?, see Figure lOa, below).
About 91 % of the engraved fine ware shcrds where typological identifications are possible are confidently classified as being from Ripley Engraved vessels (Tahle 5), mostly carinated bowls, based on tht: kinds of engraved motifs found on the rim panel of vessels (see Thurmond 1990:Figurc 6). There is also a smattering of Taylor Engraved and probable Hodges and Womack Engraved types in the Tuinier Farm fine ware sherds. There is one shell-tempered Avery Engraved vessel sherd from a trade vessel that likely was manufactured on a McCurtain phase Caddo site along the Red River, well to the north of the Stouts Creek area. Taken together, the co-association of these engraved fine ware types suggests that the Caddo occupation at the Tuinier rarm site postdates ca. A.D. 1550, and certainly lasteJ into the 17 1 h century A.D. The occupation could have lastl.!d as late as the miJ-to late 17 1 h century given the known chronological age range of Titus phase sites (see Perttula 2005:364-370). The same range of fine ware types has been recovered in the vessels placed as funerary objects in the Culpepper site cemetery (Scurlock 1962:Tablc 1 ).
Seven different Ripley Engrave<.! carinated howl rim motifs arc represented in the Tuinier Farm rim and hody sherds, with equal numbers of the pendant triangle (n=5 ). scroll (n=5 ), and interlocking horizontal scroll (n=6) motifs 8a,. Less common rim panel motifs incluJe the scroll and semi-cin: le. the continuous scroll, the nested triangle, and the scroll and circle motif.
The pendant triang le motif (see Figures 7c and 8e) is particularly chronologically sensitive, ns it is a distinctive stylistic element signifying post-A.D. 1600 Titus phase occupations (see Perttula ct at. I 998) in the Big C ypress Creek basin; Maud and Taka points, especially the latter, typically occur on sites with Ripley Engraved vessels having the pendant triangle motif. The scroll motifand the many scroll clement sherds (scroll lines and hourglass-shuped scroll filler clements seen on several distinct and different rim motifs, sec Thurmond l990: Figure 6a- The remainder of the engraved sherds have simple geometric elements or straight line designs (although both of these elements may be from more complex but unidentifiable scroll motifs). These include: horizontal lines (n=ll, including seven rims; may be from interlocking horizontal saoll motifs, but no scroll elements identifiable on speci fie sherds); parallel lines (n= 1 0); opposed lines (n=2); horizonlal and diagonal lines (n=l rim sherd); horizontal and vertical Iim:s (n=l rim); a hatched zone (n=l); small excised triangles (n=l rim); panel dividers (n=2); and one body sherd with both circular and rectangular elements (see  Almost 5% of the engraved sherds from the Tuinier Farm site also have a red-slipped surface. This includes sherds from Ripley Engraved (n=6), Taylor Engraved (n= l ), and shell-tempered Avery Engraved (n=l) vessels. Another 6.1% (n=10) of the engraved fine wares have had a pigment rubbed in the engraved design. The vast majority of these sherds have a hematite-rich clay pigment (n:::::9), but one has a white kaolin clay pigment.
The red-slipped body shcrds include five from carinated bowls with a slip on both interior and exterior surfaces and two from bottles that have only an exterior red slip.
Fine ware rim shcrds (n=39) at the Tuinicr Farm site arc almost exclusively direct or vertical in profile (94.8%) and with rounded, exterior folded (4lU%) or rounded (33.3'1o) lips. There is one inverted rim fine ware sherd as well as one with an everted rim profile. Other distinctive lip forms noted in the fme wares include one with an exterior thickened lip and two other sherds with a flat, exterior folded lip. In toto, exterior folded lips comprise 53.8% of the fine ware rims, compared to only 13.8% of the plain ware rims and 21.4% of the utility ware rim sherds.
The utility ware sherds from the Tuinier Farm site arc from jars that were likely used for cooking and storage tasks Juring the Caddo occupation there. As previously mentioned, utility ware vessels decorated with appliqued or neck banded elements arc most prevalent (sec Tahle 4).
The appli4ued sherds from McKinney Plain vessels include one lower rim sherd with curvilinear appliqucd strips forming a lug handle (Figure 11 a), large nodes (n=3), straight appliqued ridges-up to three closely-spaced parallel ridges (n= 19, Figure  11 c, f)-that apparently extend from the lower rim vertically down the vessel body, single to multiple curvilinear appliqued ridges nn the vessel body (n=6, Figure II d), and appliqucd fillets (n=3). There are also Lwo sherds of Harleton Appliqucd with ap-pli4ued chevrons (applied beginning immediately below the rim-body junction and extending in some cases well down the vessel body, see Suhm and Jelks 1962:Plate 33d, f-g) and two others wiLh clusters of small appli4ued nodes ( Figure II a, e).
The LaRue Neck 13anded sherds, rims and lower rim (tabulaLed with the body sherds), have hroad horiz.ontal neck handed or corrugated coils that    There is one distinctive incised rim sherd from the Tuinicr Farm site, from the borrow pit area, that is not from the 16 1 h and 17 111 century Caddo occupation. This is a grog-tempered Coles Creek Incised rim with a single lip line. Not enough of the rim remains to determine the placement or execution of horizontal incised lines on the rim itsdf, and there are a number of varieties of Coles Creek Incised that have lip lines (Phillips 1970). Those varieties that may have only a single lip line include t•m: Stoner, var : Phillips, and vw: Campbellsville (Brown 1998:8). According to Brown (1998:52-53) (Figure 15b-c, c), and there are six everted rims (Figure 15d) from plain jars. Lip forms arc very commonly rounded (n=21, 72%), likely from bowls and jars, fiat (n=4, 13.8% ), and rounded and exterior folded (n=4, 13.8% ); these latter rims ( Figure 15a) arc likely from plain carinated howls. The ceramic vessel sherds from the Tuinier Farm site are tempered almost exclusively with grog, either as the sole temper, or in small amounts in combination with hematite, bone, or charred organic materials ( Table 6). Less than 2% of the sherds have a shell temper, and these are from Red River McCurtain phase trade vessels. Between 7-17% of the sherds by ware have a naturally sandy clay paste, with the highest proportions among the utility wares and the plain ware sherds. The fine wares arc more commonly tempered with bone or hematite than either the utility wares or plain wares (see Table 6).   Depending upon the ware, hetween 59.6% (plain wares) anc.J 87.1% (tine wares) of the ceramic sherds from Tuinicr Farm are from vessels that had been tired in a low oxygen or reducing environment (Table 7). Tht:-vast majority of thest:-vessels were pulled from the fire to cool in the open air, leaving them with one or both surfaces of the vessels with a lighter (usually a chocolate hrown color, at least in the case of the fine wares) color. Those that were left to cool down in a low oxygen environment turned a gray to hlack color. Only 13% of the fine wares were fired, or at least were partially tlred, in an oxidizing environment. Much higher proportions of the utility wares (31.5%) and plain wares (40.4%) were fired in an oxidizing environment or fired under less wellcontrolled firing conditions (see Table 7).

R. A. Watkins
There are a total of 183 sherds in the collection from the R. A. Watkins site (see Table 2). This includes six plain rims, 130 plain body sherds, one drilled body sherd (possible spindle whorl piece) with a I I .6 mm perforation, four base sherds, and 42 decorated sherds. The plain to decorated sherd ratio is 3.36. Plain vessels arc apparently a common constituent in the vessel assemhlage in usc at the site given the recovery of six plain rims compared to I 0 decorated rims (three from utility wares and seven from engraved fine wares), accounting for 38% of the rims in the collection. The plain rims are uniformly direct or vertical in profile, with rounded (n=2), rounded-exterior folded (n=3), or rounded-interior hcveled (n=l) lip forms. These rims are probably from undecorated bowls and carinated bowls.
Half of the decorated sherds are from fine ware vessels (55clo, n=23), including engraved (n=20) and red-slipped (n=3) sherds. The engraved sherds appear to he from at least nine different vessels, seven carinated bowls of the Ripley Engraved type (Figure 16c-e), one Hodges Engraved bottle, and a Taylor Engraved carinated bowl ( Figure l6a). Four or the Ripley Engraved vessels recognized in the sherds have had a red pigment rubbed into the engraved motif. Three of the sherds also have an interior/exterior red-slip ( Figure I6b). Rim forms are primarily direct in profile, but one has an everted rim (Figure l6e), with rounded (n=3) and roundedexterior folded (n=4) lips. The principal decorative motifs on the Ripley Engraved vessels include scrolls, either from continuous scroll or scroll motifs (see Thurmond I Y90: Figure 6). These have vertical and hour glassshaped scroll dividers defined primarily through either excision or cross-hatched engraving as well as vertical engraved lines (see Figure 16b-c, f). Two rims have sets of horizontal engraved lines (see Figure 16d-e), and these either arc used to delimit the engraved rim panel motif or may be from Ripley Engraved compound bowls with an upper panel with horizontal engraved lines and a lower panel with a more complicated engraved motif; the sherds are not large enough from the R. A. Watkins site to determine this.
The possible Hodges Engraved bottle sherd has a curvilinear engraved line from a scroll element with a series of tick marks on the line. The Taylor Engraved vessel has a graceful series of intersecting ~.:oncentric engraved lines (sec Figure 16a). The red-slipped sherds include one with only an exterior slip and two othcrs-hoth-shell-tcmperedwith interior and exterior red-slipped surfaces. These latter an~ probably from the undecorated portion of a shell-tempered Avery Engraved ur Taylor Engraved vessels or from a plain red-slipped shell-tempered Clement Rcdware vessel (ef. Flynn 1976).
The other decorated sherds (n= 19) are from utility ware jars: neck banded (n=9, including Lwu rims); appliqued (n=5); brushed (n=l); brushed-incised (n=2); and punctated (n=2). The neck banded sherds arc from at lc.ast two different LaRue Neck Banded jars, one with a direct rim and a flat lip (Figure 17c-d) and the other with an everted rim and a rounded lip. The five appliqucd sherds are from McKinney Plain jars with nodes placed around the rim hut under the lip (Figure 17a), as well as narrow appliqued ridges and fillets that run vertically on the rim and un the vessel body (Suhm and Jelks 1962:Piate 49e, h).
The less common utility wares include une brushed sherd from the hody of a jar, and two brushed-incised sherds: one of these has parallel brushing and incised lines, while the other is a rim with horizontal brushing and a diagonal incised body (see Figure 17h); this rim also has a crimped and notched lip. Tht: two punctated sherds have either tool or fingernail pum:tated rows.
The ceramic sherds from the. R. A. Watkins site are tempered primarily with grog or crushed sherds (Table 8), including both the plain wares and the decorated sherds. Decidedly minor tempers used by Caddo potters include crushed and burned bone, hematite, charred organic materials, and crushed and burned mussel shell; the latter arc from red-slipped Avery Engraved vessels made hy McCurtain phase Caddo groups on the Red River in northeastern Texas (see Peruula. ed. 2008).
Equivalent amounts of a naturally sandy clay were used by Caddo potters fur the manufacture of plain and decorated vessels at the R. A. Watkins site: 17 .2-18J~% of the shcrds examined in deLail (sec Table 8). In general, the more heterogeneous temper-paste combinations are characteristic uf the plain ware sherds.
The ceramic vessels at lhe R. A. Watkins site were fired under a diverse set of firing conditions (cf. Teltscr l993:Figurc 2; Perttula, cd. 2005). Most were fired under a low oxygen or reducing environment (51.7% of the plain sherds and 6~.8%  or the decorated sherds), especially deriving from vessels that were subsequently removed from the fire and allowed to cool in the open air (Table 9). Less well-controlled firing (i.e., incompletely oxidized or sooted/ smudged/reheated firing conditions) was apparently more prevalent among the plain wares than among the decorated sherds analyzed in detail.

