MISSISSIPPIAN OCCUPATION OF PHILLIPS COUNTY, ARKANSAS,

AND COAHOMA COUNTY, MISSISSIPPI

PREFACE

This article originated in a contract report prepared for the Memphis District of the USA Corps of Engineers by Garrow and Associates, Inc. Limited test excavations were undertaken at the Ellis Mound site in connection with the development of a slackwater harbor south of Helena. I worked on the excavations, directed by Doug Prescott and David Crampton, and, some years later, analyzed the materials recovered by this testing. In light of recent work on Mississippian sites both to the north (lower St. Francis basin/Barrett and Kent phases) and to the south (Arkansas River/Quapaw phase/Menard complex), I thought it would be appropriate to expand the discussion to include other Mississippian components near Ellis mound. I presented a summary of this research to the 1996 MidSouth Archaeological Conference.

The Ellis mound excavations provided a sequence of radiocarbon-dated construction phases and associated artifactual evidence of ceramics and structural technology. This excavation is almost the only modern archaeological work undertaken on the Mississippian sites of Phillips County, despite there being a number of important sites in the county. Some additional information for the area can be derived from pothunted collections, late nineteenth and early twentieth century sources, the Lower Mississippi Survey of the mid twentieth century, amateur archaeologists of the Arkansas Archeological Society, and cultural resource management surveys. The records and collections of the Arkansas Archeological Survey in Pine Bluff and Fayetteville and the U.S. Forest Service in Russellville were also reviewed for this study.

This report attempts several scales of comparison. The discussion begins with a recounting of Mississippian research in Coahoma County, Mississippi, which is intended as an introduction to the problem of Mississippian research in this portion of the Mississippi Valley from the perspective of the adjacent county your author is most familiar with. A brief introduction to the history of research in Phillips County and surrounding parts of east Arkansas up to the time of the Ellis Mound excavations in 1989 is followed by a summary of the results of that excavation based on a chapter from a Memphis District, Corps of Engineers (COE) contract report (Childress et al. 1995). The scope of the discussion is then widened to consider sites contemporary with Ellis in an about 50 mile radius around Ellis, an area that extends from Chucalissa and Parkin in the north to Menard, Powell Bayou, and Winterville in the south. This section is also drawn largely from the contract report. After this excursion through the fifteenth century, the focus returns to Phillips County, with a recounting of all the other known Mississippian sites in the county, including more detailed information on sites mentioned in the discussion of the origin of the Old Town phase concept. The definition of Mississippian components on these sites was almost always based on the presence of shell tempered ceramics. The article concludes with a discussion of just how little we know about the area after a century of research and presents suggestions for future research based on the State Plan for the Conservation of Archaeological Resources (Davis 1982). 

INTRODUCTION I: A COAHOMA COUNTY, MISSISSIPPI, PERSPECTIVE ON MISSISSIPPIAN RESEARCH

Writing from the vantage-point of Mississippi, I feel it is appropriate to preface this discussion of the Mississippi period occupation of Phillips County, Arkansas, with an overview of the better-developed archaeology of Coahoma County, Mississippi. Phillips and Coahoma counties have nearly coterminous northern and southern boundaries and in many ways what we know of the late prehistoric occupations are similar. However, while Phillips County has seen very little modern investigation, Coahoma County has seen a great deal of research conducted largely under the direction of John Connaway of the Mississippi Department of Archives and History (MDAH) Clarksdale office. This summary of the research in and prehistory of Coahoma County is presented for the Arkansas readers who may not be familiar with the highly relevant Mississippi literature. It can also be taken as a chronology of how research has been conducted in the county by those interested in contrasting the processes of developing local archaeological sequences. 

Phillips (1970:939) defined a single phase, Parchman, encompassing Late Mississippi period occupation in northern Coahoma, southern Tunica, and northern Quitman counties, Mississippi. The Parchman phase appears to have been based on geography as much as ceramic type frequency; Phillips noted that there was little to differentiate it from the Kent phase to the north, the Old Town phase to the west, and the Hushpuckena and Oliver phases to the south. These phase definitions were based on sherd collections made by the Lower Mississippi Archaeological Survey (LMS; Phillips et al. 1951), and so were limited by the original ceramic typology which was simpler than the later type-variety system (Phillips 1970). The two system (types and type-variety) are still used interchangeably by area archaeologists, although analytic emphasis has now shifted to modes, vessel forms, and ceramic sets. 

Based on my own experience in the region, I concur with Phillip's assessment of Late Mississippi period ceramic homogeneity across much of the St. Francis-Sunflower portion of the Delta. In the following analysis, I will argue that Phillips County archaeology has close similarities to these regions as well. To the south of the area with which I am most familiar, Brain (1988:265), in reference to that part of Arkansas south of the Arkansas River, notes widespread similarities in domestic ceramics of the upper bayous Bartholomew and Macon and Coahoma County.

Ian Brown (1979) has examined, in one way or another, 37 previously-reported Mississippian cultural components in Coahoma County. He focused on Protohistoric components, visiting 17 sites and examining private collections from the area. I have also examined collections and records from these and other sites in an evaluation of the ceramic content and site plans of the Parchman phase (Starr 1984), but much work remains. We do see that Coahoma County has a wide range of site types, ranging from multi-mound groups to hamlets and farmsteads. Apparently these span the entire millennia from A.D. 750 (the beginning of the development of Mississippian traits in the Late Woodland Coles Creek period) to 1750 (the approximate end of significant occupation by peoples native to the Delta).

In 1988, Jeffery Brain published Tunica Archaeology, wherein he reevaluates Phillip's Hushpuckena, Parchman, and Oliver phases, suggesting a chronology and diagnostic ceramics. In the process, he redefines the Old Town and Quapaw (I and II) phases for east Arkansas. The basis of the chronological assignment of the surface collected materials is unstated, and could be nothing but shaky given the dearth of investigations since the original LMS surface collections were made, but at any rate Brain's work remains the most thorough attempt to integrate a large number of components throughout the Yazoo and Arkansas portions of the Delta in century-interval phases. He includes many illustrations of materials from the LMS collections and assigns a number of sites to these new phases. Many of the sites are almost completely unknown and their inclusion in phases should be contested. Brain (1988:265) recognizes that the "utility of [widely traded mortuary ceramics] for ethnic interpretations is practically nil" and that there existed a wide field of similarity, "a general background commonly shared by many groups", formed by the domestic (typically coarse shell tempered) midden ceramic types Barton Incised (rectilinear incised), Winterville Incised (curvilinear incised), Parkin Punctated (punctations only), and Owens Punctated (combined punctations and incising). Surface collections from Carson, which has Cahokia horizon through Choctaw components, figure large in his reconstructions, as do the Oliver site materials, but this latter site also represents a long occupation span. To appreciate Brain's proposed reconstruction, it is necessary to be aware of two underlying assumptions: 1) that the De Soto entrada crossed the Mississippi around what is now the Montezuma Bend (Friars Point vicinity), and 2) that the people of the province of Quiz Quiz were the ancestors of the people the French some six or eight generations later knew as the Tunica nation. As in much historic or biblical archaeology, written documents are given priority and archaeological remains are required to fit into the framework derived from the sources.

The ca. A.D. 1400 Hushpuckena I phase, opposite the mouth of the Arkansas and east of the Sunflower River, is marked by Owens Punctated with motifs similar to Barton Incised, Avenue Polychrome/Nodena Red and White, and Winterville Incised var. Ranch (imbrications or fish-scale pattern). Hushpuckena II, ca. A.D. 1500, Brain's province of QuizQuiz of the DeSoto entrada, is identified based on wide-line varieties of Winterville Incised, Old Town Red, and curvillinear incised/tooled punctated varieties of Owens Incised. Hushpuckena II is defined as contemporary with "the classic late prehistoric phases of Old Town, Kent, Walls, Parkin, and Nodena [which] were in full flower" (Brain 1988:269), but which he believes to have been most closely related to the Hog Lake and Tillar complexes of southeast Arkansas, a relationship necessary to interpret the people of QuizQuiz as Tunicans. According to Brain's reconstruction, by the sixteenth century, Hushpuckena people were migrating to the south, to be followed by, or contemporary with, immigration from south Arkansas across the Mississippi into the Coahoma County territory of the Hushpuckena II phase. The implied behavior in the term of socio-cultural anthropology is evidently common bride-exchange across the River, by people of related or allied villages, or high residential mobility of households.

Brain's re-defined ca. A.D. 1600 Parchman phase has indications of both local continuity and major population movements. The demographic changes are seen to be echoed in the Parchman phase's neighbor across the river, the Quapaw I phase which "may represent the dissolution and displacement of the Old Town, Kent, and perhaps Walls phases…the remnant populations of which moved to the Arkansas and mixed with local peoples" (Brain 1988:272). The ceramic varieties recognized as diagnostic of the Protohistoric Parchman phase are seen to be closely related to those of southeast Arkansas' Tillar complex, but "similar decorative ideas…found in eastern Arkansas" are "aberrant" and are referred to as "Tunican" rather than "Tunica", without a specific (Koroa?, Brain 1988:285) ethnic identification indicated. The Tunican mode (punctations bounding incised fields at vessel inflection points) is the primary marker for the Parchman phase, along with certain varieties of Winterville Incised and Barton Incised. Finally, the post-Tunica, ca. 1700 Oliver phase was marked by a new lithic complex, possibly identifiable with Dhegiha Siouan groups such as the Quapaw, but with pottery derived from the Quapaw I phase, the pottery-making population of which was still operating in the long Yazoo tradition (Brain 1988:278-279). The mechanism sometimes proposed for this seemingly unlikely scenario is an aggressor nation, the ancestral Quapaw, immigrants from the Ohio confluence and northeast Arkansas, decimating the Coahoma County area villages' warriors and taking over the surviving women and towns as their own. Other equally unverifiable scenarios can be proposed. For instance, the new lithic complex could be due to young men from an area with a shortage of women seeking wives of down-stream, matrilocal allies or to the spread of a sophisticated buffalo hunting technology without much change in the donor and recipient peoples' permanent residences. This last explanation is growing in popularity as it becomes more widely recognized that the endscraper at least is widely distributed from the eastern Plains and Missouri and Ohio confluences south not just among supposed Siouan-speakers, but among the Natchez and Chickasaw as well. Marvin Jeter (1990) has pointed out many of these same objections to Brain's equation of QuizQuiz with some Tunican speakers in his review of Tunica Archaeology.