Anglin
About 30% of the 4606 sherds from the Anglin site are decorated, including 74% or the rim sherds (see Tables 2 and 3). As with the other Stouts Creek sites, the sherds from the Anglin site are primarily from fine wares (especially Ripley Engraved), as well as McKinney Plain and LaRue Neck Banded vessels, with some brushed and Anglin Impressed jar sherds. With the larger sample size of decorated shcrds-both nne wares and utility wares-there are several different classes of shcrds found only at Anglin that haw distinctive decorative elements and methods of decoration (sec Table 4).
The tine wares at the Anglin site total800 sherds, including 1 RO rims, primarily if not principally from engraved carinated bowls of scvt:ral different sizes, along with a few shcrds from com pound bowls and boule sherds. Engraved sherds comprise 75%  Tables 3 and 4).
As with the Tuinier Farm and R. A. Watkins sites, Ripley Engraved is the primary engraved fine ware type at the Anglin site. Almost 89% or the engraved carinated bowl, howl, and compound bowl sherds from the site that can be identified to a defined type are from Ripley Engraved vessels, including 91% of the rim sherds (Table 10). Simms Engraved is a far distant second (5% ), followed by a Womack Engraved variant (2.3%), Hodges Engraved (1.9% ), Taylor Engraved ( 1.2% ), and Avery Engraved (0.8%). With the exception of the absence of Simms Engraved shcrds at the Tuinicr Farm site, the proportions of the key engraved types arc virtually identical to that seen in the fine ware sherd assemblage from the Anglin site: Ripley Engraved (91% ), Womack Engraved variant (2.6% ), Avery Engraved ( 1.3°/r') , and Taylor Engraved (3.9%) (sec Table 5).
A variety of Ripley Engraved carinated bowl, howl, and compound bowl rim motifs (see Thurmond 1990: Figure 6) have been identified in the fine ware sherds from the Anglin site (see Table  10). The prin~.:ipal motifs include the interlocking  Table 5). The interlocking horizontal s~:roll is not a common Ripley Engraved rim motif in Thurmond's (1990) compilation for Titus phase sites in the Big Cypress Creek basin in Northeast Texas, heing found    usually only in low amounts (1-7% or the whole vessels) in Titus phase sites in the Big Cypress Creek basin and in western Titus phase cemeteries in the upper Sabine River basin (Perttula et al. 1993). At the P ilgrim's Pride site, vessels with interlocking horizontal scroll motifs comprised 16% of the whole vessels (Pcrttula 2005, ed.:272). The predominance of the interlocking horizontal scroll motif at the Anglin site (see Fi g ures 18d,19a,c,and 21c), as well as in the Tuinicr Farm sherds (30% or the Ripley Engraved sherds with an identifiable motif) and the Culpepper site vessels (3 I .6%, sec Scurlo~:k 1962:294 and Figure 6d), clearly set the Stouts Creek sites apart from all other well-documented Titus phase vessel assemblages.
At Anglin, 13% of the sherds with identifiable Ripley Engraved motifs have a continuous scroll (see Fi gure 20b); 21% of the Culpepper vessels have a continuous scroll rim motif (Scurlock 1962: Figure  6c). Other Titus phase cemeteries where vessels with the continuous scroll motif are relatively abundant includes sites in the upper or western reaches of the Big Cypress Creek basin, particularly at the Tuck Carpenter site (41 CP5, 40%) and Mattie Gandy (41FK5, 29%) ( Pcrllula 2005, ed.:272). This suggests some level of contact and interaction between the Caddo peoples living in the Stouts Creek and western parts of the Big Cypress Creek drainage during the time of the occupation at the Anglin site. Perttula's ( 1992:tablc A.2) analysis of Ripley Engraved motifs suggests this interaction may have taken place during the earlier part of the Titus phase occupation at the Stouts Creek sites, perhaps in the middle part or the 16th century.
The scroll motif is present in considerable numbers on vessels in Titus phase sites throughout the Big Cypress and upper Sabine river basins, as well as in sites in parts of the Sulphur River basin, from early to late Titus phase contexts (sec Perltula et al. 1998;Pertlula 2005, cd:272, 274;Thurmond 1990). At the Culpepper site, occupicu during the latter part of the 17th century, vessels with the scroll motif account for 31.6% of the Ripley Engraved vessels ( Scurlo~.:k 1962:Figurc 6b ). At Tuinier Farm and Anglin, sherds with the continuous scroll motif represent only 5-13% of the identifiable Ripley Engraved sherds (see F igure 18a-c).
A bit more than 9% of the Ripley Engraved sherds at the Anglin site with an identifiable motif have the nested triangle motif (see Thurmond 1990:Figurc 6h). As with the interlocking horiwntal scroll and continuous scroll motifs, Titus phase cemeteries with Ripley Engraved vessels having the nested triangle motif arc more abundant in western Titus phase sites in the western reaches of the Big Cypress Creek hasin (Pcrttula 2005, ed.:274-275;Perltula and Sherman 2008:Figure 9-27). This includes the A.P. Williams (41TT4, 15.1%), Pilgrim's Priue (41CP304, 10%), and Mockingbird ( 41 TT550, 9.4%) sites In addition to the many Ripley Engraved vessel sherds from carinated bowls, bowls, or compound bowls, a small percentage of the engraved fine wares are from other types, including Taylor Engraved (n=3), Hodges Engraved (n=4), Simms Engraved (n=13, all from hubcap-shaped carinated howls), Avery Engraved (n=2), and a variant of Womack Engraved (n=6) (see Table 10 and Figure 2lb). These arc all post-A.D. 1500-1550 fine wares in the southern Caddo area, as was discussed above with respect to the fine wares from the Tuinier Farm site. The hubcap form of Simms Engraved ( Figure  22a-d), including several that are lip notched, was made during the latter part of the McCurtain phase (ca. A.D. 1500-1700) (Pcrttula 1992:Tablc 11); none of the Simms Engraved shcrds from Anglin are shell-tempered, and thus it is likely that they were not from Red River contexts, but from a more local production locale. The two Simms Engraved vessels (including one hubcap-style form) from the Culpepper site (Scurlock 1962:296, 298) arc also not shell-tempered.
The possihle Womack Engraved sherds from the Anglin site indude two with inverted rims and four body sherds (see Table 10). These shcrds have opposed and offset rows of either hatched or excised pendant triangles, with the upper row of triangles pointing downward and the lower row pointing upwards; the apexes of the triangles do not match. Except for the fact that the pendant triangles are hatched and excised, rather than cross-hatched, these sherds closely resemble Design A of Womack Engraved (Duffield and Jelks 1961 :Figure I 0;Story ct al. 1967:Figurc49).
None of the cf. Womack Engraved shcrds from the Anglin site are shell-tempered; only 3.3% of all the sherds from the site have shell temper (sec below). At the nearby <lnu contemporaneous Culpepper site, a recent examination of the vessels at the Texas Archeological Research Laboratory indicates that only 6.1 o/o of the vessels are shell-tempered. Given the increased use of shell-tempering in Womack Engraved vessels in later 18th century contexts (Perttula 2007:137, 142), and an increased use of shell tempering in general in the manufacture of ceramic vessels, this suggests this design variant of Womack Engraved dates from the latter part (ca. A.D. 16 70) of the 17th century, ncar the end of the Titus phase. About g.s% of the Womack Engraved sherds from the early 18th century Womack site have shell temper. Later sites with Womack Engraved vessels and sherds have more shell-tempering in the sherd assemblages as a whole: 24% at the Pearson site on the upper Sabine (mid-late ]gth century) and 56% at the Gilbert site, thought to date from ca. A.D. 1730-1770. At Gilbert, more than 70% of the Womack Engraved sherds have shell tempering (Story ct a!. l967:Tablc 7).
There arc 313 other engraved sherds at the Anglin site, mostly smaller pieces, that have simple straight, geometric, or curvilinear clements that cannot be associated with larger decorntive clements or di~tinctive rim panel motifs (Table 11). I suspect that almost all of these sherds are from Ripley Engraved carinated bowl , based on a consideration of the more obvious dt:corative elements and motifs recognized in the larger body and rim sherds listed in Table 10. The shell-tempered and red-slipped shcrds with single straight, parallel, or curvilinear engraved lines (see Table ll) are probably from Avery Engraved vessels.
Engraved bottle sherds (n=22) are not at all common at the Anglin site, accounting for only 3.7% of the engraved fine wares (see Figure 21 a). At the Tuinier Farm site (see above), almost 10% of the engraved line ware shcrds are from bo!lles.
The most common decorative elements idenlificd on the bottle sherds include curvilinear and concentric lines (n=6; two of these are from shelltempered vessels) ami curvilinear lines along one side of an excised area (n=4). One bottle neck has horizontal lines on it, and two others have simple straight or curvilinear opposed engraved lines, but the remainder include the following elements: curvilinear lines and zigzag lines (n= 1); circles (n=l ); excised scroll divider/filler (n= 1 ); excised negative oval (n= 1 ); curvilinear lines and excised triangles that are part of a scroll motif (n;;;:l ); scroll elements (n=1); cross inside a circle (n=1, sec Figure 19b); and an oval with an attached excised triangle.
These engraved bottle clements, except for the excised negative oval and the two shell-tempered bottle sherds, would not he out of place on a Ripley Engraved boule. The shell-tempered bottle sherds are likely from Avery Engraved vessels (and one of them is red-slipped), while the excised negative  Suhm and Jelks 1962:Plates 37 and 38). Finally, there is a large Hodges Engraved bottle sherd from the Anglin silt: that has wide cross-hatched zones and triangular elements (see Figure 2lb). Two of the buLL!e sherds (9.1 %) from the Anglin site have a red slip on their exterior surface, and one has a red pigment ruhhed in the engraved motif.
Appro.'l.imately 10.5% of the engraved sherds from the Anglin site also have a red-slipped surface (ahout 90tfl: uf these have both the interior and exterior surfaces coven:d with a red slip); red-slipped engraved vessel sherds are twice as common here when compared to the Tuinier Farm assemblage. This includes sherds from Ripley Engraved, Taylor Engraved, and shell-tempered Avery Engr<Jved vessels; 14.~% of the rtd-slipped engraved sherds arc shell-tempered. Fine ware sherds with pigments ruhbed in the engraved designs arc also more common al Anglin ( l0.4C;-;,, n=62) then at Tuinier Farm (6. Jl';h). Must (67.7%; 90% at Tuinier Farm) of these sherds have a hematite-rich clay pigment (n=42), but 32.3% of the Anglin fine wares with a pigment have a white kaolin clay pigment compared to unly I 0% of the pigment-covered engraved shcn.ls at the Tuinier Farm.
The red-slipped fine ware sherds (n=198), all hody shcrds from rim decorated fine wares, almost always (91.4%) have both surfaces covered with a slip. Another 7.6% hnve only an exterior red slip (and are probably from bottles) and 1% have only an interior red slip. Approximately 6.6% of the redslipped sherds are from shell-tempered trade vessels, primarily Avery Engraved vessels.
Other fine wares frum the Anglin site include om: shell-tempered Keno Trailed budy sherd (probably from a bowl) and two lip notched and burnished rim shcrds. One of these also has a red slip on hoth shcrd surfaces.
Fine ware rim forms arc almost exclusively direct or vertical in profile (97.3%), with rounded (44.1 %) and rounded, exterior foldcll (45.3%) lips. There are four inverted rim engraved vessels in the Anglin site ceramic assemblage.
The Anglin site neck banded pottery from LaRue Neck Banded vessels includes 4~ rims and 86 body sherds. These sherds he1ve broad and crimped horizontal coils or neck bands encircling the rim that were not smoothed over (Figure 23a-h). Suhm and Jelks ( 1962:93) indicate that there may be as many as four to eight neck handed coils at the vessel rim.
One of the neck banded body sherds appears to also have corncob impressions, and four others have roughened rim and body areas (see F igure 23a). A single body sherd with pinching e1ppears to represent a decorative element that simulates the use of neck banding, but without the crimping of coils.
Several of the neck banded sherds also have appliqucd elements. This includes body sherds with neck banding above an appliqued ridge, appli4ued lug handles and neck banding (Figure 24c), and appli4ued nodes amidst neck handed coils ( Figure  24a). Finally, a rim has a row of tool punctates under the lip and above the neck banded coils, and an appliqued node is set amidst the neck banding.
The appliqued shcrds from the Anglin site are dominated by narrow and straight ridges of clay applied to vessel bodies (Table 12 and Figure 2Sa-c, e), single nodes, or sherds with an appli4ued ridge and node. Most of these are from McKinney Plain vessels, where the appliqued ridge served to quadrate the vessel body (Suhm and Jelks 1962:Piate 49). Other appliqued elements that may mark McKinney Plain vessels are straight appliqucd Jillets ( Figure  2Sd and Figure 26a, c).
Curvilinear appliqued ridges, parallel ridges, ridges with dusters of small nodes, parallel fillets, chevrons, and curvilinear lug handles on body and rim shcrds are decorative clements (see Figure 26b, e; see also Figure 2Sf) associated with the more complicated Harleton Appli4ued designs seen on Titus phase jars. These comprise about 22% of the appliqued sherds from the Anglin site. The node clusters and row of small nodes may also belong    with this group of poLLt:ry, rather than with the McKinney Plain vessel sherds.
There are seven appliqued-punctatcd sht:rds, including four rims (see Figure 26d and Figure 27c). The rims may he from Mockingbird Punctatt:d vessels as they have aL kast one row of tool punctatcs on the rim, as well as a single appliqued node; in half the sherds, the appliqued node was placed above the punctated rows, just under the lip. The thret: body sherds include ont: with an appliqued ridge next to a row of punctations; another with a row of linear punctatcs alongside an appliqucd node; and the third body shcrd has an <lppliqued fillet alongside a row of tlngernail punctates.
Three body sherds at the Anglin site have appliqut:d and brushed decorative elcmt:nts. Two have a single straight appliqued ridge and an adjacent area with parallel brushing. The third sht:rd also has a single straight appliqued ridgt:, hut with opposed brushing marks on eitht:r side of the ridgt:.
Sherds with tool punctations account for almost 60% of the punctated rim sherds from the Anglin site, as well as 77% of the body sherds (Table 13). Other punctated elements represented on sherds have heen exccutt:d with either fingernails, a small circular tool (not a cane), or other forms of instrument punclations.
The tool punctales, with one exception, include at least one horizontal row of punctmes encircling the vcsst:l rim (Figure 27a-d). One rim has horizontal and vertical opposed rows of vt:ry small tool punctates. The other rims have similar horizontal rows of punctations. Six sherds with shallow and diagonal stab and drag punctates (Figurt: 27f-g) may bt: from the lowt:r part of the rim of certain McKinney Plain vessels (sec Suhm and Jelks 1962:Piate 49j).
Tht:re are 40 Anglin Impressed sherds in the Anglin site ceramic assemblage (Figure 28a-e), including nine rims. These sherds have horizontal rows of impressions made by rolling a corn cob over the wet paste surface of a utility ware jar. One of the Anglin Impressed rims also has an appliqucd node under the vessel lip.
Vessels with brushing decorative clements (including those with brushed-incised and hrushedpunctated elements) are not common at the Anglin site, comprising only 8% of the utility wares. By contrast, at the Tuinier Farm site, 20% of the utility wares have brushing, and 15.8% of the R. A. Watkins utility wart: ceramics arc brushed.
The Anglin site brushed sherds arc both rim (n=4) and body shcrds. Three of the rims have horizontal brushing marks, while the fourth is horizontally brushed, but with rows of tool punctations pushed through the brushing. The latter decorative element is known on Pease Brushed-lncist:d vessels (Suhm and Jelks 1962: 119), which do occur in Titus phase sites in both mortuary and domestic contexts (Pt:rttu Ia 2005:Tables 11-10 and 11-11). Body sherds have parallel brushing ( n=25), parallel brushed-incised (n=7 ), overlapping brushed (n= 1 ), overlapping brushed-incised (n=1, similar to Spradley Brushed-Incised, a btc J7lh_early 18th century utility ware type st:en in Caddo sites in the Neches-Angelina   The few incised utility ware sherds from the Anglin site have simple straight line decorative elements (Table 14). This includes horizontal and diagonal incised lines on jar rims (figure 29a) and opposed ( Figure 29b) and parallel lines-closely-to widelyspaced-Dn vessel bodies (Figure 29d). One of the parallel im:ised body sherds is from a shell-tempered vessel made along the Red River, most likely Emory Punctated-Inciscd, a common shell-tempered utility ware in Late Caddo McCurtain phase contexts (sec Perttula 2008:352 and Figures 25, 51, and 5!k).
One body sherd (probably from the lower part of the rim) from Anglin has an inciscd-appliqued decorative element (see Figure 29c). This sherd has diagonal incised lines on one side of an appliqued lug, part of a lug handle.
Incised-punctatcd decorative elements arc very rare in the sample of utility wares !'rom the Anglin site (they are absent from the Tuinier Farm and R. A. Watkins ceramic collections), comprising less   Table 4 and see Figure 27h). The inciscd-punclated sherds, including one sherd, from the Anglin site have a lou! punctated row framing a single broad incised line. Incised-punctated sherds are not common in other Titus phase ceramic assemhlages in the Snbine, Big Cypress, and Sulphur river basins, based on an analysis of the decorative composition of the domestic ceramics from 19 Titus phase sites (PerLLula 2005 :Tahle I l-11 ). In these group of 19 sites--each with su bstantial numbers of decorated shcrdsincised-punctated sherds account for less than 3.6% of all the decorated shcrds from each of the sites; at II of the sites, incised-punctnled shcrds comprise less than 1% of all the decorated sherds, and five of the sites had no incised-punctatcd shcrds.
The utility ware rim sherds in the Titus phase ceramics at Lhe Anglin site are dominated by everted rim profi les (59.4 % ) and direct/vertical profiles (39.1 %). One rim has an inverted rim prot1le. Most of the utility ware rim sherds have a rounded lip (86.6%), with a few that have flat lips (6.1%). Not many utility ware vessel rims (6.1 %) have had their lips folded to the exterior of the vessel as otherwise commonly noted on the fine ware vessels and a signi fie ant number of the plain ware vessels (sec below). One utility ware rim has an interior beveled lip.
The plain ware sherds from the Anglin site include 94 rims, 3051 body shcrds, and 114 base sherds; as previously mentioned, plain ware rims account for almost 26% of all the rim sherds from the site, indicative of a substantial plnin vessel assemblage. Three sherds are from a roughly molded and poorly formed small plain vessel, possibly a vessel designed to hold pigments (Figure 30a-c' ).
The variety in rim and lip profiles of the plain rims suggest that plain jars, bowls, and carinated bowls were made and used al the site. Of the rims, 81.5% arc from howls and carinated bowls with direct or vertical proftlcs (Figure 3la-h, d), and 18.5% are everted rims from plain jars. Lip forms are very commonly rounded (58.3%), likely from bowls and jars, flat (9.4%, from jars, howls, and carinated bowls), and rounded and exterior folded (28.1 %); these latter rims (see Figure 3lc, e) are likely from plain carinated bowls. Other lip forms present in the plain wares are rounded and exterior thickened (2.1%) and flat and exterior folded (2.1% ): these arc from bowls and plain carinated howls.
The use of grog temper is pervasive among all three wares at the Anglin site (Table 15). The de-tailed analysis of a sample of 546 sherds from Lhe site indicates that beL ween 86.1% and 99':'/ n of all the shcrds are from vessels made with grog temper inclusions. In most cases, grog was the sole temper inclusion. The highest proportions of grog temper occur in the plain wares and utility wares.
Other temper inclusions used by Caddo potters who lived along Stouts Creek include crushed and burned bone (with the highest proportions seen in the fine wnres; bone-tempered pollery is more common in the tine wares at the Tuinier Farm site, see Tahle 6); hematite (most abundant in the Anglin site plain wares; at the Tuinier Farm site hematite-tempered pollery is most prevalent in the fine wares); and charred organic remains (most common in the fine wares) (see Table 15). A naturally sandy clay paste was used for some of the vessels manufactured in all three wares, particularly in Lhe utility wares, but sandy paste grog-tempered pottery is slightly more common overall at the Tuinicr Farm site, although the utility wares al that site also have the highest proportion of sandy paste sherds (see Table 6).
The most uislinctive aspect of the Anglin sherds is Lhe considerable number of shell-tempered sherds in the fine ware class (13%) (see Table 15); al the Tuinicr Farm site, only 1.9% of the fine ware sherds were made with shell temper. Although no chemical analyses have been conducted on any of the sherds from the Stouts Creek sites to confirm the supposition, previous instrumental neutron activation analyses (INAA) or shell-tempered sherds from Northeast Texas Caddo sites, including Titus phase sites, indicate that she!Hempered vessels were made by Late Caddo McCurtain phase groups that lived on the middle reaches of the Red River, in the vicinity of the conll uence of the Kiamich i and Red rivers (Cogswell et al. 2008). Outside of the Red River valley, shell-tempered vessels are quite rare. These INAA findings indicate that a number of engraved shelltempered trade vessels-typically Avery Engraved, but also including utility wares-had been obtained in the course of contact and exchange by the Caddo people~ living at the Anglin site on Stouts Creek.
The sherds from the Anglin site arc from vessels fired in approximately the same manner as the ceramic sherds from the Tuinier Farm assemblage (see Table 7). That is, technologically, the majority of the sherds in the Anglin ceramic assemblage are from vessels fired in a low oxygen or reducing environment-especially the fine wares-with the greatest proportion of those then pulled from the fire and allowed to cool in the open air (Table 16).