These three review efforts (Brown 1979, Starr 1984, Brain 1988) have resulted in small new surface collections from Coahoma County at best. Unfortunately, there has been little focused surveying for new sites or testing to validate suggested dating of recorded sites. However, in addition to the 37 components reported in Coahoma County by Brown (1979), another 59 Mississippian components can be found in the state site file, for a total of 96 confirmed or likely Mississippian components in Coahoma County as of January, 1997, out of 253 numbers assigned, or 38%. In Phillips County, Arkansas, the figure is 51 out of 339 total sites, or 15%. About 2000 acres have been surveyed in Coahoma County in the course of 51 cultural resources surveys dating from 1977 to 1996 (the figure is upwards of 4000 acres in Phillips County, but often at fairly low intensity). At least 34 of the Coahoma County surveys failed to report cultural resources, but seven of the surveys have recorded 14 probable Mississippian components. The majority of the county's sites were reported by the LMS or the Mississippi Archaeological Survey (Sam McGahey, John Connaway, and Sam Brookes) beginning in 1968, but some were reported by members of the Mississippi Archaeological Association.

Based on the files of the MDAH as cited below, it appears that cultural resource management (CRM) is contributing a little to the archaeology of the Mississippi period in Coahoma County. With 2000 acres surveyed and 14 Mississippian manifestations identified (1 for every 142 acres surveyed), only two sites have been investigated, the McNight village and the Sunflower Landing mounds. CRM having produced 14 of the 96 total Mississippian occupations indicates that prior surveys, the MDAH, and a few private individuals reported about 85% of these sites. Much of the CRM archaeology in Coahoma County has been associated with the ongoing four-laneing of U.S. Highway 61.

In 1983, Mississippi Department of Transportation (MDOT) survey for the proposed widening of U.S. Highway 61 recorded 22CO683, which was believed to be ineligible for the National Register due to lack of integrity. In 1987, John Connaway, in the course of surveying borrow pits, encountered buried wall trench and midden remains just north of the National Register boundary of the Salomon site. He recommended moving the pit location to avoid the deposits and no pit was dug there. In 1991, additional MDOT survey for the widening of Highway 61 reported a Woodland and Mississippian site (22CO731) on the Yazoo Pass north of Salomon. The site was not considered significant. In 1993, Connaway surveyed a borrow pit north of the Yazoo Pass, along Highway 61, between Salomon and Barbee, and found scattered daub. He equivocated in his recommendations, saying that either the area should be stripped with the expectation that quite a few houses might be uncovered, which he did not want to have to deal with single-handedly in a rushed fashion. The borrow pit was moved to avoid this area. In 1994, a 250 acre Farmers Home Administration (FmHA) foreclosure resulted in minimal investigation of one of the LMS sites, Hull Brake (22CO515), which was written off as not eligible for the National Register. Finally, in 1996, MDOT resurvey of part of the Highway 61 widening resulted in the recording of seven small sites with shell tempered ceramics, triangular arrows, and/or daub (22CO745, 748, 749, 750, 755, 759, and 761). Of these sites, 22CO670, 749, and 764, south of Clarksdale, are to be monitored during grading.

As Weinstein et al. (1985) note "every once in a while a small-scale cultural resources survey locates something completely unexpected." Such was the case when Coastal Environments, Inc. (CEI) located the Late Mississippi period Sunflower Landing mound group (22CO713) between segments of the 1850s and 1890s levees during a berm survey near Rena Lara. Weinstein (1985:7-1), following Brain's (1984) proposal of a southern crossing of the Mississippi, once proposed that the site was one of the towns of Quiz Quiz occupied by the DeSoto expedition, but he has since reevaluated this hard-to-support position (cite his O. chapter). Much of the Sunflower Landing occupation area has probably been removed for fill and part of the large mound has been spread out by farming, but Mound A still measures about 40 by 30 m, with the sides oriented to the cardinal directions and the summit being 4 m high (Weinstein et al. 1985:6-35). Little cultural material was recovered by CEI due to poor visibility and siltation, but in addition to Woodland types, Mississippi Plain (N=40), Addis Plain (N=7), Bell Plain (N=5), Barton Incised (N=10), daub, cores, flakes, and bone were collected from the surface and limited testing (Weinstein et al. 1985:6-40).

This work shows that CRM survey does have some utility in gathering data that can be used in more specifically directed research. It would have been much more valuable, however, if there had not been such readiness on the part of surveyors to discount the potential of small, cultivated sites. Somehow, only 2000 acres surveyed by CRM efforts seem rather low for 20 years of work. This can perhaps be accounted for by the fact that the Coahoma County economy is still dependent to a great degree on cotton cultivation. Indeed, it is the center of one of the world's premiere cotton-growing districts and the presence of so much high, loamy land should instantly alert us to the fact that it also saw large Mississippian populations. Phillips County, while it has some prime alluvial soil, particularly around Old Town Lake, has a higher percentage of low-lying, heavy backswamp land as well as braided stream surfaces, an important landform not present in Coahoma County. It has also seen somewhat more industrial development than Coahoma County, indeed most CRM work in Phillips County has been done in association with such development.

Early Mississippi period components are fairly well documented in Coahoma County, although they are perhaps more common than has been recognized. Early sites are dominated by coarse shell tempered plain jars and bowls, but these materials do not stand out in surface collections from large, multicomponent sites and the small villages and farmsteads that appear to characterize the period are probably underrepresented in the database. White Crescent Quarry/Burlington chert blade cores, flake tools, and debitage from the St. Louis, Missouri, area may be a significant marker for the ca. A.D. 1100-1200 era, although the material is also associated with Marksville occupations (Johnson 1987). Northern shell tempered pottery types like Cahokia Cordmarked, Ramey Incised (broad-line incising in complicated patterns on a distinctive high-shouldered vessel form), and Varney Red (coarse shell tempered, hematite slipped) occur but are scarce. At any rate, a great number of Late Baytown and/or Coles Creek period, Plum Bayou culture-related components are known from the area. Many of the traits (other than shell tempering) that are commonly found in definitions of the Mississippian culture began during the Coles Creek period of the Lower Valley. Coahoma County Late Woodland components ca. 750-900 have many of these characteristics, such as rectangular wall-trench houses and platform mounds (Connaway 1981:32-35).

A date of 875+85 (UGa-280) from a pit at Barner (22CO542), a village site on Big Creek tested by Sam Brookes, appears to place the transition from Woodland to emergent Mississippian in the county (Connaway 1981:82). The site also had a 15 by 16 foot wall trench house surrounded by pits, most or all attributable to the Woodland period. Barner does have a small amount of shell tempered pottery and a small, plowed-down mound where human bone and flat disc shell beads have been observed (John Connaway, personal communication 1997). The white Missouri chert mentioned as an Early Mississippi period diagnostic is generally believed to have been used for drilling shell beads, particularly around Cahokia, where it originates.

Craig (22CO566) was another small Early Mississippi period site salvaged by John Connaway and Sam McGahey in 1969; the following description is based on their notes on file at the MDAH Clarksdale office. This two acre village on Barkley Bayou had a low sandy mound with a heavy daub scatter. There were three other daub scatters nearby, and the shell tempered pottery appeared to be most concentrated near the daub. No wall trenches were identified, but there were some suspicious linear stains and several unaligned postholes and pits, one of which produced corn. The soil was very dark, hindering feature interpretation. Terminal Woodland ceramics included Mazique Incised (grog tempered with rectilinear incision) and Soloman Brushed (grog tempered with rough finish from wiping with grass or a coarse fiber brush); typical Early Mississippi period types identified were Cahokia Cordmarked, Ramey Incised, and Mississippi Plain with Powell Plain-like rim, all Cahokia-area types with occasional Lower Valley examples or locally-made analogs. Craig also produced some typically later types: Bell Plain (fine shell and grog tempered), Avenue Polychrome (red and white painted with a black stain on unslipped parts), Barton Incised, Parkin Punctated, and Ranch Incised (curvillinear incised now subsumed under Winterville Incised) as well as a petrified wood chisel. The use of petrified wood for Late Mississippi period chisels is a consistent low-frequency trait of the region.

Hasty and limited salvage at the Bobo Site (22CO535) resulted in the recording of four house patterns out of 80-100 uncovered at this large village site with an 8 foot high mound (Connaway 1981:36-40). Bobo was a 6-8 acre Coles Creek and Mississippi period midden on the Sunflower River. Overlapping and rebuilding of structures were noted, along with internal features (roof support and bench/bed posts, hearths, storage/refuse pits, and graves). Structures measured from 15 to 32 foot on a side. The larger, seemingly contemporary, structures were oriented north-south. Rows of thatched, wattle-and-daub, wall-trench houses around a plaza west of the mound were indicated, but not documented. Artifacts recovered included Mississippi Plain, Winterville Incised, and an oval-shaped Mill Creek? chert hoe. Two dates, here shown uncalibrated, were run from the site: A.D. 890+90 (UGa-560) on a large, deep house post under the mound and A.D. 1275+100 (UGa-559) on corn from a pit outside overlapping structures. Occupation for several centuries of the late Coles Creek and Early Mississippi periods is indicated.

In what is probably the most important CRM work to have import for the Mississippian archaeology of Coahoma County, Panamerican Consultants, Inc., has recently conducted excavations at the McKnight village site (22CO560) adjacent to Barbee Cemetery Mound (22CO510), a presumed Marksville or Coles Creek period conical mound along Highway 61. I was fortunate to be able to volunteer some time to this work and was struck by the large number and size of the Woodland pits on the site. Two wall trench structures and associated Mississippian pit features were encountered along with a Mississippian assemblage dominated by thin, plain, coarse shell tempered ceramics similar to Powell Plain and Mississippi Plain var. Coker. One of the houses was oriented with the sides to the cardinal directions and the other with the corners to these points. Both were about seven meters on a side and roughly square. There was a small pit in the corner of one house and a large bell-shaped pit near the same structure. The Powell Plain-like medium jar has the thin body, high shoulder, and everted rim typical of the Cahokia area. The only other noteworthy Mississippian sherds were a small loop handle and two Barton Incised sherds, one with cross hatching and one with oblique, opposing parallel lines. A small, flaring bit Mill Creek hoe was also recovered from the surface (Shawn Chapman and Richard Walling, personal communication 1997).

This site is particularly interesting for a site discovery problem it poses: while it was recognized from the earliest stages of planning to four-lane Highway 61 that this site might be significant, only about half of the visits to the site for surface collecting and shovel testing have produced Mississippian artifacts, despite the fact that very good sub-surface contexts were revealed when the plowzone was stripped. Indeed, from the days of the LMS, the Mississippian component visible on the surface at Barbee-McKnight has been archaeologically ephemeral, but the features 35 cm under the surface of the cotton field were anything but ephemeral.