314
At Anglin, hetween 56.6% (plain wares) and 73.9% .. (fine wares) of the sherds nre from vessels fired in a reducing environment. Sherds from oxidized and incompletely oxidized vessels, and from vessels thiH appear to have been sooted, smudged, or reheated, arc must common in the plain wares (43.4%) and utility wares (35.8%) at the Anglin site (see Table I  having been fired under wdl-t.:ontrolled and lengthy firing conditions, limiting the number of vessels that were in...:ompletely fired ur reheated as well as producing vessels that would have been harder and more durable. The firing would also have led to the production of vessels that had the interior and exterior surface colors preferred by the Stouts Creek Caddo potters (i.e., chocolate brown, dark brown, and dark grayish-brown).

Elsbeth Dowd, George Morgan, and Beau Schriever
The purpose of this analysis was to investigate the analytic potential of the electron microprobe (EMP) for examining poltery shcrds from a Caddo archaeological site. Electron microprobe analysis is useu to determ ine the qualitative and quantitative chemical composition of solid materials. There an: several advantages of using the microprobe to study pottery shcrds. First, the microprobe can be useu to Jetermine the chemical composition of very small locations, ranging from 0.2 to 20.0 um. This makes it possible to take separate readings of the clay and temper, analyzing both the clay size fraction of the paste and the composition of the temper. Second, the microprobe can analyze all clements with atomic numbers greater than or equal to 5, including silica. All of the major clements that make up most rocks and seuiments can be idemificd, which could potentially be useful for differentiating and sourcing clays and tempers. Third, the microprobe has excellent digital imaging capabilities, accompanying the precise compositional readings.
This project was conducted at the University of Oklahoma Electron Microprobe Laboratory, with the assistance of Dr. George Morgan. The pottery shcrds were provided by Dr. Timothy K. Perttula. They are from Tuinier Farm ( 4 LHP237), a 16 1 h to J7 1 h century Caddo site probably affiliated with the Titus phase. Analysis of these sherds demonstrates that the electron microprobe is useful for determining temper composition, and may be useful for differentiating the clays in each sherd.

METHODS
Samples were prcpareu for analysis as thkk sections. A cross-section of each shcrd, roughly 0.5 to 0.75 inches in length, was removed and embedded within a l-inch PVC ring using a two-component epoxy. The rings were cleaned and one end taped closed to produce the form for holding the epoxy. The surface of the shcrd sample to be analyzed was ground flat anu this surface was placeu face down in the ring and pressed down to adhere to the tape.
Due to the friable nature of the ceramic samples, they were placed under a low vacuum to help the epoxy impregnate the ceramic body.
Once the epoxy has set, the samples were then hand polished llat using a sequence of progressively finer grit films and diamond slurries, with the final grit a 0.25 micron diamond slurry on a cloth pad. The polished thick sections were then sonically cleansed in water to remove all loose material. Following the cleaning, the thick sections were dried at low temperature in a lab oven. Finally, they were carbon coated to both ground the sample and make it electrically conductive, required conditions for microprobe analysis.
For each of the thick sections, microphotographs were taken for use as reference maps during analysis. This step was necessary because the microprobe is only capable of imaging a small portion of the sample at a time. The microphotographs provided a means to record the location of acquireu backscattered electron (BSE) images and to identify temper. BSE imaging was used to select clay matrix and temper locations for identification using the Energy-Dispersive X-ray Analyzer (EDXA) and to capture windows on the sample documenting the analysis. The BSE image windows were acquired, saved as TIFF files, printed, and then useu to mark and record the readings of clays and tempers.
Preliminary EDXA readings were taken of selected clay and temper lm.:ations on each sample to acquire a general understanding of the composition of each she rd. After this, I 0 additional readings were taken of the day portion of the matrix for each shcrd. Minerals analyzed incluue Si0 2 , Ti0 2 , Al_p 3 , FeO, MgO, CaO, Nap, and K 2 0. This provided a larger sample for the chemical analysis of the clay portions, so that the composition of the clays could he more accurately compared.

RESULTS
Sherd #I (a carinated howl body sherd with engraved ovals, likely Ripley Engraved) has a multigenerational grog temper. Although there is not much quartz in the body matrix, the grog does contain quanz, along with smaller pieces of grog. Eight BSE images were taken in six separate areas of the sample (Figure 32a-c). Two EDXA readings were taken in Area I, one of the matrix clay in the body and one of the matrix clay in the grog (Table L 7).