A final Early and Middle Mississippi period component that has seem some work is the Carson-Stovall-Montgomery complex, one of the largest and most significant sites in this portion of the Mississippi Valley (Figure 1). It was probably already an old and very important center for the hamlets and villages I have just described when the Mississippianization of the region began. The site complex extends for a mile near the modern Mississippi River channel. While most of the material collected from the site dates to the Late Mississippi period, it has a few sherds of Cahokia Cordmarked and Kimmwick Fabric Impressed (coarse shell tempered pans with imprints of coarse cloth on the exterior) as well as a full range of Woodland components. Apparently, part of the site is quite late, producing Owens Punctated, Wallace Incised (broad lines, generally on helmet-shaped bowls), Chickachae Combed (fine paste curvilinear and rectilinear decorated pottery made by the Choctaws), Fatherland Incised (fine paste curvilinear incised pottery of the Natchezean people), engraved, complicated stamped, and painted ceramics (Brown 1979); but historic trade goods have not been reported from the site, to my knowledge. Col. Norris of the Bureau of American Ethnology reported 85 mounds in 1884 even though the plantation had already been in cultivation and subject to looting for a generation or more (Thomas 1894), today five large mounds remain at this National Register site.

In 1951 the Memphis Geological and Archaeological Society dug 30 inches down in a ten foot square in the westernmost area of the Carson complex. This is a village area just inside a fortification line and near the largest platform mound (Beaudoin 1952). They encountered a large refuse-filled pit, part of a house floor, and burials "within or near the circles of dark earth indicated on the Smithsonian Institution plat as low mounds" (Beaudoin 1952:10). There were 12 bundle burials and 3 flesh inhumations, only one of which (still?) had a pot with it, and also a shell bead at its hand. A lithic collection had triangular points, ground stone and chipped celts, scrapers, hammerstones, grooved pumice, and a fine-grained bluish diorite discoidal or chunkey stone. Some of the lithic material was " a much finer quality of flint or novaculite" (Beaudoin 1952:14), perhaps a reference to Burlington chert which occurs in relative abundance at Carson. Beaudoin offers a late Middle Mississippian date for the area of the testing, and while he does not describe the ceramics, he states that this village may have moved to Old Town Lake, in what is now Phillips County, Arkansas, with a channel shift, "for the similarity of the potsherds indicates both sites to have been occupied by the same people (Beaudoin 1952:10)."

The Clover Hill site (22CO625) is a small Late Mississippi period village on a high, sandy natural levee (Connaway 1981:45-49). In 1973, the largest of several daub scatters on the site was excavated, resulting in the documentation of the floor of a large wall-trench house. Smooth-surfaced daub in contact with the floor, along with the lack of wall poles in direct contact with the floor, indicates that the structure may have had interior plaster. A small pit within the structure contained corn, beans, and persimmon seeds. The structure had a large cylindrical hearth at the center, 1.5 foot in diameter and 2 foot deep, with prepared clay walls and rim, and, possibly, a clay dome indicating that it was used as an oven. Carbon from three internal green ash posts were used for radiocarbon dating, resulting in dates of A.D. 1510+60 (UGa-1889), 1360+65 (UGa-1890), and 1525+55 (UGa-1891). A number of reconstructable vessels were recovered near the excavated structure: a horizontally striped Avenue Polychrome bottle; a deep nicked-rim bowl; two plain, short-necked bottles or jugs; and a standard jar.

Two Coahoma County Late Mississippian sites have a type of structure that is, so far as I know, unique in the archaeological record of our area, in that they appear to be raised on piles. Such buildings are, however, described in Garcilaso "The Inca" de la Vega's oral history/novelization of the De Soto entrada's time in Aminoya, which was likely in the lower White/Arkansas river basin. French reports of the Mississippi Valley also appear to describe such structures. Baird (1980:11) mentions that around 1700, the Quapaw built platform 15 to 20 foot high to sleep on in summer, to escape the mosquitoes and to gain some breeze. Marquette may have also observed a similar structure, as he described scaffolds, with floors of poles and covering of bark, which provided protection from mosquitoes when a fire was built underneath, and shade and breeze (Thwaites 1959:146).

Wilsford (22CO516), between the large Parchman and Salomon mound groups, had a small rectangular mound (large enough to be noticed by the LMS), daub-covered rises, and very sparse surface material. Like the Clover Hill work, the Wilsford excavations were in response to land leveling. The excavation of this site has been reported in detail (Connaway 1984). Daub was the primary material recovered. It is thoughtfully and thoroughly covered by Connaway's daub typology. The Wilsford pile structures show evidence of rebuilding in the same location. They have wall trenches, and within them, a closely spaced grid of support posts. Each had a deep, ramped pit for a large centerpost, and probably, a surrounding porch or balcony and entryway. A Nodena Red and White bottle fragment and an Addis Plain (a southern analog of Bell Plain, generally with little or no shell) carinated flaring-rim bowl were recovered from one centerpost pit. Other ceramics from the site include a typical Barton Incised jar with curving shoulder, lug, and line-filled triangles. Dates on the Wilsford structures were 560+75 BP (UGa-4713, House 1 wall trench post), 720+55 BP (UGa-285, House 1 centerpost), 485+60 BP (UGa-283, House 1 interior support post), 525+60 BP (UGa-281, House 2 wall trench post), and 375+90 BP (UGa-4714, House 3 interior support post) (Connaway 1984:162). The later dates are on smaller posts, and probably are fairly close to the construction date, while the earlier centerpost date can be accounted for by the large size of this timber. The main occupation date for the site appears to be around A.D. 1425.

The smaller, presumably similarly elevated structure at the Hays farmstead site (22CO612) produced a date of A.D. 1705+60 (UGa-279) on the centerpost (Connaway 1981:84). The one acre site had only been disked when it was excavated by the MDAH in 1969 and it was scheduled for land-leveling. The site had two areas: a concentration of debitage and unfinished pebble-core triangular points and a concentration of daub. The structures uncovered had been rebuilt at least two times on the same spot. One construction appeared to be a standard wall trench building with internal posts indicative of beds around the walls. The other construction had single set wall posts and a large number of interior posts, a ramped center post placement pit, and a set of short, close-spaced wall trenches perpendicular to the southeast wall probably part of the stair or ladder. The five foot deep, tapering center pit had daub with grapevine impressions, split woven cane impressions, grass temper, and smooth surfaces. It also contained cypress charcoal, corn, and persimmon seeds. Pecan and white oak were also apparently used in the buildings. Ceramics were mostly Mississippi Plain with some Barton Incised and a few small painted sherds, possibly Carson Red on Buff (a late type with designs made by painted and plain areas). The only vessel fragment was a small wide-mouthed or hooded bottle. Lithics included, by the 1969 tabulation, microliths, end-scrapers, a hoe fragment, triangular Madison points, and a piece of ground sandstone (notes on file, MDAH Clarksdale).

Another very significant site is Humber-McWilliams, a early Protohistoric village and ceremonial center with perhaps several thousand graves that have been extensively pothunted. The two-mile-long site lies along a natural levee adjacent to the modern Mississippi. Brown (1977:17) describes the settlement thus: "houses seem to have been…concentrated along the…easternmost portion of the ridge edge, just below where it slopes down. Burials were placed both in this area and along the eastern slope, primarily the latter. The ridge to the west was, with but few exceptions, unoccupied. The depression between the two ridges may have been a [stone-tool using] workshop. Presumably the land to the west of this linear settlement was cultivated." Several mounds were removed for levee construction in 1929, opening the site to pothunting that was almost continuous until the landowner began to put an end to it in the 1970s, due to equipment falling into unfilled potholes. One of the collectors kept detailed notes on his work, which are on file at the Clarksdale MDAH office. The site produced spectacular "dog pots" and other elements of the popularly defined Quapaw ceramic complex, including Bell Plain teapots, Old Town Red bottles, and bottles with red and white interlocking scrolls and horizontal bands, many examples of which are on display at the Cottonlandia Museum in Greenwood, Mississippi, and at the Carnegie Public Library in Clarksdale. A small area with meter-deep midden was tested in a year-long program sponsored by Cottonlandia Museum (Tesar 1975). The site was evidently occupied intensively over an extended period, as there are many examples of overlapping sequential buildings. A five meter square rectangular structure made with singly set posts, packed gumbo floor, straw daub, and large southeast facing entrance was recorded. It perhaps sat on a house mound. Parts of 3-to-4 m diameter arcs of postholes were also noted, along with fire pits and dumps containing corn, beans, persimmon, deer, and fish remains.

Tesar excavated 33 pits with the remains of about 46 people, both primary inhumations and bundle burials, and recovered a number of associated vessels. Nancy Ross-Stalling (Mitchell 1977) examined some of these for her thesis work at the University of Mississippi, refuting many of the assumptions Tesar made about his finds. These are the only autopsies of a late population in the county available, although Ross-Stallings has also examined some cemetery collections from Tunica County to the north and the Oliver site. The Humber people were in the rather poor state of health typical for the era. Anemia, bone abscesses, dental caries, and osteoporosis, all evidence of nutritional deficiencies and other stresses such as bacterial infections, were prevalent. Indeed, given the apparent high rate of birth defects such as spinal bifida, they may have been more malnourished than we generally think about for the period. Life expectancy at birth was 22 years, and at age 6, 35 years for males and 33 for females, but a few of the Humber people lived into their 70s (Mitchell 1977:115). The statures of the men ranged from 5'3" to 5'9" and the women, 5'2" to 5'5". The onset of arthritis in jaws, limbs, and the spine was at age 35 to 40, and affected 40% of the men and 66% of the women, perhaps indicating that the women were the farmers, although the sample was too small to have much confidence in the significance of the finding (Mitchell 1977:110). The Humber people practiced cradleboard cranial deformation and had the associated Wormian bones, as well as some instances of the Inca bone (Mitchell 1977:108).

Two pieces of European sheet brass (copper-tin-zinc alloy) have been recovered from graves at the site; the first is a small perforated strip with preserved cord from a subadult bundle in a multiple bundle grave, the second is larger and came from the forehead of an adult male skull. It has native-made punctations along the margins of the rectangular strip and has embossed nested squares or rectangles on it (Leader 1990). This is a very "Muskogean" looking motif, and a similar variant on the motif occurs on a late jar at Oliver. A source, admittedly farflung, for such material is the Alabama River and Mobile delta, which came into contact with coasting European vessels on a sustained scale before the de Soto entrada. The majority of these landings for water, timber, corn, rape, and slaving, as well as shipwrecks, are not documented in the European records of the sixteenth century.

The Oliver site, excavated in 1901-1902 (Peabody 1904), 1941 (Phillips et al. 1951), and 1991 (Starr 1992), has produced more abundant evidence of Middle and Late Mississippi and Protohistoric period occupations. The site had extensive Woodland occupation and probably Early Mississippi components as well. Most of the Oliver site has been destroyed. First estimates of the European materials attributed them to the de Soto entrada, later they were thought to date around 1680-1720, when the French established relations with the Quapaw. Current interpretation attributes the trade goods to Florida and Mexican Spanish sources, brought inland by natives in the still-active native exchange system during the sixteenth and seventeenth centuries (Starr 1994). This is supported by the intermingling of glass and brass beads with marine shell beads in several necklaces. A Mexican-Southwestern United States source is perhaps supported by the presence at Oliver of a small child with a necklace of the tiny turquoise disk beads very much like those common in the Southwest, but it should be remembered that Oliver lies opposite the mouth of the Arkansas River, which heads in the turquoise-producing areas of New Mexico. Belmont (1961) reviewed Peabody's notes and established that the main mound had eight strata, that it began as a Woodland conical mound, evolved into a platform mound, and finally, was reduced from this rectangular shape when it served as the resting place of a great number of early contact period bundle burials.

The 1991 land leveling salvage recovered 40 complete or reconstructable vessels, some in sets around hearths, in addition to about 78 whole or largely complete pots dug at the turn of the century. Most of Peabody's collection consisted of plain helmet, hemispherical, and rim effigy bowls, with some examples of Barton Incised, Winterville Incised, Parkin Punctated, and Old Town Red. Most of the 1991 salvage vessels were also plain, but Old Town Red (bowl), Winterville Incised (very large jar), and Owens Punctated (small jar) vessels were recovered. Vessel form include a wide range of jar sizes (4 to 46 liters) and small (1 to 3 liter) helmet bowls; flaring rim, hemispherical, and scallop rim bowls; a hooded bottle; and a small vessel that appears to be an imitation of ca. A.D. 1700 brass kettles (Starr 1992). There may be a second "colonoware" vessel from the Peabody collection, an imitation of the skillets exported to Louisiana by 1701 (Brain 1979:140, 181), although this is a tentative identification, as there are some dippers from the site, but these are small and globular. The hooded bottle came from the grave of two youths, one beheaded and apparently killed by a Madison-tipped arrow, accompanied by European material and with a Nodena arrow being found in their grave as well. The historic materials are generally similar to those Ford (1961) reported from the Menard excavations: occasional glass beads and cut copper alloys (tubular beads, cut and rolled bracelets and coils, and bells). The Oliver ceramics also somewhat resemble those of the Lower Arkansas Protohistoric-contact period sites, although they seem to show greater affinity with the preceding Parchman, Kent, and Walls phases. A distinctive tool-kit from the site has given origin to the term "Oliver lithic complex" (Nodena arrow points, large triangular knives, endscrapers, "pipe" drills, and, I think we can add, catlinite pipe bowls for calumets). John Connaway is currently editing a comprehensive volume on the Oliver site that will include a detailed description of the ceramics, lithics, features, and a physical anthropological study. 

Besides the Oliver assemblage, other evidence indicates that occupation of Coahoma County extended into the Protohistoric or Contact period. On a surface collecting visit to the already discussed Clover Hill site, I found a blue glass bead. Marvin Smith (personal communication 1992) has examined the bead and places it in the Ichtucknee Plain/Early Blue/IIa40 type, a common French and English trade item dating from ca. 1575 into the eighteenth century. Based on the morphology of this single specimen, it probably in the first half of this interval. Bramlett (22CO551) in extreme southern Coahoma County, has produced Wallace Incised sherds (2, versus 67 Winterville Incised, 34 Barton Incised, 12 Owens Punctated, 6 Leland Incised, and 4 Parkin Punctated), Madison points, chipped and ground chert and fossilized wood celts, a sandstone discoidal, and a ground spud. Most of the occupation at this large site appears to have been in the fifteenth or sixteenth centuries, but part of it could be contemporary with the latest occupation in Phillips County.

INTRODUCTION II: THE ORIGIN OF THE OLD TOWN PHASE CONCEPT AND ITS PLACE IN MODERN LOWER VALLEY STUDIES

Recorded investigation of the Mississippian cultural remains in Phillips County began with the Bureau of American Ethnography (BAE) Division of Mound Exploration's work. The 1891 catalogue (Thomas 1891) maps five mounds or mound groups and lists eight. Mounds at the mouth of the St. Francis River (Moore-Steagal locality) had been investigated by C.H. Boyd (Mason 1880). Goach Mounds, with a recent cemetery, heard of and perhaps viewed by Edward Palmer, seem to have been in the Fitzhugh area. The locations of the Barney and Rodgers mounds are in the vicinity of Modoc. Both had mounds with more than one level and contained "beds of clay burnt to a brick red", indicating a probable Mississippian cultural affiliation (Thomas 1894:234).

The Smithsonian Institution had in 1872 received ceramics collected by Jerome Pillow during the construction of a levee across Old Town and Long Lake bayous (Devereax 1873), perhaps leading to Col. P. W. Norris' visit to Old Town. The BAE description of the Old Town Works included several illustrations. The works portrayed probably represented two separate towns, occupied either sequentially or by neighboring, but unamalgamated, allies. "Nothing of interest" was revealed in excavation of the large mound and excavation of the houses revealed "the usual fire-bed, charcoal, and fragments of pottery" (Thomas 1894:235). It now is apparent that the site is Mississippian, based on mound form, content, and site plan. The BAE's mention of bark coffins and house depressions is particularly important, as these were destroyed soon after cultivation began. The degree of preservation may indicate a very late prehistoric/Protohistoric component, although no mention was made of glass beads or other trade goods.

C. B. Moore (1911) excavated the Avenue site in southern Phillips County, recovering a number of vessels. These are illustrated and described in his usual extensive manner, but the county was largely neglected by professional archaeology until 1940, when the LMS mapped and surface collected additional sites. The Old Town site was found to be almost completely destroyed by levee construction and alluviation. Burials and pothunting were still evident and a surface collection was obtained from borrow pit outwash. The Buie site, the southern part of Thomas' Old Town Works, was revealed where a drainage ditch cut through an old levee. The Fitzhugh site was also largely destroyed by scouring from a levee break.

The 1961-62 Alden Redfield Dalton Project collected a number of "arrows" and other typical Mississippian lithics but without access to the collections or better provenience information, this study is of little relevance, although four of his Phillips County sites had Mississippian components. Arrowpoints were collected, along with eccentric objects, knives, scrapers, adzes, picks, hoes, chunkey stones, celts, and pipes, many of which were undoubtedly Mississippian. In 1975, Michael Hoffman compiled a list of prehistoric material from Arkansas at the Smithsonian Institution, and while the provenience seem confused, those attributable to known sites are discussed below. 

Based on the very limited LMS collections, Phillips (1970) defined the Old Town Phase of the Mississippi Period. Seven sites (Old Town, Ellis, Fitzhugh, Buie, Tinsley, Baytown, and Rodgers) on the L, M, and N quadrangles of tier 15 are shown as having Old Town Phase components. All of the sites lie within the Mississippi floodplain between the Mississippi and White rivers. Baytown, the only of these sites to be located in Monroe County, is not considered here, but the others are. To the north are the Kent phase sites (mostly in Lee and St. Francis counties, Arkansas, but also including a few in Phillips County, Arkansas, and Tunica County, Mississippi). To the east is the already-discussed Parchman phase. To the south are the Hushpuckena-Oliver phase sites in Coahoma and Bolivar counties, Mississippi, and the Quapaw phase (now called Menard complex) sites in Arkansas and Phillips counties, Arkansas. All of these phases except the last one are characterized exclusively by very similar ceramic complexes, and are defined largely by reference to each other. Sites in each of these surrounding groupings have been investigated since the LMS and all of the phases except Old Town are somewhat better defined today. The 1989 Ellis Mound excavations are the only significant contribution for the Old Town phase. Phillips' only discussion of the ceramic attributes of the Old Town phase states that these assemblages "do not fit well with the Kent phase to the north, still less with Parchman across the River, principally through lack of Bell Plain…The only excuse for an Old Town phase at this stage is to fill out the distribution of Mississippian culture in this part of the area" (Phillips 1970:940). As later discussion will show, the characterization of the sites of the Old Town phase area as lacking Bell Plain was erroneous and based on small collections.

Since 1970, there has been a tendency to accept these phases as proposed, although there have been some attempts to redefine them, particularly in House's (1987, 1993) long-term Kent phase research in Lee and St. Francis counties and McGimey's (1989) continued efforts to synthesize a 1600-1800 Quapaw phase along the lower Arkansas. Most research, however, has been largely fortuitous, as we have seen to have been the case in Coahoma County, and has focused on individual sites rather than regions. Hence, while we were not without indicators of what was to be expected at the Ellis site when the work there began, a modern framework for the area in which to place the Ellis site was lacking. Background information had to be drawn from some fairly distant locations, but a general sequence of events, as outlined below, could be drawn for eastern Arkansas.

At about A.D. 800, when transitional Late Woodland-Emergent Mississippian components are recognized in the Sunflower basin, northeast Arkansas was seeing the development of the Big Lake phase, best known from the Zebree site (Morse and Morse 1980). By A.D. 1000 agricultural villages were to be found in a variety of habitats, ranging from the Wilson phase sites, such as Burris, in the Cache and upper White rivers basins (Jeter 1988) to the Brougham Lake farmsteads in the Eastern Lowlands (Klinger et al. 1983). Some research in the late 1960s revealed the ceramic, lithic, and architectural content of Early and Middle Mississippi period sites, such as the salvage work on the mound and houses at Hazel (3PO6) (Morse and Smith 1973), the Dumond mound (3AR40) on the Grand Prairie (Scholtz 1968b), and the Roland midden (3AR30) on the lower White River (Scholtz 1991). Salvage excavations of Barrett Mound A (3LE3), along with other Kent phase research, has documented an early phase just to the north of Phillips County (House and House 1987). To the south and west, a few smaller sites, such as Buford (22TL501) (Marshall 1988) and French (22HO565) (Potts 1976, Smith 1987) have provided indications of the extent of the emerging Mississippian culture deep into the Yazoo-Mississippi Delta.

Around 1400, much of the Western Lowland began to be abandoned in favor of settlement in densely populated, fortified villages along the St. Francis and Mississippi Rivers (House and McKelway 1982, Morse 1981, Morse and Morse 1983). Farmsteads seem to disappear in many areas, and, while the reasons for these population shifts are not clear, they are thought to be linked to increasing regional population density and increased levels of organized intergroup warfare. Political rivalry over travel routes, hunting grounds, and high quality agricultural land are also thought to be associated with these shifts. House and McKelway (1982) have also suggested that the shift to reliance on permanent field agriculture and large, long-term villages may have increased the need for a sustainable protein source such as could only be provided by large oxbow lakes, the typical setting for the large late towns. One such nucleated village on Bee Bayou, Dupree (3PH1), in southern Phillips County was briefly tested during land leveling in the 1960s (Moselage 1965, McGimey 1965), but it was a long time before professional archaeological attention focused on the Mississippi period in the area again.