Bone temper
Sherd #2 (engraved carinated bowl body sherd with a hatched triang lc pendant from a series or wrvilinear lines) has quartz and feldspar temper. This includes three different feldspars, which arc mostly end member K-spars (90-100 Orthoclase). Four BSE images were taken in four areas of the sample (Figure 33a-c). Two EDXA readings were taken in Area 1. The first was ofKfs-1 temper, which originally developed from high-temperatu re magma. The second was of Tourmaline (schorl) temper. This was probably originally part of a paraluminous granite, which is generally derived from the melting of preexisting sediments. Two EDXA readings were taken in Area 2. The first was of the gray matrix clay in the body. The second was of white clay, which may be part or a grog temper, or whi~.:h may be part of a non-homogeneous section of the body paste.
Sherd #3 (engraved carinated bowl body sherd with a panel ftlled with short vertical lines; a red pigment had been rubbed into the engraved lines, probably Ripley Engravt:d) has a temper of either grog or of crushed fired clay, much like Sherd #2. As in Shcrd #2, there are a numbt:r of end member Kspars, but no noted Tourmalines. Both shcrds #2 and #3 also contain high quantities of quartz. Allhough they look very similar in mineral content, however, Shcrd #3 has a lower silica content and a higher iron oxide content than Shcrd #2. Six BSE images were taken in six areas of the sample (Figure 34a-c). Two EDXA r~:adings were taken in Area 1. The first was of matrix clay in the potential grog temper, and the second was or matrix clay in the body.
Sherd #4 (a Ripley Engraved carinated or compound bowl sherd with a scroll element) has a dark grog tt:mper with denser, finer-grained clay particles than those in the body matrix. We arc uncertain why the grog is so dark, but it could be due to carbon or organic matter. Four BSE images were taken in four areas of the sample (Figure 35a-c). Two EDXA readings were taken in Area 2. The first was of matrix clay in the body, and the second was of matrix clay in the grog temper. Sherd #4 also contains a large amount of quartz, made up of smaller, denser particles than in Sherds #2 or #3.
Sherd #5 (probable Avery Engraved body sherd from a Red River trade vessel with a hatched ladder clement; macroscopic examination by Pcrllula suggested it did have an exterior red slip) has a shell temper in a range of sizes, from relatively large pieces down to clay-size particles. There are also some bone, hematite, quartz, and Bryazoan inclusions. Six BSE images were taken in six areas or the sample (Figure 36a-c). Four EDXA readings were taken. The first was in Area 1, or the shell temper. The second reading was in Area 2, of the clay matrix. This spectrum showed a high level of calcium, but this was probably from minute particles of ground shell. The third reading was in Area 4, on the edge of the shcrd. At first we thought that there may have been a slip applied to the vessel, but the composition a b c Figure 32. EMP sherd 237-1: a, areas 1-6; b, Area 2 at SOx; c, Area 3 at 50 x. looks the sam\: as that of the clay matrix in Area 2, so we think that the edge was simply burnished. The fourth reading was from Area 6. of a piece of bone, identifiable by the spike in phosphorus.
Following the initial EDXA readings, I 0 additional reading were taken from the clay matrix in each shcrd. The samples can best be compared by looking at the normalized weight percent oxides of the different minerals (the center columns in Table 18). The clay matrices on two of the sherds (#2 and #5) are different from the other three. The silica content is higher in Sherd #2 than in any other sherd, making it distinctive. The calcium content is high in Sherd #5, but this could be due to the large quantity of crushed shell in the matrix, rather than to any properties of the clay. The other three sherds all look relatively similar, up to the 1-sigma level. Shcrd #I may be somewhat distinct based on iron content, hut this may not be effectively distinguishable at the 2-sigma level.

CONCLUSIONS
Analysis of these shcrds using the electron microprobe demonslralt:d the instrument's utility for dose identification of temper and paste composition. The method was also used to identify the chemical composition of the matrix clay in each sherd. While it was possible to differentiate the shen.ls based on chemical composition of the clays, it is uncertain whether this would be useful in a broader analysis. The electron microprobe may have great pott:ntial to complement other analytic methods, such as instrumental neutron activation analysis and laser oblation, in the analysis of pottery sherds from the Caddo area.
More information on electron microprobe analysis can he found on the following websites:    Two complete elbow pipes have been found with one or two of the three Late Caddo burials at the Tuinier Farm site. The first pipe has had the ba~.:k enJ of the stem turned up vertically against the back end of the bowl, with indentations where the bowl and wrapped-around stem meet. Tt is decorated with four hatched engraved triangles pendant from the bowl (Figure 37a Six pipe sherds or pipe sherd sections, all from grog-tempered elbow pipes, have been found in the 2007 excavations in the southern midden (Midden 1) at the Tuinier Farm. These include a plain stem fragment and two plain bowl rim sherds. The other pipe sherds are from probably two different decorated pipes. 0 1 2 3 4 5 centlmet~r~:.,.:!: · · The first decorated pipe (Unit 4, 20-30 em bs) has two horizontal engraved lines and rows of small circular punctates on the elbow pipe stem ( Figure  39b-b'). The punctates occur in two rows between the engraved lines and in a third row underneath the engraving. In addition, there is at least one row of circular punctates that extends vertically down the stem towards the bowl-stem attachment. The stem is a maximum of 37 mm in height, with an exterior orifice diameter of 24.9 mm; the stem is 5.6 mm thick. The second pipe is a bowl with diagonal and semi-circular engraved elements separated by a narrow band of rocker stamping (Figure 39a).

Anglin
The excavations at the Anglin site have recovered four elbow pipe sherds from four different grog-tempered pipes; two of the pipe shcrds have a naturally sandy clay paste. Three of these sherds arc undc~.:orated, including a pipe bowl rim (direct profile with a rounded lip), a nat-lipped stem, and a sherd from the lower portion of the stem. The fourth elbow pipe sherd is a flat-lipped stem (grog-tempered, with a sandy paste) with a single horizontal engraved line below the lip and at least one hatched triangle pendant from the horizontal line ( Figure 40). One of the complete pipes from the Tuinier Farm site has the same engraved motif, except executed on the bowl rather than the stem (see Figure 37a).

FIGURINES, EAR SPOOLS, AND OTHER CLAY OBJECTS
The Stouts Creek sites have an assortment of day objects of varying forms, including fragmentary pieces of low-fired clay ligurines from both the Tuinier Farm and Anglin sites and a number of ear spools from the midden excavations at Anglin. Such objects, especially figurines, are very rare on Caddo sites of any age, and car spools when found in Titus phase contexts arc usually recovered in burial features (see Turner 197R). Although the function or functions of figurines found at the Stouts Creek sites is not known, it is doubtful that they were used as toys, an explanation offered for the figurines found on Plains Village sites on the southern Plains (Bell 1984:320). Their rarity on Caddo sites suggests use as anything other than toys. Newell and Krieger ( 1949: 151) note that the animal and human figurines found at the George C. Davis site were intentionally broken across the neck or torso, and they hint at both their ceremonial and magical use by Caddo peoples at that site.
Aboriginal societies in the Southeastern U.S., including the Caddo area (see Swanton 1942: 163-166, 211-216), had ideological systems that defined a dose relationship between human~ and animals, perceiving both to occupy a conceptual category of "intelkctual beings." Thus, beliefs and myths would often allude to the descent of humans from animal ancestors (in the case of the Caddo, this would include bears, dogs, beavers, and coyotes [Swanton 1942:215 J), and then attribute a host of anthropocentric characteristics to animals, including powers or qualities to which humans aspire. Animals arc often responsible in myths for defining or illustrating cosmic relationships. Therefore, a closeness between humans and animals, disclosed in myths and demonstrated in rituals, suggest that animal and human figurines (miniature animals and humans, cf. Laugrand and Oosten 2008) may well be powerful symbols of religious and cosmological beliefs for the Stouts Creek Caddo peoples. Such figurines may also have held transformativc properties in myths and rituals, transforming beings in life and death.

Tuinkr Farm
There arc two possible clay figurine fragments in the general collections at the Tuinier Farm, both possible leg or limb pieces (Figure 41 c, f). Om: is at least 39.5 rnm in length and 11 mm in width, while the other is 19 mm in diameter.
There is also a llat spatula-shaped fired clay piece (see Figure 4le) in the general collections; similar pieces have been found at the Anglin site. The one from Tuinicr Farm is 58 x 18 x 7.2 nun in length, width, and thickness. Its function is unknown.
One of the clay objects previously found in a general context at Tuinier Farm is a 25 mm long clay bead (Figure 42a-b). A second bead-15.5 111111 in diameter-was found in Unit 2 excavations in the northern midden (see Figure 41 b). A small clay ball or bead (16.5 mm in diameter) was also recovered in the southern midden (see Figure 41a).

R. A. Watkins
The collection has a single clay object. It is a small clay ball approximately 14 mm in diameter. Similar clay balls have been recovered from the excavations in the Anglin midden (see below).

Anglin
A wide assortment of clay objects have been found at the Anglin site, in numbers not previously seen in Late Caddo Titus phase sites. These clay objects im:lude figurines and figurine fragments, small clay balls, clay beads, and several ear spools (Table 19), as well as other pieces of uncertain function or use.
The one notched clay piece, with three notches, is a tabular piece of clay at least 25 mm in length and 8.9 111111 in thickness ( Figure 43a). The clay has pieces of temper in its paste.
There are two oblong pieces of clay from the Anglin site that are referred to as clay squeezes because they both have fingerprint impression on them (Figure 44a-b). These range from 32-48 mm in length, 18-23 nun in width, and 17-22 mm in thickness. These may be the beginnings of unfinished figurines, or extra wide and thick clay coils, rather than morphologically purposeful clay artifacts. Eight small anJ roughly round clay balls have been found in the Anglin midden, five with protrusions (Figure 45a-d); the purpose of the protrusions on some of the clay halls is not known, though they may have been designed to a~sist with the attachment of the clay halls to another object. These are not well-shaped or smoothed, but are lumpy; nunc have perforations. Two of the three clay balls without protrusions have Jlngerprint impressions ( Figure  46a) and another is hollowed-out on one side of the piece (sec Figure 43c). These clay halls range between 20-24 mm in diameter; the clay balls with protrusions are slightly larger, ranging from 21-43 mm in diameter.
There are three clay perforated beads designed for suspension on a string. One is tubular-shaped, and 15 x. II x II mm in length, width, and thickness, while the other two are rectangular-shaped (sec  Figure 43b). These range from 14-19 mm in length and 10-14mm in width.
Four clay pieces are relatively flat and spatulashaped, with one rounded end (Figure 47a-c). One of these has a raised clay protrusion or attachment at one end of the piece. These clay pieces range from 15-20 mm in width, 7.7-12 mm in thickness, and arc at least 29-35 mm in length.
Another interesting category of clay objects from the Anglin site are five clay pieces that have clearly defined tapen.:d points on them ( Figure  48a-d). These range from one rounded piece (38 x. 26 mm in length and width) to finely-shaped and narrow tuhular pieces (8-15 mm in width anJ 23-26 mm in length), each with a point at one end. One of the narrow tubular tapered point clay uhjects has a hole at one end, as if it was meant to fit onto a stick or some other sort of holder. No.

Figurines
One of the figurines from general midden contexts at the Anglin site (and nul included in the counts in Table 19) is a small zoomorphic figure, possibly a bird or owl, sitting down with two legs in the front of the body. The head has two eyes and a mouth ( Figure 49). There are 21 other figurine pieces from the Anglin site (see Table 19). None of the figurines from the Anglin site resemhle horses, which are a common form of figurines in post-1720 archaeological contexts in parts of the Southern Plains and the Red River, including at least two historic Caddo sites, Womack ( 41 LR I, Harris et al. 1965:303) and Roseborough Lake (41BW5, Miroir et al. 1973: Figure 6e). Figurines on Caddo sites predating the llSth century arc quite rare, as discussed below.
The other figurine fragments from the Anglin site consist of unidentified rectangular or tabular fragments (n=8) that may be body or torso pieces (one has fahric impressions on one side and another has fingerprint impressions), another six blocky torso or body pieces (anirnal or human), including one with a protrusion or appendage, three that are lcg/limh or basal pieces (including one identical to a figurine fragment from I Hth century Caddo contexts at the Roseborough Lake site [Miroir ct al. 1973: Figure 6g]), and two small rectangular pieces of uncerlain location on the figurine (Figure 50a-c, e). One of the rectangular or tabular fragments has a small hole at its base, probably to facilitate fitting it on a stick or other kind of holder. Marlin Hawley (2008 personal communication) suggests that these holes arc put into solid clay objects in order to keep them from shattering during firing.
A more complete figurine is anthropomorphic, with two legs and the area for a head (see Figure  SOd   Webb (l<J4S:127-128 and Plate 16:4) has reported on figurine fragments from several Bossier phase sites in northwestern Louisiana. They arc human torso fragments; Webb ( 1948: 128) estimates that complete figurines would have been from 5-S em in height. There arc human and animal figurines (bird and dog) at the Belcher site on the Red River in northwestern Louisiana (Webb 1959: 176-177 and Figures 13f, 22a, and 35g); two may have been attached to pottery wssels. The one free-standing figurine is a small human figure ( Historic Wichita sites along the Red River and elsewhere in the north central part of Texas (see Smith 1993:Figures 24j-l and 26g-h) do have quantities of day figurines, as do some prehistoric Plains Village sites in the Washita River basin in south central Oklahoma (Bell 1984: Figure 14.3d-h). Hundreds of mostly broken figurines have been reported from the Spanish Fort complex of sites, which date from the mid-to late-18th and early 19th centuries (i.e., Bell 1967:Figures 47a-j and 57n-p). Identifiable pieces from these sites include complete and fragmentary human effigies, as well as quadrupeds, particularly horses and horses with riders. Bear, deer-like animals, and bird (probably owls) forms may also he depicted. Some of the anthropomorphic figures have separately modeled limbs, pa~ticularly legs.
An I Sth century site on the Colorado River in west Texas reported by Skinner ( 1978) had a number of fragmentary fig.urines depicting humans (n= I 02), dogs, horses, and horses with saddles. According to : All of the figures appear to be handmade by rolling and pinching the clay to form the desired shapes. There is no evidence of molding .. . Appendages arc not well made and no attention was paid to creating fingers or feet. Most of the human figures are estimated to be about 10 em high although on~ example is considerably smaller.