The area's later Mississippian sites did not, however, escape everyone's notice. As in Coahoma County, several Phillips County sites have been extensively pothunted; Brown (1978) had documented the displayed portion of one of these collections, which will play an important role in the following discussion of the Phillips County sites, but at present it is enough to say that the collection can be taken as illustrative of the content of the Old Town phase's most spectacular burial furniture, also illustrated in the many vessels attributed to Phillips County in Hathcock's (1983) volume on the popularly defined ancestral Quapaw culture. However, House (1991:19), like Brain (1988), in describing burial assemblages pothunted from sites in Arkansas County, has cautioned against giving too much importance to these late burial-derived assemblages, stating that mortuary vessels "appear to be ill-suited to the task of placing archaeological components into a high level cultural unit such as Mississippian, Caddoan, or Tunican [because] there is reason to believe that specific design and vessel form modes--and probably vessels as well--freely crossed ethnic boundaries". While these comments were made about the mouth of the Arkansas region, "an open field of diffusion and interaction among small-scale, locally diversified cultural entities" (House 1991:20), these cautions are well taken in the Old Town phase area immediately to the north, too.

Protohistoric and contact period archaeology in the area is only beginning. Occasionally, ca. A.D. 1550-1700 burials and occupation areas with small amounts of European materials are reported at locations as divers as the Parkin, Kent, and Clay Hill sites to the north and at French (22HO565; Potts 1976) and Menard (Ford 1961) to the south, but, once again, professional attention has not focused on the area, despite the extensive, systematic looting of the contexts. These finds provide glimpses at otherwise unrecorded contacts. 

Mississippian archaeology in Phillips County has been implicitly studied through CRM surveys. Projects that have reported evidence pertinent to the questions addressed here will be summarized and the sites themselves will be described in detail below. As in Coahoma County, the data gained has been distributional rather than chronological and even when Mississippi period sites were located, few more intensive investigations have resulted.

Two projects have been of watershed scale, one on the braided stream terrace and the other south of Old Town Lake. Fleniken's (1974) preliminary survey of the Lee-Phillips drainage rechanelization project reported one small Mississippian component in the braided stream area among the 17 sites found, but this low-intensity survey undoubtedly missed many sites. Most of the area examined was vegetated land in the Lick Creek ditch system (Figure 2). The majority of the sites were on higher terraces, outside the direct impact zone, and all were on agricultural land. In 1977, John Dorwin and Judith Stewart conducted background research and cursory fieldwork for Soil System, Inc. (SSI) around the Old Town Lake berm and along Long Lake Bayou; parts of the berm were examined again, without much result, by Historic Preservation Associates (HPA) (Klinger and Cochran 1982). HPA attributed their failure to locate cultural resources to very poor field conditions along the 5.6 miles of channel improvement. The 27 miles of reconnaissance along Long Lake, conducted by SSI resulted in the recording of 3PH153 (a Mississippian site outside the direct impact zone) and 26 other sites (11 prehistoric components, 5 of which were isolated finds, and 15 historic components) but no other Mississippian materials (Dorwin and Stewart 1977:20). They also examined an area where a gated culvert was placed at Old Town, noting that construction without prior cultural resources survey had damaged site deposits and opened them to looting.

Four projects have provided information from large, fairly homogeneous tracts. Also, in 1989, a small borrow pit survey reported a site in a setting similar to the one Fleniken (1974) had recorded (Williamson 1989). Beginning in 1978, timber sale and recreation development surveys in the St. Francis Ranger District, Ozark-St. Francis National Forest, began to provide a wide coverage of Crowleys Ridge, which appears to be devoid of Mississippian remains except along its edges. Finally, large surveys for Helena Chemical (Santeford and House 1983), Helena Slackwater Harbor Industrial Park (Harcourt 1991), and Helena Slackwater Harbor (Childress et al. 1995) have provided a great deal of distributional evidence from the alluvium east of Old Town Lake and across from Friars Point, Mississippi. The Old Town Lake alluvium was apparently extensively occupied during the Baytown and Coles Creek periods, with numerous small sites being reported along many watercourses in the area. The survey of selected portions of a 720 ha/1800 acre proposed Helena Chemical plant site located 28 new sites (8 Late Woodland, 5 Woodland and Historic, and 15 Historic). Harcourt's (1991) preliminary pedestrian survey of 3000 acres/5 square miles proposed for industrialization recorded 26 new and 6 previously reported sites, 9 with potentially significant small Woodland habitation deposits.

While the finding is not unequivocal, it appears that the Old Town Lake-area alluvium, the setting of several mound groups with village middens, largely lacks farmstead-size Mississippian components. Nearly 5000 acres has been surveyed, but at a generally low level of intensity. For instance, while the proposed chemical plant site study considered some 1760 acres, only about 120,000 foot of pedestrian survey transect was walked by two crew members in four days, or 220 acres per person per day (Santeford and House 1983:24). It seems likely that Mississippian components might be located on these tracts if more intensive survey was undertaken.

THE ELLIS MOUND (3PH19) EXCAVATIONS

After having seen how Mississippian archaeology has developed in the area, we now turn to information from the only Old Town phase site to have been tested in recent decades, the Ellis Mound (3PH19), a cut-in-two half of a mound lying just outside the Mississippi River levee at Montezuma Bend. Construction of the mound appears to date mostly to the fifteenth century. After fuller consideration of the fifteenth century in this part of the Delta, I will return to the 50-odd other Mississippian sites that have been identified in Phillips County to try to identify the contemporaries and cultural affines of Ellis Mound.

The first recorded mention of the site is associated with the late nineteenth century Smithsonian collections (Hoffman 1975), which include bone fragments, potsherds, and burned clay from mounds on Eugene Ellis' place and J. Perry's ranch, ten miles below Helena near the Montezuma Levee. Hoffman's (1975) index also notes cabins on a lowered mound and another mound 300 foot to the south (a location now under the mainline levee), citing "Corrigan 1895 and Shinault 1882", references I have been unable to locate. This note provides the only indication that Ellis was a two mound and plaza town. The Smithsonian materials are "shell tempered coarse, gray brown ware Fine sand tempered or non tempered Smooth slip gray or buff-on one or both sides Decoration grooved diagonal lines, double slipped, smooth surface Punctate impressions, gray ware, sand or grit tempered, traces of charcoal on crude ware. All surface material. Chunks of burned clay showing grass and reed impressions Flint-red and brown Chunk of clear quartz" (Hoffman 1975). Most of the ceramics appear to be what would now be termed Mississippi Plain and perhaps Bell Plain, along with Barton Incised, Parkin Punctated, and Old Town Red. Daub and Citronelle gravel are typical Mississippian finds, but the quartz is perhaps attributable to a Marksville period occupation documented in the 1990 testing.

Ellis Mound was next recorded in 1940 by James Griffin, Fisher Motz, and Philip Phillips of the LMS as 15-N-2, a large rectangular mound with Baytown and Middle Mississippi components, owned by H. Rightor and Mrs. Chinault of Helena. The LMS site description reads "Mound, now rectangular with tenant house on it, about 180 x 120 ft. and 6 ft. high at the highest point. May have been shaped up to accommodate house. Tenant says Rightor (owner) said mound was built by one Pillar, a former owner. This may refer to the shaping up, which is particularly in evidence on the N side. There is considerable material on the mound, both Baytown and M. Miss., to N and W considerable portion of Baytown" (LMS file copies, University of Memphis, Department of Anthropology). The LMS collection included 232 Neeleys Ferry Plain (now Mississippi Plain), 6 Parkin Punctated, 6 Barton Incised, 14 Bell Plain, 4 Old Town Red, and 1 Carson Red on Buff. Mulberry Creek Cordmarked, Baytown Plain, Larto Red, Marksville Incised, and Mazique Incised (Woodland period types) were also collected.

In 1982, John House visited the Ellis site. The mound had been partly leveled and a cattle pen, barn, and paved hay storage lot built where the west half of the mound had been. A grab collection (AAS 82-584) from the eroding cut face of the mound confirmed a Late Mississippian date. The collection is now stored at the University of Arkansas, Pine Bluff (UAPB). In addition to grog tempered ceramics, a cylindrical Old Town Red bottle neck sherd, a Parkin Punctated sherd, and a Bell Plain bottle base were recovered. Five other Bell Plain sherds, and 17 Mississippi Plain sherds complete the Mississippian ceramics. One unusual piece of grass tempered, vitrified daub shows six parallel bent canes. Rarely are more than three or four parallel split canes observed, and these are generally woven. Another grassy piece of daub shows wood grain on two adjacent surfaces. Riven wood is infrequently, but consistently, observed in collections of Mississippian daub; it is worthwhile to note that Buchner (1996) reports a door jamb of planks on the moundtop structure at the West Mounds site (22TU520), in adjacent Tunica County, Mississippi. There are also three pieces of daub with a flat exterior surface with a 1 mm sand coat over a 7 mm plain clay plaster on a grassy core and 11 fragments of grassy core daub. More of this sand and clay surfaced daub was recovered in the 1989 excavations. A small amount of stone, equally attributable the Woodland components, two bone splinters, and a few turn-of-the-century item complete the UAPB collection. From this limited site revisit, it seemed obvious that the site deserved attention during the planned industrial development of the area south of Helena.

As part of the Helena Harbor survey, HPA conducted limited work at Ellis. In the bean field west of the levee fence they found grog tempered pottery, debitage, and daub. HPA also made five auger tests and three 50x50x50 cm column samples in this area, which showed 20 to 40 cm of dark grayish brown soil with artifacts, indicating that it is probable that features survive in the cultivated portion of the site. Also, about half a kilometer to the north of the mound, on a slight ridge, separated from the main site by 200 m, is a 100 by 30 m ridge where the HPA party found a shell tempered plain sherd and an arrow point. They extended the site boundary to include this scatter and noted that there had been historic structures along the south boundary of the site (Childress et al. 1995).

The 1989 Corps of Engineers work at Ellis was intended to do as little harm as possible to the mound remnant while assessing the extent of the remaining deposits and obtaining dates and artifact samples. As a result, the excavation was limited to cutting three trenches into the slump perpendicular to the cut face of the mound. Most of the excavation of undisturbed deposits was limited to mound fill stages; when intact structure floors were encountered, the undisturbed materials were investigated in a limited way and then the trenches were stepped back to the west. Therefore, only a small volume of the deeper, earlier portion of the mound and submound surface was investigated. Disturbed materials were dryscreened and most undisturbed deposits were waterscreened through quarter inch hardware cloth. We identified construction stages, which are tentatively numbered from the mound base up. It appears that at least six occupation/construction stages are still present. The AAS revisited the site after our work and recommended that the site be nominated to the National Register of Historic Places.

Trench A, two meters wide, was placed perpendicular to the crest of the mound remnant, and extended to the edge of the pavement. The nearly vertical face of the uppermost remaining fill episode (Stage VI) was cut back only far enough to expose an unweathered surface. The upper 20 cm of this very dark grayish brown silt contained many small brick and daub fragments along with other historic materials from the house that sat on the mound at the time it was recorded by the LMS (Figures 3 and 4). If the existing surface was in use at the time of the site's abandonment by the Mississippian people, all traces of the final aboriginal occupation were probably obliterated by the twentieth century occupants. It seem likely that one or more sixteenth century summits are missing from the record, given that both the Smithsonian informant and the LMS recorded the mound as lowered and reshaped.

Ninety cm below the present surface at the east end of Trench A, the first intact daub and charcoal lens was encountered. This burned rubble was 10 to 20 cm thick, representing the remains of a collapsed structure that sat atop the Stage V living surface. This material and the fill directly under the rubble were excavated separately to obtain a separate sample of the Stage V material culture. Excavation of the Stage V mound fill continued to 40 cm below the burned rubble (120 cmbs) before the trench was stepped to the west. A 10 to 12 cm diameter post was identified under the Stage V rubble, and a charcoal sample was taken from it. Excavation of the post halted upon encountering ash, daub, and charcoal of the next lower mound stage. The total thickness of Stage V was 60 cm; it consisted primarily of very dark grayish brown silt with loads of light yellowish brown silt and mottled yellowish brown silt with charcoal flecks.

At 130 cmbs, a horizontal, 5 cm thick charcoal lens was noted and sampled; this probably corresponds to the second "floor" noted in Trench C, described below. About 30-35 cm below this charcoal, beginning at 175 cmbs, was a third layer of structural remains (Feature 4) consisting of daub overlying cane in a shallow baked depression (Figure 5). The cane was consistently oriented in the same direction and followed the eastward slope of the baked depression. A flat rock was found on the baked surface and by 180 cmbs, the depression was found to have a distinct outline indicating the corner of a sunken floor structure. Most of the structure appears to be intact, extending to the east into the mound remnant. At 190 cmbs, two postholes were encountered at the edge of the depressed floor. One yielded a carbon sample. Feature 4 seemed to overlie a second basin filled with light yellowish brown basket loads intruding a fill zone of very dark grayish brown soil mixed with small daub fragments. I interpret this as indicating that there were two consecutive structures atop Stage IV without the addition of mound fill, although, alternately, Stage IV may have been a very thin mound fill, as a second group of three postholes began at a level slightly below the base of Feature 4.

Interpretation of the structure(s) atop Stage IV as having semisubterranean floors is also supported by the north wall profile, which shows daub rubble lying on a sloping surface. The surface bearing the rubble rises from the west to a crest at the presumed wall line and then descends into a depression which slopes to the east. The second mentioned basin seem to be in the same location and orientation as the first. The mound surface at the time of the initial basin's burning lay at about 180-190 cm below the present mound surface.

Under the rubble in these basins is a lens of dark brown basket loaded silty clay and several lenses of yellowish brown silty soil, construction Stage III. The basin was also partly filled by the Stage III loads, along with fragmented daub, not the solidly massed rubble generally associated with collapsed walls. The Stage IV fill seem to be mostly midden soil (darker, with charcoal, bone, and artifacts) with lesser amounts of subsoil (lighter, yellowish soil). The Stage III fills are mixed lenses of light and dark yellowish brown silts, distinct from the grayish brown silty clays that make up most of the observed part of the mound. The final burned structure encountered in Trench A was a bright orange baked surface overlain by a thin white ash and charcoal lens at 210 cmbs. This corresponds approximately with the top level of the asphalt pavement. Not much Stage II fill was excavated below this level, but postholes from the Stage II occupation beginning at 200-205 cmbs were recorded (Figure 6).

The maximum depth reached in Trench A was 220 cmbs, just above the submound surface, based on the results of Trench C. A recent posthole along the edge of the pavement, however, revealed that cultural deposits extend to 250 cmbs. Thus, it is likely that undisturbed Woodland and earlier Mississippian deposits still exist below the pavement.

Trench B was one m wide and lay 4 m north of Trench A. The existing exposed mound fill was cut back to the first burned clay floor. Subsequently the trench was stepped westward and down following the burned material as each intact construction stage was encountered (Figures 7 and 8). About half of the volume of Trench B was disturbed slump from the last three or four construction stages. Artifacts, carbon, and floatation samples were collected from three baked occupation levels. Each had a daub/charcoal/ash sequence indicative of rapid covering with soil after the burning structure collapsed (Figure 9). The charcoal is mostly cane, although wood was also noted. These thin burned debris layers, each representing the end of an occupation stage lay on compact, baked surfaces and no attempt was made to cut through these compact soils, which are believed to be the floors or, perhaps, heavily trafficked immediate exterior eaves or yards of structures. They correspond well in depths to the portions of the Stage III/IV structures encountered in Trench A. As in Trench A, cultural deposits extend below the pavement where excavation ceased.

Trench C, also one meter wide, lay south of Trench A and did not exhibit the pronounced floors seen in the uppermost levels of Trenches A and B (Figures 10 and 11). It was much more disturbed in its upper zones. There were no distinct lenses of burned material above the level of the pavement. It was, however, the only trench to significantly explore below the level of the pavement, to ostensibly sterile soil, and the only one to result in the recovery of a radiocarbon sample from premound (Stage I) context. About one third of the volume of Trench C excavated was slump from the upper construction stages, and some of the remainder appeared to be mixed by pothunting and/or twentieth century occupation, which included reshaping the mound.

Between 220 and 230 cmbs, distinct areas of light colored, nearly sterile soil and a darker, loamier, organic, artifact bearing soil were found (Figure 12). The fill strata above were similar to the darker soil of the overlying mound fill. The sloping contact of these two stratigraphic units may represent a sunken, banked floor similar to the one recorded in Trench A, Stage III/IV, however, not enough of this surface was uncovered to state for certain that these units are not just representative of distinct sources of fill. The lowest level of the trench, at about 230 cmbs, or 40 cm below the surface of the asphalt, revealed a wall trench containing postmolds. When excavated, the 20 cm-wide wall trench was found to have a series of six posts, with diameter of 5, 12, 11, 8, 6, and 8 cm. The combination of charcoal flecks from these posts yielded less than 2 gram of material, which was submitted for accelerator mass spectrometry (AMS) radiocarbon dating. Other posts encountered at this level, but not associable with the northeast-southwest oriented wall trench provide a further indication of the intensity of premound occupation, of either Late Woodland or Middle Mississippi period construction.

ELLIS ARTIFACT ANALYSIS

Ceramics, lithics, daub, charcoal, and bone were recovered from the Ellis mound excavations. Unfortunately, much of the material recovered was from secondary deposits. Very little material can be directly associated with the occupation stages, and in the case of all classes besides ceramics, it is impossible to sort the Woodland and Mississippian materials. A standardized method of analysis was preferred where one exists to make the Ellis assemblage of greater comparative value to other researchers. A standard typology exists for prehistoric ceramics (Phillips 1970). There is also a detailed descriptive typology available for daub, the burned plaster from wattle-and-daub or jacal construction, (Connaway 1984), but unfortunately it has seen little use, even though Connaway's typology has significant potential for partial reconstruction of building superstructures. Lithics were analyzed according to a scheme constructed for the Ellis site, but that uses generally accepted technological features for assignment of artifacts to classes. The Helena Harbor contract report (Childress et al. 1995) contains tabular appendices of floral and faunal material recovered (compiled by Andrea Shea and Emanuel Breitburg, respectively).

Ceramics

The entire collection was first passed through a one-half inch screen, with small sherds and pieces of burned earth being sorted into temper categories and then weighed. The remainder of the collection consisted of 2,717 larger sherds (633 being from the Woodland periods), some of which were subsequently found to be conjoinable. Ceramic typology follows the LMS or "type-variety" system, which is basically a two variable (paste and surface finish/decoration) classification. Where sherds were large enough to fully observe these characteristics, they were classified in the type-variety system. Where the system is inadequate to describe the Ellis materials, they were described in term of temper and surface finish. Ceramic fabrics (pastes) include clay or grog tempered, sand and clay tempered, coarse shell tempered, fine shell tempered, and fine clay and/or fine shell tempered. The first two are attributed to the Woodland period. The major Woodland types recovered were Mulberry Creek Cordmarked, Baytown Plain, Larto Red, Marksville Incised, Marksville Stamped, and Indian Bay Stamped. Possible examples of Salomon Brushed, Withers Fabric Marked, and Cormorant Cord Impressed were also noted. A primary Middle Woodland (Marksville period) component is indicated with Early Woodland (Tchula period) and possible Late Woodland (Baytown and/or Coles Creek period) components as well. The interested reader is referred to Childress et al. (1995) for further details about Ellis and other Marksville assemblages from the Helena Harbor project.

Mississippi Plain, the ubiquitous coarse shell tempered ware of late prehistory, is represented by 1126 sherds larger than one half inch. Although the Ellis materials provide a large enough sample to support definition of a local variety for the Old Town phase, the descriptive parameters have not been established; the material is sufficiently like var. Neeleys Ferry. Some sherds do provide additional information on vessel form, which is discussed below.

We recovered 14 painted sherds assigned to the types Avenue Polychrome/Nodena Red and White (as Avenue, but lacking the black varnish or lacquer) and Old Town Red vars. Old Town (coarse shell tempered) and Beaverdam (fine shell and grog tempered). No obvious sherds of Carson Red on Buff were recorded, but the much-eroded material does not preclude the presence of this late type. An attempt was made to identify vessel form, although small sherd size precludes definitive statements about the original vessels represented. Red pigments derive from ferruginous (iron-bearing) minerals like hematite while white slips most likely derive from kaolinite, a highly weathered clay. Black stains or varnishes are generally poorly preserved and are often assumed to be of vegetable origin and to have been applied after firing.

Four bottles and two flaring rim bowls were identified among the painted ceramics. Identification of flaring rim bowls is based on a markedly curving, interior red slipped body sherd with the point of recurve defined by a broad, shallow incised line in one case and on a squared, possibly nicked, rim, red filmed on the interior in the other. Both are on coarse pastes. Four other coarse paste sherds have interior red filming and two are slipped on the exterior. One is possibly a basal sherd from a flattened base bowl. Two bottles were identified from body-neck junctions and two others are identified as bottles based on size and curvature of large body sherds. One neck-body junction is exterior red filmed. The other is more difficult to interpret due to apparent burning. This latter example may be a red-on-white sherd, a rare mode (Phillips 1970:142), however, the white color appears to be due to open (oxidizing), high-temperature firing or destructive burning. The red elements on this sherd are small circular spots surrounding either the neck or a larger red circular element. The red and white body sherds probably represent curvilinear (spiral/swastika or interlocking scroll/meander) motifs.

Barton Incised (N=25) and Winterville Incised (N=2) rims are limited to coarse shell tempered jars. The specific morphology and size of the jars is variable, as will be discussed below. Size characterizations were based on orifice diameter where the sherds represented a large enough arc to yield accurate measurements. Many sherds can only be classified within a five to ten cm range, and some large sherds (orifice diameter around 45 or 50 cm) are nearly flat. Incision was also characterized by line width and spacing.