Ear Spools
The nine clay ear spools or ear ornaments from the Anglin occur in several different forms and sizes, with different ear attachments; none of them are decorated. Form A includes one large circular spool, 31 in diameter and 6 mm thick with a small interior conical plug or allachment (Figure 5ld-d').
Form B (n=2) arc tubular-shaped car spools, with equal-sided tlanges or sides, a shallow central groove, and no interior plug (see Figure 51 b-b'); Turner (1978: Figure 2ld The Form E ear spool (n=l) at the Anglin site is circular in shape (sec Figure 5lc-c'), 19 mm in diameter, and very thin (2 mm), with a central interior plug or attachment that is 11 mm in diameter and height. A fragmentary ear spool piece from Unit 8 in the midden excavations may be from a second FormE ear spool. This piece is 17 mm in diameter and has a central interior plug.
The last car spool form (Form F) includes two large circular disks (18-20 mm in diameter) with large central interior plugs (see Figure 5lf-f'). These attachments stand 17-1 R mm in h~:ight.

Perforated Sherds
Spindle whorls arc disk-shaped sherds (usually base sherds) that have a central perforation or hole drilled in them. The spindle whorl would have been affixed on a spindle to help maintain its rotary motion during spinning activities. The presence of spindle whorls on these Caddo sites suggests that Caddo women at the Stouts Creek sites were processing fibers to produce textiles (cf. Alt 1999). Materials that could have been used incl~dc ani mal hair and various vegetable fibers, among them hemp, slippery elm, mulherry, milkweed, and nettle, as well as the hark of trees.

Tuinier Farm
A single perforated sherd (with one complete perforation and a second partial perforation) comes from the northern ~· 1idden 2 at Tuinier Farm (see Figure 41 d). The perforated sherd is from the hase of a grog-tempered vessel.

Anglin
There are four perforated plain body and base sherds from the Anglin midden excavations. Each has a single perforation that ranges from 8.0-11.6 mm in diameter.

Tuinier Farm
A singlt: clay coil was recovered from excavations in the southern midden at the Tuinicr Farm. Its discovery suggests that the Caddo were engaged in c~ramic vessel manufacture at the site, because clay co!ls are the discarded remnants of the manufacture of coiled pottery vessels by Caddo potters that became exposed to lire and were preserved. They provide incontrovertible evidence for on-site ceramic vessel manufacture. The coils are roughened and unsmlX)thed.

Anglin
Clay coils and fragments of clay coils with rounded ends are numerous in the Anglin midden, as 27 clay pieces have been recovered in previous investigations here (Figure 52a-e). At Anglin, the clay coils and fragments are preserved as both narrow (n= 18), between 6-L 2 mrn in width, and wide (n=9) coils. The wide coils range from 13-24 mm in width.

Burned Clay and Daub
Tuinier Farm A single piece of daub and at least 48 pieces of burned clay were found in the 2007 investigations at the Tuinier Farm. These pieces were found in both midden areas. The virtual absence of daub in the archaeological deposits suggests that the Caddo structures at the Tuinier Farm site may not have had a wallle and daub cover.

R. A. Watkins
There are six pieces of burned clay and six pieces of daub in the collection from this site. Their recovery suggests that clay-lined hearths, ovens, and daubcovered structures are likely present at the site.

Anglin
In addition to a piece from a mud-dauber nest, pieces or daub (n=214) and burned clay (n=638) are relatively abundant in the midden deposits at the Anglin site. As at the R.A. Watkins site, the recovery of daub and hurned clay suggests that clay-lined hearths, ovens, and daub-covered structures are likely present in the area of the midden or at other locations at the site not far removed from the trash midd<.:n accumulation.

LITHIC ARTlFACTS
Tuinier Farm A number of Late Caddo period triangular an·ow points of the Maud and Talco types with conc~ve bases have been found on the surface from the tmdden areas at the Tuinier Farm site (Figure 53). They are typically made of a heat-treated local quartzite.
In the borrow pit area at the southern end of the site, a wider range of arrow point forms made from a diverse range oflithic raw materials have been found in investigations led by Lee Green (Figure 54). They include triangular Maud and Talco points and stemmed arrow points ranging from Late Woodland/ Early Caddo in age (Scallom and Alba types) to Late Caddo forms (Perdiz and Bass<.:tt). Shafer and Green (2008) also document a range of Late Paleoindian to Archai<.: projectile points in this same area.
A 70 mm long beveled knive of a non-local gray chert was one of the grave goods found with one of the Caddo burials at the Tuinier Farm site (see Figure 38a, bottom). It was found in direct association with a plain elbow pipe. Beveled knives have been found in other Titus phase mortuary contexts (Perttula 2005:287 and Figure 6-41 ).
In the 2007 investigations, from hand excavations and surface collections, we recovered 51 pieces of lithic debris and seven tools, both chipped and ground. The chipped stone tools (n=5) include chen and quartzite bifa~.:e fragments from a general surface context, as well as lwo expedient flake tools and a side scraper rrom the southern midden units. One of the expedient flake tools and the side scraper are made on flakes of local quartzite, while the other flake tool is on a non-local gray novaculite flake. The ground stone tools are a ferruginous sandstone abrader (Unit 3 in the southern midden) and a greenstone celt fragment (general site collections).
The lithic debris from Tuinier farm is dominated by quartzite (n=38, 74.Y10 from local gravel sources, mainly heat-treated to improve its knappability (cf. Shafer and Green 2008). Petrified wood is another local raw material that was knapped to make chipped stone tools: this material comprises 13.7% of the lithic debris. The remaining pieces of lithic debris produced during the manufacture of chipped stone tools include a light gray chert (n=2, 3.9% ), brown chert (n= 1, 2% ), and a grayish-hrown  chert (n= I, 2% ), as well as a piece of quartz. All these materials may be available in local stream gravels, but likely not in large quantities or as more than small pebbles. Finally, there is a piece of debris from the resharpcning of a celt.

]{, A. Watkins
Only a handful of lithic debris from chipped stone tool manufacture is in the site artifact collections. These includes pieces of quartzite (n=5) anti dark grayish-brown chert (n=l) pieces.

Anglin
Previous excavations in the midden deposits at the Anglin site have recovered a number of Maud and Talco arrow points (Figure 55). Most of these appear to have been made from the local coarsegrained and heat-treated quartzite.
In our 2007 investigations, we recovered 118 pieces of lithic debris and two core fragments. One of the core fragments was on a heat-treated qum1zitc pebble (ST B, 20-40 em bs), while the other (Unit 1, 10-20 em) is on gray chert. Both core fragments have a smooth cortical surface, indicating the raw material was collected from stream gravels.
The lithic debris is overwhelmingly dominated by flakes from local lithic raw materials, including quartzite (n=96, 81.4%) and petrified wood (n=16, 13.6%). The remaining pieces of lithic debris arc black chert (n=l, 0 .8%), yellow chert (n=l, 0.8%), gray novaculite (n=3, 2.5% ), and claystone/siltstone (n= I, 0.8% ). With the exception of the yellow chert, which can likely be found in low quantities in local stream gravel pebbles, the black chert (Big Fork chert), novaculite, and claystone/siltstone are non-local lithic raw materials gathered from gravel sources no closer than the Red River, about 110 km to the northeast. From this lithic raw material data-incomplete though it may be-the use of non-local lithic raw materials during the Titus phase occupation of the Anglin site was minimal. These Caddo apparently did not have much in the way of a dependable access to higher-quality Iithics and had to rely on difficult to knap quartzite anti petrified wood materials.
About 70% of the quartzite lithic debris from the Anglin site came from previously heat-treated pehbles. About 38% of the quart£ite pieces are cortical, with a stream-rolled surface. indicative of the earlier stages of lithic pebble reduction.

Timothy K. Perttula and Lee Green
A single Clements-style marine sht:IJ disk (Perttula and Green 2006:22), probably used as an ear disk, is in the collections from the Anglin midden (Figure 56a-h). This particular disk is 22 mm in diameter. 3.5 mm thick, and has a central dot and a single engraved circle that is 16 mm in diameter. A second and smaller engraved shell disk (Perttula and Green 2006: Figure 3) from Anglin was not available for examination for this article.
Clements-style marine shell disks have been found at two sites in the Stouts Creek valley, both from midden contexts, and at only six other Caddo sites in the Ouachita, Red, and the Big Cypress stream basins in Northeast Texas, Northwest Louisiana, and Southwest Arkansas. Ceramic vessels found in burials at these other six sites imlicate that the Caddo occupations there took place from ca. A.D. 1650-1700(Pcrllula and Green 2006 Only two gastropods were submilled from the collections, both from the Anglin site. One was a Rabdutus dealbatus, which can be found in floodplain forests or in prairie grasslands. The second is Mesudon thyroidus, which indicates the presence of trees. Twenty-three freshwater mussel valves were identified from the three sites. Normally, the Minimum Number of Individuals (MNI) would be less, hut with so few shells, the valves could be compared to one another and no matches were found. The identi ficd valves are Lam psi/is hydimw ( 12, 52.2% ), Leptodeafragilis (6, 26.1% ), Uniomerus declivis (3, 13% ), Trunci /Ia tmncata (1, 4.4%) and Quadrula pustulosa (1, 4.4%). Of the valves, one L. fragilis valve was recovered from Tuinier, one U. declivis valve was from the R. A. Watkins site, and the rest nre from the Anglin site. The greatest number of valves (n=l2) and the widest variety of species (n=5) were recovered from Unit 4 at Anglin. The unit was dominated by Lwnpsilis hydiana with seven valves.
Uniomerus declivis can stand dewatering, but based upon the presence of the other mussel species, it is unlikely the stream was dry. The rest of the mussels arc such generalists that no other environmental information can be discerned.
Fragments consisting of umbos and shells were abundant and ranged from unburned to heat-treated (gray in color) to burned black. The valve count for L. fragilis may be deceiving because of the amount of thin shells present within the fragments . The shells appear to break along the lateral tooth. At least one freshwater mussel, either Putamilus purpuratus or Amhlema plicata, is present based upon a few very thick shell fragments, hut no identifiable umbo of these species could be found.
Based upon the range of sizes of the L. hydiana and U. declivis, the site inhabitants were not selective in their choi(;e of species but were taking whntevcr freshwater mussels were present. It does not look as if freshwater mussels were a major part of the diet and probably were exploited at one time or very cautiously over time because it takes generally four years for a freshwater mussel species to replenish an area once it has been depleted.
Interestingly, sexual dimorphism could be discerned in the Lampsilis hydiana shells. As far as I know, no studies in Texas have been done to determine if the aboriginal inhabitants were selective sexually in their choice of freshwater mussels.
The fish host for Leptudeafragilis is the freshwater drum (Aplodinotus grunniens) whereas Quudrula pustulosa has several hosts, the shovelnose sturgeon (Scaphirhynchus platorynchus), black bullhead (Ameiursu me/as), brown bullhead (A. nebulosus), channel catfish (lctalurus punctatus), flathead (;at fish (Pylodictis olivaris) and white crappie (Pomoxis annu/aris). The saugcr (Stizosdedion canadense) and the freshwater drum are hosts for Truncil/a tnmcata (Howells ct al. 1996:76, 122, 146) (Fullington 1994(Fullington , 1995Yates 1993;Zimmerman 1999). However, the mussels were recovered from sites along the South Sulphur River and its tributaries whereas the Tuinicr Farm, R. A. Watkins, and Anglin sites were found along the spring-fed Stouts Creek and its tributaries, which is mapped as intermittent on the Purley, Texas 7.5' USGS quadrangle. An analogy, however, can be found in Tarrant County along White's Branch, an intermittent drainage within the Fort Worth Prairie. A small mussel shell lens site (41 TR 132) was found along the drainage just south of where a tributary flows into the branch (Skinner and Whorton 1993 ). lt was postulated that the creek was spring-fed; otherwise, the presence of the shell lens site was more problematic.

LeeAtma Sclmiebs
Introduction Archaeological investigations over the last few years by several parties at three 16 111 to 17th century Caddo sites in Hopkins County, Texas, have yielded a combined total of 1,297 identifiable faunal specimens. Unidentifiable fragments were not recorded. Colle~:tions from the surface and in test excavations in the context of a recently plowed tleld at Tuinier Farm (41HP237) resulted in the retrieval of 337 bones. Nine hundred twenty seven pieces came from the Anglin Midden (41HP240), located in a lightly wooded area adja~:ent to a fence row next to another plowed lield. Thirty-three fragments were found on the surface of the R. A . Watkins site (41HP238), a smaller third midden in the same area. The assemblage from each site is well preserved, and taxonomic recovery is diverse. All classes of vertebrates are represented, but mammals are clearly dominant. A complete inventory of the faunal remains is in Appendix 2.

Methods
Standard zooarchacological techniques have been used. Attributes of the identifiable pieces consist of taxon, element and portion of that element, anatomical location of the clement, any notes on age, burning, and presence of modification if applicable. Provenience information was recorded when available, but most specimens (at least from Tuinicr Farm) were surface colle~:ted. The prehistoric vertebrate remains were inventoried using Excel 5.0 to manipulate the generated data. Weights of specimens and burning were recorded, but are only provided as documentation for future reference. Identifications were made to the most specific category possible depending on the condition of the bone and available comparative skeletal material. Only positive identifications resulted in the assignment of elements to genus or species. Bonnie Yates at the U . S. Fish and Wildlife Forensic Lab in Ashland, Oregon, ~:on firmed the identification of several specimens when comparative material was unavailable or osteological references were inadequate.
Quantification of the assemblage from these three sites is summarized as number of identified specimens per taxon (NISP) and as minimum number of individuals (MNl) for identified clements from each site (Table 21). The MNI method was chosen as the most suitable analytical measure of abundance. "It involves no hypotheses and is purely factual. The minimum number of animals that the bones could have ~:orne from is an indisputable fact" (Chaplin 1971 :69-70).
MNI estimates were calculated according to the most fre4uently occurring element, based on symmetry and element portion (Munzel 1986). In the mammalian class, teeth are usually used whenever possible (teeth still retained in socket were counted but not weighed). However, post-cranial elements were often used in this collection. In some cases, the presence or a single element constituted an MNI of one.

Results
The sites are located on the extreme eastern edge of the Post oak Savanna, and the western edge Fishes of the Pineywoods, on Stouts Creek, about 15 km south of White Oak Creek, <1 large tributary of the Sulphur River. This area includes a wide variety of habitats exploited by the Caddo. The following section discusses the animals recovered from each of the sites and their preferred habitat (Table 22). Burned specimens are listed in Table 23.
Found in a borrow pit area at the Tuinier Farm site, gar (Lepi.sosteus sp.) is represented by ont: scale. Gars are cigar-shaped predatory fish with thick diamond-shaped scales and beak-like jaws with sharp pointed teeth. They are known to  Two otoliths from freshwater drum (Aplodi-IIOIIIS grwuziens) were recovered from the Anglin Midden. Based on the measurements of these specimens (Witt 1960), one individual was 317 mm long, weighing approximately 400 g. The other fish was much larger, at 647 m long, and weighed approximately 4,440 g. These large fish would have provided several pounds of meat. The preferred habitat of the freshwater drum includes lake shallows and large rivers, and it produces a grunting sound that is audible (Collins 1959).
Four unidentifiable fish remains are also included in the collections. A flotation sample taken in a 40 x 40 em unit (Unit 4, 10-20 em bs) at the Tuinier Farm had two very small unidentifiable fragments from a minnow-sized fish . A vertebra from a medium-size fish came from Unit 21, and a second vertebra from a very large fish was found during general collection at the Anglin site. The second The Archaeology of the 16th And 17th Cellfury Caddo in the Post Oak Savannah of Northeast Texas 69  specimen has been drilled slightly off-center, and the edges arc smoothed from usc-wear, possibly as an ornament.

Amphibia11s
The only amphibian identified in the collection, bullfrog (Rana catesbiana) is represented by two pelvic elements. They were recovered from Unit 3, 10-20 em bs, at Tuinier Farm. The largest of all frogs, it prefers larger bodies of water, residing in lakes, ponds, bugs, and sluggish streams, hiding in vegetated areas (Conant 1975).

Reptiles
Only one plastron fragment from musk or mud turtle (Kinostcrnidae) was identified, and this is from Unit 23 at the Anglin Midden. The musk turtle is commonly called ''stinkpot" because of the glands that secrete an offensive odor as a defense mechanism. "Bottom crawler" is another common description, as they are strongly aquatic turtles generally preferring slow-moving or shallow waters with soft bottoms and abundant vegetation (Behler 1995 ). Distinction between the two is difficult based on a single clement. as there are two genera north of Mexico: Stemotherus, with four species of musk turtles, and Kinostemon, with five species of mud turtles. Currently, the mud turtle (K. subrubrum), the musk turtk (S. carinatus), and the stinkpot (S. odoratus) occupy this part of Northeast Texas.
One pelvic element from a large pond turtle (Pseudemys sp.) was found at the Anglin Midden. They are part of a large group of turtles (including sliders and cootcrs) that range from coast to coast, preferring areas where the water is shallow, the aquatic vegetation profuse, and the bottom son and muddy: in ponds, marshes, ditches, edges of lakes, backwaters of streams, and in prairie sloughs, cattle tanks, and river pools (Conant 1975 ).
Box turtle (Terrapene sp.) is represented at all three sites, with a combined total of 15~ specimens. Four units, one shovel test, and surfm;e collections at Tuinier Farm yielded 46 shell fragments, with a site MNI of four based on hyoplastron clements. One pelvic element and 111 shell fragments came from !7 units at the Anglin Midden, and the site MNI is six (also based on hyuplastron pieces). One nuchal element was found on the surface of the Watkins site.
Two shell fragments from softshell turtle (Trionyx sp.) were recovered from units 7 and 18 at Anglin. Two species occupy the area: the smooth softshell (T. muticus) and the spiny softshell (T. spiniferus). Specific subspecies in Northeast Texas are the Midland Softshell (T. m. muricus) and the Pallid Softshell (T. s. pallidus). All species are aquatic, and the preferred habitat includes small marshy creeks, farm ponds, and large, fast-flowing rivers and lakes (Behler I 995 ). They are powerful swimmers, and they can run on land with startling speed and agility (Conant 1975 ). The carapace is circular, and covered with soft, leathery skin instead of horny scutes. They have long necks, strong jaws, and sharp beaks.
High quantities of indeterminate turtle were also recorded. One toe bone and 45 shell fragments were found at Tuinicr Farm, recovered from three shovel tests and three excavation units (including fine screen and flotation samples taken in Unit 4 ). The Anglin Midden yielded 74 shell fragments from 15 units and general collections. Based on specimen size, most of these pieces are from terrapins or musk/mud turtles. The exceptions include the toe bone and one shell fragment from Tuinier Farm, and two pieces of shell from Anglin: they are from very large individuals (sec Appendix 2). These three shell fragments arc notahle, as they are very water-worn, unlike other pieces in the assemblage. They could only be the remains of snapping turtle (Chelydra sp.) or pond turtle, as they arc tht: only turtles in this size range.
The Anglin site had the only snake bones in the t:ollection, comprised of nine large vertebrae. This includes one poisonous snake (Viperidae) from Unit 1 R, indicated by the long spur protruding vertically from the centrum. The other eight elements were recovered from five units and general collections. Unfortunately, absence of diagnostic attributes, specifically the centrum spur, prevented specific identification. However, they are all similar in size and may be the remains of a single individual. There are four species of poisonous snakes in Northeast Texas: rattlesnakes, copperheads, cottonmouth/ water moccasins, and coral snakes.

Birds
All three sites had the remains of turkey (Meleagris gallopavo). At Tuinicr Farm, four pieces came from three units (including two fragments from a Rotation sample taken in Unit 4 ), and 17 bones were retrieved during general site surface collections. A minimum of two individuals were present at this site, based on distal ends of the tarsometatarsus.
Fourteen units and general collections at the Anglin Midden had 42 specimens. Based on proximal ends of the tibiotarsus, this site also has an MNI of two. Two turkey bones were found on the surface of the Watkins site, with a site MNJ of one. Turkey occurs as wild fowl in open woodland environments (Robbins 19R3), and its presence indicates exploitation of the grassy areas along the edges of the woods.
Tuinier Farm had one tiny specimen from a very small perching bird (Passeriformes). This distal tibiotarsus fragment was recovered from a flotation sample taken in Unit 4 (10-20 cm bs).

Mammals
Opossum (Didelpl1is \'irginiana) is represented by five specimens at two sites; both sites have an MNI of one. At Tuinier Farm, a vertebra was recovered in Unit 3 ( 10-20 em bs), and an upper canine tooth was found during general surface collection. The tooth is drilled through the root area, and is highly polished from use-wear, probably hecausc it was used as an ornament similar to the previously mentioned drilled lish vertebra. Opossum teeth are naturally very sharp, and this piece could have also functioned as a punch-type tool. Unit 17 at the Anglin site contained fragments from a scapula and pelvis, and an ulna fragment came from Unit 20. The opossum is widespread throughout Northeastern Texas, occupying a wide variety of habitats. This includes wooded areas, prairies, and marshes, preferring wetter areas near streams, swamps, creeks, and river bottoms (Schmidly 1983).
Nine-banded armadillo (Dasypus novemcincrus) is represented at two sites. Surface collection at the Watkins site had one vertebra, one cranial fragment, and live scutes. At the Anglin Midden. four pelvis fragments were found in general collections, and a humerus fragment came from Unit 18. The specimens arc modern intrusives.
Eastern cottontai I rabbit (Sylvilagus fioridanus) is represented at all three sites. Seven specimens were recovered from Unit l (0-10 em and 20-30 em bs) at Tuinicr Farm, including a mandible with four teeth. A femur fragment was found on the surface of the Watkins site. These two sites each have an MNJ of one. General site collections and 12 units at the Anglin site yielded 30 specimens. Anglin has a site MNT of three, based on several bones (proximal femur, mandible, and lower second molar). An unfused femur and vertebra indicates that at least one individual is immature. The preferred habitat for the eastern cottontail is brushy areas with grasses and herbs for food and protection from predators; it is found in all vegetated areas of Northeast Texas, occasionally occurring in swamps and woodlands (Schmidly 1983 ).
Black-tailed jackrabbit or swamp rabbit (Lepus ca/ifornicus or Sy/vilagus aqualic:us) is represented at two sites. One femur shaft fragment was found in Unit I (20-30 em bs) at Tuinier Farm. The Anglin site had 36 specimens from general site collections and 13 units. Based on proximal humerii and upper third premolars, the Anglin site MNI is two. Because these two rabbits arc similar in size, distinction between them is difficult, especially based on fragmentary remains. The jackrabbit is rare in the oak-hickory and pine-oak regions of Northeast Texas; the more common swamp rabbit prefers the marshy areas bordering floodplains, woodlands, and grasslands (Schmidly 1983). Based on the location of the sites in the region and their close proximity to water sources, it is likely that most of these bone fragments are the remains of swamp rabbit.
Squirrel (Sciurus sp.) was found at two sites. Eight specimens were recovered from two levels in four units at Tuinier Farm, and the site MNI of two is based on scapula fragments. Six units and general collections at Anglin Midden yielded nine bone fragments, with a site MNJ of one. In Northeast Texas, gray squirrels (S. caro/inensis) are rare in the pine woods and upland forests; fox squirrels (S. niger) arc found in all timbered habitats (Schmidly l9lB).
Pocket gopher (Geomys sp.) is represented by 29 specimens from six units and general site collections at the Anglin Midden. Based on mandibles, there were a minimum of four gophers in the faunal assemblage, including an immature individual. These arc probably the remains of Louisiana pocket gopher (G. bre~·iceps). These may be intrusive remains, although during times of stress could have been dietary supplements. Pocket gophers occur in sandy soils with a low clay content.
Raccoon (Procyon lotor) remains were recovered from two sites. Five specimens came from two levels in two units at Tuinier Farm, and a finely crafted awl manufactured from a fibula was found during general site collections. The Anglin Midden yielded 16 elements from three units and general collection areas. Site MNT at Tuinier Farm is one, and a minimum of two individuals were at Anglin (hased on mandibles and teeth). Raccoons arc found in all vegetated regions in Northeast Texas, The Archaeology of the 16th And 17th Cet!IILry Caddo in the Post Oak Saramwh of Northeast Texas 73 including lloodplains, bottomlamls, and hardwood-Limbered habitats (S<.: hmidly 1983). They seldom occur far from water, and do much of their foraging near or in bodies of water (Davis 19n).
Dog (Ca11is sp.) is well represented in the faunal collection. Tuinicr Farm had 57 specimens, recovered from one shovel test (0-20 em bs), three levels in three units (0-30 em bs), and a general site collection. A minimum of three individuals were present at this site. Three teeth and one foot bone were found on the surface of the Watkins site. Twelve units at the Anglin Midden (including four levels in Unit 24) yielded 208 specimens. Based on the upper first molar, the Anglin Midden has a site MNI of uve, and at least two of these dogs are immature. Domestic dogs (C. familiaris) are often found in prehistoric contexts. Their only domesticated animal, the Caddos used dogs to hunt buffalo and found them particularly useful for routing out bears. and they were eaten in times of extreme scarcity or possibly on a few ritual occasions (Newcomb 1993); most likely the dog remains arc from disturbed burials.
Cougar (Felis concolor) is represented by three bones, recovered from the Anglin Midden. Identifications were confirmed by Bonnie Yates of lhe U.S. Fish and Wildlife Service. Also known as the puma, panther, or mountain lion (although there are no mountains in East Texas), reports of the so-called "black panthers" are common in eastern Texas; they probably occurred throughout the region prior to settlement by Anglo-Americans but have been consistently eliminated over must of the region since the end of the nineteenth century (Schmidly 19~3 ). Deer is the cougar's preferred prey (Davis 1978). One element that compared favorably to cougar was also recovered from the Hurricane Hill site (Yates 1999:346) in the upper Sulphur River basin. Otherwise, bobcat is usually the most common feline found in Northeast Texas faunal assemblages.
One tooth from pig (Sus scrofa) was recovered from Unit 7 in the Anglin Midden. Feral hogs have been present in the United States since the first settlers brought them to Florida in 1539, and there is a sizeable population of feral hogs, European wild hogs, and hybrids in Texas; free-ranging hogs occur throughout the timbered country of Northeastern Texas (Schmidly 1983). This element is probably an intrusive faunal specimen.
Whitetail deer ( Odocoileus virginiwws) is the most common large game animal found in Caddo faunal assemblages. Not only are they the main game animal in Caddo diets, but their hides and bones are also utilized as clothing and louis. A total of 452 specimens were recovered from the three sites, ranging in age from about six months to 4 years old. This is based on tooth eruption, tooth wear (Severinghaus 1949), and epiphyseal fusion of post-cranial elements. The Tuinicr site has an MNI of four, the Watkins site has an MNI of two, and the Anglin site has an MNI of three. Deer occur in all vegetal regions, but in Northeast Texas they are found in larger numbers in timbered areas (Schmidly 1983).
Bison (Bisoll hison) is represented at all three sites. Sixteen specimens were found at the Anglin Midden, including two drilled incisors and several post-cranial elements. The Watkins and Tuinier sites yielded only one drilled incisor each. These teeth were probably worn as pendants. Bison once ranged over almost the whole of eastern Texas, except for the densely wooded Big Thicket, and were probably numerous in the post oak woodlands, which were covered with woods and open prairies; they became extinct very soon after Anglo-Americans occupied the land (Schmidly 1983 ).