Two vessel fragments of Winterville Incised var. Winterville (Brain 1988:383, 1989:161) were recovered. One is a very large jar with a short, outflaring rim and incision on the neck-shoulder (Figure 13). The motif is the over-and-under meander (guilloche) comprised of eight lines. The other is a much smaller neckless jar, possibly with a lug, with the same motif composed of four lines. The incision is on the upper body. Both are sooted on the exteriors above the point of maximum diameter. The incisions were made with a sharp pointed tool while the paste was still quite wet. The treatment is more uniform and precise on the large jar and rather haphazard on the smaller vessel.

A very small sherd with fine, heterogeneous paste may be Walls Engraved or some other (Tunican or Caddoan?) engraved type. It comes from a sharply inflected, small to medium, polished black vessel.

Nine sherds are classified as Parkin Punctated. One has closely spaced and aligned punctations on the body below a plain rim, while another has sets of punctations arranged in rows oriented obliquely to the lip. Besides these two sherds showing alignments similar to Parkin Punctated var. Castile (formerly Castile Linear Punctated) or Pouncey Ridge Pinched, there are five sherds with closely spaced, overall, unaligned nail punctations on the body; one sherd with wide spaced nail marks; and another sherd from the inflection point of a jar that has a single row of punctations broken in half which may represent the lower border of punctations that sometimes occurs as a Barton Incised mode. Combined punctated and incised ceramics (Owens Punctated and miscellaneous unclassified) are fairly common in other parts of Phillips County, but are not found in the available collections from Ellis.

Of the vessel forms indicated by the mostly plain ceramics, eleven flaring rim bowls were identified, with six additional possible specimens. The majority of the sherds indicating vessel morphology were of coarse shell tempered paste, but a few specimens of fine paste (Bell/Addis or a moderately fine approximation of these pastes) were also diagnostic. Modifications on flaring rim bowls are: cutouts or eared/scalloped (N=2), nicked rim (N=4), interior incision along the inflection point (N=7), and red filming (N=2).

The scalloped flaring rim bowl form (two positive and two probable examples) varies in depth and width of the cut-outs, and probably in the number of cut-outs, although complete vessels of this form do not have less than four (Figure 14). One example is from spoil, while the other, larger example is from Trench C in or below the charcoal concentration at 179-200 cmbs (Stage III/IV). The smaller polished Bell Plain bowl has an orifice diameter of 15-18 cm and a square lip. The well-smoothed Mississippi Plain bowl was 35+ cm at the rim and 7 cm deep, with outslanting wall without marked inflection, a rather flattened base, and square lip.

The nicked flaring rim bowls all came from Trench A, with one each from the spoil, from 130-145 cmbs, from loaded fill at 160-210 cmbs, and from 206-216 cmbs. The one from the spoil had a diameter of about 35 cm, with a rolled out, squared rim with wide dents resulting in a wavy rim. The one associated with the charcoal has about the same diameter. Vessel walls below the grooved inflection point are steep and the rim flares out sharply, so the vessel could be assigned to the class "helmet bowl" intermediate between jars and bowls, but apparently functionally more like cooking jars. This last is Mississippi Plain and the nicks are wide. The 160-210 cmbs specimen is from a Mississippi Plain bowl with a diameter of 30-35 cm, with nicks on top of the rounded rim and a groove marking the inflection point. The final example comes from a vessel at least 5 cm tall, with an orifice diameter of 26 cm, and a diameter of 22 cm at the inflection point, which is sharp, but not marked with a line. The rim is square, with the excess being rolled down and back from the rim before the application of deep, sharp-edged nicks. The paste is well-smoothed Mississippi Plain.

Vessels with incision-marked inflection points vary from having sharp cut overhanging lines to a wide groove. Two specimens occur in conjunction with nicked rims and a third comes from a red filmed body sherd. Red filming is a common mode on deep flaring rim and helmet bowls. One specimen from Trench C, 90-113 cmbs has interior filming on a moderately coarse shell tempered paste, possibly with other tempering agents added, and has a somewhat squared lip. The second, the just-mentioned body sherd with wide groove marking the inflection point; a coarse paste deep, slightly flaring or helmet bowl, is from Trench C, around 90 cmbs.

Of the 17 flaring rim bowls, sometimes called "plates" or "platters", 15 are of Mississippi Plain, one is heterogeneous Bell paste, and one is heterogeneous rather fine Mississippi Plain. Although the temper is generally coarse, seven are polished or well-smoothed. All but one are too small to accurately measure diameter, so estimates were recorded in 5 cm increments. There are one 15-19 cm, three 20-24 cm, seven 30-34 cm, one 35-39 cm, and one 40-44 cm diameter bowls. There are six additional possible flaring rim bowls that may actually represent unrestricted jars or simple slanting wall bowls.

Five rim sherds were identified as hemispherical-to-inslanting bowls. All have some polishing on the interior and exterior. Although these sherds are really too small to measure orifice diameter, they seem to have smaller diameter (but not necessarily capacities), on average than the flaring rim bowls. Two have broad, perpendicular nicking on the lip exterior. The "lands" between the nicks are abraded. Two have the top of the lip flattened and three others have rounded lips.

Seven sherds are identified as parts of bottles (Figure 15). Three are from painted vessels already discussed. A red sherd with heterogeneous Bell paste comes from the neck-body junction. The provenience is Trench C, 1-90 cmbs. A second red neck-body junction comes from Trench C, ca. 175 cmbs. An Avenue Polychrome body sherd comes from the surface. Three bottles may be called "jugs" or bottles with wide, short necks. One wide mouth bottle neck has the Memphis rim mode (House 1993) on Mississippi Plain paste. The beveled, thinned flange or sharply flaring lip is 0.9 cm wide and perpendicular to the wall of the neck. It is from Trench A, spoil. Another jug with orifice diameter of 6 cm has a less pronounced interior bevel (House's intermediate rim mode). The third jug has an orifice diameter of 12-14 cm, heterogeneous Bell paste, and a squared lip. One Bell paste sherd from Trench A spoil is from a pedestal base bottle with a diameter of 8-10 cm.

There are 16 sherds identified as coming from jars and another 16 probable examples. All are on coarse pastes, although one is finer than usual and may include other tempering agents and another has sparse moderately coarse shell temper. The sixteen jars include five Barton Incised, three Parkin Punctated, and two Winterville Incised. The largest Winterville fragment is reconstructed from sherds from the occupation surface of Stages III/IV. Estimated orifice diameter for the sixteen jar rim are three 10-14 cm, two 15-19 cm, two 2-24, three 25-29 cm, one 30-34 cm, three 35-39 cm, one 40-44 cm, and one 45-49 cm. Lip morphology is variable. There is one example that seem to be an extreme expression of the Memphis rim mode, although it could be an extremely large lug. However, nearly perpendicular flanges encircling the entire lip are known from this region, for example a 1.8 cm wide flange on a very large Winterville Incised jar from the Oliver site (Starr 1992). Deep flaring rim beakers from the Walls phase are also similar (Childress 1992).

Other appendages associated with jars include a 1.9 cm wide strap handle with two nodes at either the rim or body attachment (only half of the handle was retrieved). A small lunate lug is placed at the inflection point or shoulder of a jar with 15 cm orifice diameter. The smaller (ca. 15 cm orifice diameter) Winterville jar has a broken lug rolled out directly from the squared lip. All jar appendages except the extreme Memphis rim mode come from disturbed contexts; the flanged example is from the only rubble lens recognized in Trench C, probably corresponding to Stages III/IV.

Lithics

The vast majority of the lithic material on Central and Lower Mississippi Valley Woodland and Mississippian sites is of local origin, with the exception of a few distinctive imported raw materials such as Fort Payne, Dover, Mill Creek, and Burlington/Crescent cherts. Both the gravels and the above-named cherts originate in the Mississippian limestones surrounding the Mississippi embayment. The Ellis materials are no exception to this regional pattern. No effort was made to classify the apparently local materials, as all can be assumed to originate in the Citronelle formation. This weathered, redeposited and/or residual gravel, lying under a mantle of loess or in active streambeds, is highly variable in macroscopic characteristics, reflecting the varied source formations. Ellis is adjacent to the Mississippi River and near Crowleys Ridge; both could be sources for the chert used at the site. Cherts, slate, coal, quartzites, fossilized or petrified organic materials, ferruginous sedimentary products, and a wide variety of other materials are found on modern Mississippi River gravel bars.

The lithic sample from the mound included 1,368 item (pebbles, cobbles, broken pebbles, broken cobbles, fire cracked rock, miscellaneous stone, tested cobbles, cores, core fragments, bifaces, biface fragments, debitage, retouched/utilized pieces, ground stone, and battered cobbles) It is difficult to provide a detailed assessment of the Mississippian lithic technology at Ellis due to the mixing of Mississippian and Woodland materials in the mound fill. Some similar tool forms were used throughout this entire span of time, while temporally diagnostic forms are extremely rare. The technology can be characterized as expedient or informal, based on local small gravels, and resulting in a limited array of end products. A trajectory of utilization can be proposed as: 1) selection and assessment of surface gravel, 2) occasional heat treating, 3) hammer reduction of suitable masses to polyhedral-to- amorphous cores or bifaces, 4) retouch flake finishing of blade edges, 5) hafting, if any, and use, including sharpening, and 6) reduction to broken or unusable state and discard. Outputs of each of these technological steps include debitage and broken item which could be recycled into expedient tools as well as finished or curated unfinished tools. The core-oriented technology in evidence at Ellis resulted in the following useful products: 1) tested cobbles, 2) cores and bifaces, 3) utilized debitage, and 4) stemmed point/knives, arrows, and drills.

Very few groundstone and nonlocal materials were recovered, so they will simply be enumerated below. Two small fragments of hematite (4 grams) are abraded on several faces, probably to produce red pigment for body paint, slipping vessels, and other art objects. The material is fine-grained and dense, and was recovered from Trench B, 175-220 cmbs. This material could be attributed to any prehistoric component.

A large (7.5 cm long, 3.3 cm thick, 204 gram) piece of arkosic or calcareous sandstone comes from Trench A, 131 cmbs (mound fill). It has four parallel striations on one surface. The striations were damaged during excavation, but appear to be prehistoric in origin. If so, this artifact belongs to the class generally termed "shaft abrader", presumed to function in removing knots or joints from arrow shafts, although similarly grooved surfaces are produced by modern flinknappers on such coarse stones in preparing striking platform along biface margins.

Two item may be of Dover or a related northern chert type (Figure 16). The first is a definite hoe flake with an earth-polished surface. It has a bifacial platform and seems to be retouched or heavily utilized on the lateral flake margins. It was recovered from general mound fill at 165-210 cmbs in Trench A. The other possible Dover flake also has a bifacial platform and retouch/utilization, but does not exhibit hoe polish. Large bifaces were imported into the Lower Mississippi Valley during the Mississippi period from the limestone regions of Tennessee and Illinois. These tools are generally indicative of intensive farming and wall trench and earthwork construction, although other forms ("maces", "swords") sometimes occur. Complete tools are sometimes found in cache pits and on house floors and sharpening flakes, often reused for their blade edges, are common midden inclusions in large lithic samples.