Modified Bone
Modified bone refers to faunal specimens with evidence of human alteration such as cutting, grinding, or other reshaping, as well as finished bone tools or jewelry. The three sites had 20 modified specimens (Table 24), and the majority came from the Anglin Midden. They have been grouped into four categories, distinguished by assumed function and/ or form. The system is based loosely on Kidder's (1932) scheme for bone artifacts from Pecos, New Mexico, and an adaptation of this scheme by Beach and Causey (1984) for Arroyo Hondo, New Mexico. In large modified bone assemblages, the categories arc often primarily sorted by anatomical element, animal used, then function.

Type A
This category is comprised of three sharply pointed specimens from the Anglin Midden and one from the Tuinicr Farm. A finely crafted awl is manufactured from the proximal end of a deer ulna (Figure 57a), one of the most common clements used for tools of this type. The thin, tapering of the shaft needs little shaping to form the pointed working end, and the ulnar notch provides a perfect    handle. This pie~.:e measures about 8 em in length, and originally was probably longer but was fractured during use, then resharpened at the broken edge. It is highly polished from usc-wear. The sewnd fragment from Anglin is broken at the shaft and measures 3.5 em in length, with an almost needle-like appearance ( Figure 57b ). Because diagnostic attributes are absent, determination of animal and element was not possible. It is also very highly polished from use, and may have served as an expedient tool after breakage. The third piece is also broken at the shaft, and resembles the mon: commonly found awls: llat in cross-section, a wider mid-shaft, and tapering to the point (Figure 57c). Manufactured from an indeterminate element of an unidcntitiahle large mammal, it measures 4 em in length and is not polished on the surface. The tool from the Tuinier Farm site is of particular interest, as it is in pristine condition despite its dcli~.:acy. Made from a raccoon fibula, it is 9.3 em long, and is also needle-like in shape ( Figure  57d). Perhaps this piece could have been worn as a hair pin as well as functioned as a punch-type tool. Striations from manufa~.:ture and use arc visible on all four specimens.

TypeB
This category includes six bone attifacts with semi-rounded or blunt ends that are not sharply pointed. They were all recovered from the Anglin site. The small deer ulna is broken on the proximal end, just above the finger notch handle, and the distal end is a dull point (Figure 58a). Originally it was probably very similar to the ulna awl described above (see Figure 57 a), and served as a punch-type tool. Two similar fragments from unidentifiable large mammal bones are broad and flat in ~:ross-section, broken mid-shaft, and taper into the dull pointed working ends ( Figure  58b-c). Also hroken at mid-shaft and flat in crosssection are two large mammal hone fragments that have rounded working ends, but arc not pointed at all (Figure 5Rd-e). The final specimen is a broken fragment of unknown function, but remnants of a hroad, dull working end are visible (Figure 580.

TypeC
This group is comprised of ornaments or special decorative items. A leg bone from an in-~· .. determinate animal (a bird or small mammal) is recorded as bead debris (rigure 59a). Transverse scoring is visihle mid-shaft, and the remnants have been snapped off by a ring and snap procedure. The other edgt: is ragged, and is assumed to be the discarded waste from the creation of a hone tube or bead. It was recovered in Unit 3 (10-20 em bs) at the Tuinier Farm site.
Two drilled objects also came from the Tuinier Farm site: a bison incisor (see Figure 59b) and an upper canine tooth from an opossum (see Figure   59c). The opossum tooth is very sharp, and could have also served as a punch-type tool as well as an ornament. The R. A. Watkins site yielded a large bison incisor with evidence of intent to drill a hole through th~;: tooth root, but the hole is incomplete and unfinished (see Figure 59d). Two more drilled bison incisors (see Figure 59e-f) were also recovered from the Anglin site, as well as the drilled vertebra from a large unidentifiable fish (see Figure 59g). One of the bison incisors from Anglin is also unfinished, similar to the specimen from the R. A. Watkins site.

TypeD
Three items from deer bone comprise this category of modified specimens. An antler pedicle from a small deer was recovered in Unit 24 (10-20 em) at the Anglin midden site excavations (Figure 60a). The surface of the base has been ground smooth, possibly from usc as a pestle. The antler shaft is bmken, but would have served well as a comfortably fitting handle. It measures 6 em in length.
Also from the Anglin site is a modified Jeer mandible fragment (see Figure 60b). The diastema at the anterior end (the area closer to the incisors) has been removed, then shaped and ground to form a broad working edge, evidence of usc as a rubbing or grinding implement. lt is highly polished from use, and is 9 em long. Another mollified deer mandible came from Unit 1 (10-20 em bs) at the Tuinier Farm site (see Figure 60c). It is almost complete, including all hut one tooth in socket. The diastema is intact, but the incisor sockets are absent. This is the shaped and ground working edge, much more narrow than the mandible from Anglin. It is assumed that these two implements served the same function, probably as deer jaw sickles (cf. Brown 1964(cf. Brown , 1996Krieger 1946:202 and Plate 23c), but one has been used much more extensively. Both or these mandibles fit comfortably in the hand as does the antler fragment. Krieger ( 1946:193) noted that two or three such deer jaw tools were recovered from the midden excavations at the Sanders site (41LR2) on the Red River, along with a fishhook, bea mers, shaft wrenches, and awls. Four deer jaw sickles were recovered from burial and non-burial contexts at the Spiro site in eastern Oklahoma (Brown 1996:496). Brown ( 1996:496) has indicated that deer jaw sickles are founu on Caddo and Southern Plains sites in Arkansas, Texas, and Oklahoma. is probably minimal, based on the small size of most of the animals recovered. Wooded edges were hunted for deer, cottontail, and turkey. The remaining animals were found in grasslands, woodlands, and hotLomlands.

Summary
The faunal assemblage suggests that the sites could have been occupied throughout the year because the animals identified would have been available during all seasons, specifically the fish, turkey, rabbits, and squirrel. The bison was probably obtained as the opportunity presented itself. However, the young deer indicates hunting during the summer or early fall, as offspring are born in the spring. The shed antler pedicle implies a late winter kill. Winter hunting may also be indicated by the presence of the cougar, opossum, and raccoon. Their pelts become especially luxurious and more valued than the rest of the year. Turtles were probably obtained during the warmer seasons.
The recovery of small animals, especially the turtles, suggests the possibility that entire families took part in the procurement of food. These animals could have been obtained by women and children using passive hunting techniques. Men were generally the hunters of deer and the other larger animals, hut the deer surely provided the main meat source for the Caddo because of its availability. However, turtle, opossum, rabbit, squirrel, pocket gopher, and raccoon are also important dietary resources.
Previous investigations at other Caddo sites in the area document similar patterns of animal resource utilization and general species composition, such as Hurricane Hill. Environmental areas exploited include aquatk and riparian habitats, forests, and open meadows with wooded edges. The modified bone assemblage at these three Caddo sites provide further evidence of site activities such as plant processing as well as animal procurement and subsequent processing. The ornamental pieces may suggest that ceremonial or ritual endeavors also took place at the sites using animal parts. The canine bones could be the remnants of dog burials, as these are common in Late Caddo sit~.:s in this area.
The faunal specimens from the three sites are in very good condition, despite surface exposure and agricultural activities. The information presented in this section provides a representation of broad trends in the subsistence practices of the Late Caddo occupants that lived in the Stouts Creek valley: exploitation of the diverse animal life in the rich ecosystem of Northeast Texas.

SUMMARY AND CONCLUSIONS
The Tuinicr Farm (41HP237), R. A. Watkins (41HP238), and Anglin (41HP238) sites are Late Caddo, Titus phase, domestic habitation sites on Stouts Creek in the Post Oak Savannah of northeastern Texas. The three sites were located and first investigated by Lee Green and associates over the last several years, where they identified considerable midden deposits at each site, located three burial features at the Tuinier Farm site, and recovered a substantial associated artifact assemblage (primarily consisting of ceramic vessel sherds from fine wares, utility wares, and plain wares) and an impressive amount of unburned and burned animal food debris. The Anglin and Tuinier Farm sites also have anumber of clay objects, including figurines and figurine fragments as well as clay ear spools and disk, plus Clements style marine shell ear disks; these kinds of artifacts are otherwise quite rare in other prehistoric or early historic Caddo archaeological contexts in Northeast Texas and other parts of the Caddo archaeological area.
Principally because of the excellent preservation of the midden deposits at the Stouts Creek sites as well as the character of some of the intriguing artifacts from the Anglin site (i.e., Clements-style marine shell ear disks and an abundance of clay objects, including a large assortment of clay figurines and car spools) and the Tuinicr Farm (i.e., a ca. mid-17th century inverted rim Taylor Engraved vessel), limited shovel testing and hand excavations were conducted at the Anglin and Tuinier Farm sites in February 2007 to evaluate their archaeological character in more detail than had been previously done and also to gather first-hand and controlled archaeological data on the artifact and faunal assemblages. In conjunction with this effort, and with the permission of Lee Green and his associates, we also undertook a detailed examination of the extant collections from these Stouts Creek sites. This was done primarily to hetter ascertain the likely chronological age and social and cultural affiliations of the Caddo populations that occupied the Stouts Creek sites. That is to say, it was clear that the sites were occupied by what archaeologists call Titus phase Caddo groups (e.g., Perttula 1998Perttula , 2004Thurmond 1990), likely during the latter part of the phase, or during the protohistoric/early historic period (Figure 61 ), but our intent was to clarify and refine-if possible-the chronological span of the occupations and the direction of cultural contacts A simple but dlective way of determining cultural and ceramic stylistic affiliations between contemporaneous Caddo groups in East Texas and northwestern Louisiana is to make comparisons hetween ceramic assemblages using a series of general decorative classes (i.e., brushed, ridged, incised, engraved, punctated, appli4ucd, and red-slipped) (sec Kelley 2005:61 -66) to "see how much variability occurs in assemblages from nearby regions." Kelley's examination of Belcher and Titus phase sites from different parts of the region, the Bumitt site in the Sabine River uplands in northwestern Louisiana, and sites at Toledo Dend Reservoir along the Sabine River showed "very little variation within each region and significant differences between the regions." Perforce, these similarities and differences in ceramic stylistic attributes and decorative classes lie at the heart of any conclusions about the cultural and ceramic affiliations of local Caddo groups. Determinations of cultural affiliations and dose ceramic stylistic tics between different hut contemporaneous Caddo sites clearly imply the existence of regular contact, interaction, and the sharing of ideas between Caddo peoples living at those sites.
1 employ the same approach here with rcspe~.:t to ascertaining the cultural and ceramic stylistic affiliations of the Stouts Creek sites by utilizing ceramic decorative data (proportions of key utility wares and red-slippc.J wares, since Ripley Engraved is common at just about all these sites) from contemporaneous Caddo sites (mostly of Titus phase affiliation) in the region and comparing that to the decorative class information from the Anglin and Tuinicr Farm sites (Table 25). The sites used in this comparative analysis includes several Titus phase sites in the Dry Creek and Caney Creek localities in the upper Sabine River basin; two suhstantial ceramic assemblages from Titus phase sites on the middle reaches of Big Cypress Creek, but belonging to the western Titus phase ceramic tradition; the James Owens site (41TT769) on White Oak Creek in the Sulphur River basin (Walters ct al. 2003); and the Titus phase Ear Spool site (41TT653) on a tributary stream that flows north into White Oak Creek (Pcrttula and Sherman 2008).
Not just geographically, the Titus phase ceramic assemblages at the Tuinier Farm and Anglin sites helong with the westem ceramic tradition of the Titus phase Caddo (Perttula 2005, editor:404-405): this tradition is marked by higher frequencies of plain wares than eastern ceramic tradition Titus phase sites, punctated utility wares, and La Rue Neck Banded utility wares, abundant use of red-slipping on fine ware vessels, as well as several unspecified varieties of Ripley Engraved. Western tradition Titus phase sites occur in the middle and upper parts or the Big Cypress Creek basin, as well as sites in the upper Sabine and White Oak Creek basins ( Figure  62) The basic differences in eastern and western ceramic traditions within Titus phase sites suggests that there were long-standing dichotomies in belief and cultural practices that may have existed for 150-200 years. This dichotomy suggests that there were well-defined so~.:ial boundaries between the different Titus phase communities-including the community that lived on Stouts Creek-inside and outside the Big Cypress Creek basin and that the cultural landscape across the Titus phase area (Figure 62) was complex and dynamic. Nevertheless, the sharing of a variety of Ripley Engraved motifs across the many different communities, and the basic similarity in much of the utility wares from one area to another, indicates that there was considerable intra-areal interaction and contact between each of the Titus phase communities.
The ceramic decorative category data included in Table 25 points to close stylistic and cultural affi!illfiuns between the Stouts Creek Titus phase sites and conlemporaneous Titus phase sites in the Dry Creek lucalit_v in the Lake Fork Creek basin no/ far tu the south (sec Figure 62). Sites in these two areas share the considerable usc of neck banded and appliqued utility wares, a minimal usc or brushed utility wares, and the relative importance of redslipped vessels (either plain red-slipped bowls and  Figure 62. Map of the Titus phase area, depicting the area with sites having the closest stylistic associations with the ceramic assemblages in the Stouts Creek locality. (see Table 25). Red-slipped vessel sherds are also an important part of Titus phase ceramic assemblages at the Pilgrim' s Pride and Underwood sites in the western part of the Big Cypress Creek drainage, hut here brushed jar sherds arc proportionally quite common (35-46% of all the decorated sherds), as they arc at the Ear Spool site in the western part of the Sulphur River hasin (see Table 25). Another way to measure the stylistic and cultural associations between contemporaneous Titus phase sites is with the consideration of the plain to decorated shcrd ratio (P/DR) in their ceramic assemblages. The P/DR expresses the proportions with which vessd surfaces are decorated as detected in plain and decorated sherd counts, and there are interesting spatial and temporal trends in the P/DR of specific Caddo ceramic assemblages and traditions in Northeast Texas (Perttula 2004:390). For instance, unlike contemporaneous Late Caddo groups in northwestern Louisiana and eastern Texas that made ceramics where large portions of vessel surfaces were decorated (particularly with the introduction or brushing on the bodies of utility ware jars), and the proportions of decorated sherds in an assemblage may be as much as 50-60% of all the sherds (with P/DR ratios of less than 1.0), McCurtain phase ceramics from the middle Red River area have P/DR ratios that arc greater than 40.0 (Pcrttula 2008:348-349). The proportion of decorated sherds in these assemblages is only about 2-3%, and it is dear that tht: ceramic tradition oftht: McCurtain phase Caddo was one comprised predominantly of plain vessels and large rim-decorated vessels with plain and expansive bodies. At the other extreme, in the early 18th century Deshaw site in the Angelina River basin in East Texas, the proportion of decorated sherds in the assemblage is an impressive 77% (dominated by brushed sherds), with a P/DR of 0.29 (see Fields 1995).
Of the sites listed in Tahle 25 that have some measure of ceramic stylistic relationships with the Tuinier Farm and Anglin sites, those with the most similar P/DR in their ceramic assemblages are Titus phase sites 20-30 km to the east-southeast (in the case of the Pilgrim's Pride and Underwood sites) and east-northeast (in the case of the Ear Spool site) in the Dig Cypress Creek basin (Tahle 26). The most divergent with respect to their P/DR values from the Stouts Creek sites are several Titus phase sites in the Dry Creek and Caney Creek localities in the upper Sabine River basin, which is ironic given that they are not geographically distant (see Figure 62) and are also stylistically much the same in the kinds of decorated wares, at least in respect to the prominence of Ripley Engraved vessels, the use of red slipping, and in the character or their principal utility wares (see Table 25), especially the use of neck handed and appliqued decorations on jars and the infrequent use of brushing on utility ware vessels.
Taking these two measures together (i.e., selected decorative categories and P/DR), it is apparent that the closest stylistic and cultural affiliations of the Stouts Creek Titus phase sites lie with other Titus phase communities within a 20-30 km radius to the north, south, and east-southeast. Even within these areas, however, there existed considerable local and intra-areal diversity in the character of the decorated utility ware and fine ware vessels made and used by different but socially interactive Titus phase communities.
In summary, the Tuinier Farm, Anglin, and R. A. Watkins sites are part of a very distinctive western Titus phase community that lived in the Post Oak Savannah in the Stouts Creek valley in the 16th century and much of the 17th century A.D; the Culpepper site (see Scurlock I 962) is another component in the community. Their unique an.:haeological nature rests in the character of their material culture: particularly with the fine ware and utility ware ceramics they made and used (among them Anglin Impressed, a new utility ware type), as well as the abundance of clay ear spools and figurine fragments found in domestic contexts, suggesting they were in regu Jar use within the community, and the use of marine shell Clements style car disks, also found in domes-Lie contexts. These particular kinds of artifacts arc rarely found at any other Caddo sites in Northeast Texas, much less other parts of the Caddo archaeological area, even in important mortuary or mound contexts, and speaks to the distinctive cu It ural practices and adaptive strategies employed by this Titus phase community to successfully thrive in the Post Oak Savannah of Northeast Texas. By all measures, this community thrived until \:a. A.D. 1700, alkr which they abandoned the area.

END NOTES
l. The historic artifacts found at the Tuinier Farm and Anglin sites date from the }l)th century and are not associated with the 16th and 17th century Caudo occupations. At Tuinier, a total of 25 cut nails (24 with heads and one nail shank), possibly Type 7 forms ( 1834-184 7) but more likely Type 8 cxamph:s ( 1820-189 I) (sec Wells 2000:335), had been c.:olle<:Led from the surface of the South miduen, suggesting a log structure stood in this area. At the Anglin site, there were two pieces of glazeu hand-made hrick fragments.

ACKNOWLEDGMENTS
First, Perttula would like to thank Lee Green for the opportunity to study the Late Caddo artifact collections from the Stouts Creek sites, and for working with us in 2007 at both the Anglin and Tuinicr Farm sites. Bo Nelson, Mark Walters, and Clay Newton assisted in the work done at these sites in 2007. Bo and Mark undertook the unenviable task of washing and sorting the large collectiom of artifacts.
Bob Brooks, Ann Early, Marlin Hawley, David Kelley, Daniel J. Prikryl, and Alan Skinner provided helpful leads and information on Caddo sites and other sites with clay llgurines. Pcrttula would also like to thank the various consultants that studied the faunal remains (LeeAnna Schniebs) and the mussel shells (Jesse Todd), as well as Elsbeth Dowd, George Morgan, and Beau Schriever of the University of Oklahoma for completing the electron microprobe study of five engraved sherds from the Tuinier Farm sire. Lee Green would like to thank his wife Jenny, Newman Bradford and Felicia Lewis for their help, and Jace & Rose Anglin for access to their property. Sandy Hannum prepared the maps used in this report, and Bo Nelson took the artifact photographs.

EXCAVATIONS
An examination of the north wall of the previous excavations indicated that there were buried midden deposits remaining in this area, and a pit feature was also observed in the trench wall just west of the midden The first two arbitrary levels (0-20 ern bs) of archaeological deposits in Unit 4 were s~..:reened through 1/4-inch mesh screen, hut when tht: midden deposits were encountered (20-43 em bs), the remainder of the unit fill was collected as fine-screen ( 1116-inch mesh ) samples; the unit was terminated at the base of the midden. This was done to enhance the recovery of charred plant remains, especially charred nutshells, in the hope of ohtained a sufficiently large sample of nutshells to submit them for standard radiocarbon assay at Beta Analytic, Inc. In Unit 5, the archaeological deposits above Feature I were scrt:ened through 1/4-inch mesh screen, as were the deposits from 20-40 em hs that were outside of the exposed pit feature. The fill of Feature I (20-53 em bs) was collected as a single fine-screen sample.
The archaeological deposits in this part of the Anglin site consist of a dark brown (lOYR 3/3) A-horizon sandy loam (zone I) that is between 40-43 em in thickm.:ss, beginning at the modern day ground surface (see Figure 63). These deposits overlie a dark ycllowi~h-brown (IOYR 4/4) sandy loam E-horizon (zone 2, approximately 5-lO em thick) and a strong hrown (7.5YR 4/6) clay B-horizon (zone 3). The B-horiwn is encountered between ca. 45-50 em bs. In one area of the trench profile, the A-horizon has an organically enriched and black (lOYR 2/l) midden deposit (zone 4) that is a maximum of 23 em (20-43 em bs) in thickness; the E-horizon (zone 2) underlies the midden. Feature I apparently originates in the middle part of the zone J A-horizon, at approximately the same depth (18 em bs) as the top of the zone 4 midden deposits (sec Figure 63). This pit feature is approximately 37 em in diameter and has straight walls and a rounded bottom. The pit fill is a very dark grayish-brown (IOYR 3/2) sandy loam with charcoal and bone flecking and small burned clay nodules.
The fact that top of Feature I is at virtually the same depth as the top of a ca. 23 em thick midden deposit suggests that the pit feature may have been dug about the time that the accumulation of the Late Caddo midden deposits ceased. The source of the A-horizon sediments above the midden and Feature I is not known, but may be the product of bioturbation and natural soil accumulation after the Anglin site was abandoned by Caddo peoples in the 17th century A.D.

ARTIFACTS
Prehistoric artifacts are abundant in the two small units excavated in 2008 at the Anglin site, particularly pieces of burned clay/daub and animal bone in the Unit 4 midden deposits (Table 27). By unit, the artifact density ranges from 752 (Unit 5) to 4594 (Unit 4) artifacts per m 2 in these cxcavations. Burned clay/daub and animal hone are also relatively abundant in the fi ll of Feature I. *charred plant remains-wood charcoal and charred nut,hclls-arc nut included in the artifact totals as they have not been quantified.
The ceramic sherds (n= l20) from the Anglin site arc from fine ware and utility ware vessels tempered uniformly with grog. A small percentage also have crushed and burned bone (6.3%) or hematite/ferruginous sandstone (7.8c7o) added to the clay paste along with the grog temper. A I"ew other sherds (6.3%) have charred organic materials in the paste-indicative of incomplete firing that failed to completely combust these materials in the clay paste-and 4.7% of the vessel sherds have a sandy paste. These latter sherds suggest that occasionally a Caddo potter at the Anglin site chose to use a naturally sandy clay for vessel manufacture.
The animal bone and mussel shell pieces arc refuse from hunting and gathering and food processing activities that accumulated in and near the Anglin site's midden deposits. The majority of these pieces are small, burned fragments (especially the animal bone), although white-tailed deer bone is present in tht: collection.

CONCLUSIONS
Limited investigations in October 2008 at the Anglin site (41HP240) documented midden rt:mnants at one end of a pre-e.x.isling lrt:nch profik, as wdl as a small pil feature (Feature I). Animal hones, burned clay/ dauh, ceramic vessel sherds, and lithic debris from chipped stone roo! manufacture, are abundant in these deposits. These remains arc from a post-AD. 1500 Caddo habitation. Clearly the archaeological potential of the Anglin site's archaeological record has not heen exhausted. Hopefully with the sorting and analysis of the recovered plant remains, samples of charred nutshells can then be submilled to Bela Analytic, Inc. for radiocarbon dating lo establish the absolute age range of the Anglin site midden deposits.