A very small (0.5 gram) fragment of a large "greenstone" celt was recovered from undifferentiated Stage IV or V mound fill in Trench A. The material is moderately grainy, mottled green, tan, and black metamorphic stone. There are striations perpendicular to the bit typical of axes. The artifact is most likely associated with the Woodland components.

The final exotic item to be considered is a probable Burlington/Crescent chert core trimming or platform rejuvenation flake from the uppermost fill stage in Trench B. This white Missouri material has been noted as an Early Mississippi period Barrett phase marker immediately to the north in the St. Francis basin (House and House 1985) and is also recovered in quantity at the Carson mound group 12 km to the south (Johnson 1987). It is typically thought of as a Cahokia export, generally in slabs for prepared cores used to produce microdrills for shell bead production, but similar blade cores of the same material are also diagnostic of Marksville contexts in this region (Johnson and Hayes 1995).

Most of the remainder of the lithic material consists of cores, core fragments, biface fragments, and debitage. No classifiable bifaces were recovered, although a large triangular biface identified as a preform may be a finished artifact. The mass of 24 cores ranged from 11 to 86 gram; maximum length was 5.8 cm, the minimum 2.1 cm; the maximum thickness is 3.5 cm, the minimum 1.0 cm. Other minor lithic materials include waterworn hematite, small fragments of a limy slate, ferruginous and nonferruginous sandstone, several pieces of petrified wood (occasionally used to make pebble celts), and a possible fragment of mineralized bone, (subfossil Pliestocene bones are common today in the Helena vicinity). All of these materials may have been obtained from Mississippi River gravel bars below the site. 

Daub

The typology used for the analysis of the daub follows that developed by Connaway (1984) based on samples at the Late Mississippi period Wilsford site. This typology was used for the Ellis sample because the primary variables emphasize structural characteristics imprinted on fragments of burned house plaster, but characteristics of the actual clay body (grass vs. sand temper) are also included. Gerald Smith (personal communication 1990) has suggested that sand and grass are variable components of daub, with the grass added to stiffen and strengthen the daub not being necessary with sandy clays.

Daub recovered from Ellis is a convenience sample consisting primarily of larger pieces. The collection is further biased towards those pieces having distinguishing characteristics. Daub from the cut face of the mound was generally small and badly weathered; therefore all daub and other burned clay less than 2 cm in diameter was weighed without further analysis. The daub has only moderate amounts of grass temper; many specimens have sparse temper. The paste is silty to fine sandy clay with some chopped grass temper added. Cracking during drying was observed on some poorly tempered specimens. The presence of thin coats or film (Type B.1) was noted in 5 of 14 pieces of daub that had hand-smoothed outer surfaces. These 0.3 to 0.5 cm thick film are sandy clays without grass temper. Other surfaces are only roughly smoothed while the daub was quite wet. Cane impressions cross at right angles. Groups of three or more lengths of 1.0 to 1.7 cm wide split cane are woven with their convex sides to receive the daub. The 10 pieces of Daub Type A.2.b (woven split cane on one side and irregular grass impressed opposing surface) range from 1.8 to 6.4 cm thick, with an average of 3.3 cm. The nine pieces of Type A.3 (cane impressions with unsmoothed, lumpy, unimpressed opposite side) range from 2.1 to 3.9 cm, with an average thickness of 2.6. A single piece with cane impressions on one side and hand smoothing on the other is 1.8 cm thick. Examples of Type A.5 (cane mat with opposite surface broken away) are 4.9 and 6.5 cm thick while an example of Type B.4 (hand smoothed with opposite surface unsmoothed and unimpressed) is 3.9 cm thick. The single example that bears an impression of split wood (Type C.1) is 2.5 cm thick.

The Ellis sample indicates that in addition to upright posts walls consisted of woven cane lathe plastered with silty clay often about 3 cm thick but sometimes 6 cm or more in thickness. Walls probably had a pronounced batter, with a thick base to support thin, higher daub. Some use of riven wood is indicated. The method of fastening of the cane lathe or matting to the closely-spaced wall posts is uncertain, but the specimen recovered by House that bears bent canes may give some indication of this fastening. True wattle, after the European origin of the term (branches woven between posts), does not appear to be the case in Eastern North America. Grass thatch would have of necessity overhung the mud walls in this sub-tropical environment to keep them from being constantly washed away by rain. Well-preserved Mississippian houses sometimes still exhibit a drip-line around the building as well as a small bank along the exterior of the wall line. 

Radiocarbon Dates.

Seven charcoal samples were sent to Beta Analytic by the U.S. Army Corps of Engineers. Six were processed by conventional counting and a seventh smaller sample was sent to Lawrence Livermore National Laboratory for accelerator mass spectrometer counting. Four of the dated carbon samples were from Trench A, two from Trench B, and one from Trench C (Figure 17).

 Table 1. Ellis Mound Radiocarbon Dates, Calibrated After Stuiver and Reimer 1993.

 In addition to the dated samples, a number of other charcoal samples mixed with soil and stored in aluminum foil are curated with the site assemblage for future dating. Although the dates do not have as tight a stratigraphic sequence as is desirable, they provide significant data on the individual construction stages and the duration of mound use. The latest date (Beta-41702) is the most significantly altered by calibration against the dendrochronological curve. All indicate construction during the latest part of the Middle Mississippi period and during the Late Mississippi period. A terminal date for occupation at Ellis cannot be determined due to historic reoccupation of the mound. The wall trench at the base of the mound (Trench C) seems to date to the mid-1300s. The actual date is indefinite, with multiple calibrated intercepts at A.D. 1328, 1333, and 1395, and a calibrated two standard deviation range of A.D. 1283-1441. Dates for Stages II and III are ambiguous, but a range near the end of the fourteenth century seem reasonable based on the results of Beta-41705. There are three dates from the Stage III/IV interface associated with the construction of individually set post houses with floor basins (Beta-41704,41703, and 41706). Calibrated intercepts are A.D. 1410, 1290, and 1347, with two sigma ranges of A.D. 1302-1455, 1240-1398, and 1250-1446, respectively. These dates create an interruption in the relative temporal/stratigraphic order, but there seem to be no grounds to suspect any of the dates, even though Beta-41703 does seem too early. Removing this assay does not improve the situation, as it would be just as reasonable to discard Beta-41705 as a "bad" date for Stage II. Neither is justifiable on stratigraphic grounds, and the date for the Stage III/IV interface is taken as ca. A.D. 1400. The initial accumulation of rubble on Stage IV, dated by Beta-41701, is calibrated to be A.D. 1431, with a two sigma span of A.D.1307-1616. The second continuous rubble zone, on the Stage V summit, yielded a calibrated date of A.D. 1478, with a calibrated two sigma range of A.D. 1432-1644. Daub fragments in the uppermost disturbed mantle indicate that the mound supported at least one additional occupation stage after the application of Stage VI fill. This suggests that the mound was in use until at least A.D. 1500. Using the two sigma ranges for calibrated calendar years, a maximum span of 350 years may be postulated for construction of the earthwork, but it seem more likely that the mound was built in episodes over 150 to 200 years or less between A.D. 1300 and 1500. The chrono-stratigraphic data also provide evidence that the mound stages were added at intervals of 50 years or less.

ELLIS AND THE FIFTEENTH CENTURY IN THE LOWER VALLEY In order to place the Late Mississippi period component at Ellis in a broader regional context, a comparative overview of ostensibly contemporary sites and archaeological phases within a fifty mile radius of the site follows (Figure 18). The discussion proceeds from north to south, reaching from Chucalissa to Winterville, with the Kent, Parchman, Hushpuckena, and Quapaw (or Menard complex) phases being the main foci of attention. The groundwork for this discussion has been laid with the presentation of Coahoma County, Mississippi, sites, but as I have shown, Coahoma County lacks comparable excavated mound strata. Later, after more material from Phillips County has been presented, we will return to regional comparisons and temporal placement. The basis for comparison is, first, radiocarbon dates, but in their absence, reasonable estimates of dates, and secondly, artifacts, primarily ceramics. The high degree of similarity of the Kent, Parchman, Hushpuckena-Oliver, and Old Town phases noted by Phillips (1970) appears to be borne out in the following comparison of excavated samples obtained since the LMS work. The Chucalissa Site. The presently understood Walls phase is based largely on work at Chucalissa, a small mound group just outside the northern perimeter of the fifty mile radius around the Ellis site, on the Chickasaw Bluffs in Memphis, Tennessee. It is the only well-documented Walls phase site (Childress 1992, Lumb and McNutt 1988, contra Mainfort 1996, who would exclude it from the Walls phase), although a tight cluster of other Walls phase sites are known from the alluvial valley in DeSoto County, Mississippi; the loess bluff zone of Shelby County, Tennessee, and, by some definitions, Crittenden County, Arkansas. The Walls phase (originally "Walls-Pecan Point" complex or focus in Phillips et al. 1951, which included today's Nodena phase sites) provides a valuable example of contrast in the otherwise homogeneous group of sites and phases to the south which tend to "grade" or "fade" into one another without distinct differences in ceramic assemblages. The Walls phase, with its rather distinct geographic and material parameters, is marked most visibly by the high percentages of elaborately constructed Bell paste vessels recovered in huge numbers by grave looters over the past century. Radiocarbon dates from Chucalissa indicate overlap with Ellis during the Boxtown (1250-1400) and Walls (1400-1540) phases. The Boxtown ceramic complex is dominated by Mississippi Plain, along with Parkin Punctated, Barton Incised, Owens Punctated, and Old Town Red. Lugs and loop and strap handles (some with punctations) occur on jars with gently recurved rim. Boxtown minority types (Fortune Noded, Matthews Incised, Beckwith Incised and late or "degenerate" Ramey Incised) were not recovered at Ellis and are uncommon on sites to the south of Memphis, being rather characteristic of southeast Missouri and northeast Arkansas. The early Walls phase or Bell I ceramic complex has about equal representation of Bell and Mississippi pastes, and the balance shifts more towards Bell in the late Walls phase Bell II complex (Lumb and McNutt 1988). Bell I handle forms are varied and include arcading. Parkin Punctated is the most common decorated type, and continues to be so in the Bell II complex. Some of the Bell I minority types (Owens Punctated, Rhodes Incised [spiraling swastikas], Barton Incised, Old Town Red, and Avenue Polychrome) are found at Ellis, as are parts of the Bell II complex (Walls Engraved, Fortune Noded, Manly Punctated, Parkin Punctated, Kent Incised [a variant of Barton Incised with perpendicular lines on the body], Ranch Incised, Rhodes Incised, Wallace Incised, and Hollywood White/Carson Red on Buff/Nodena Red and White). Painted ceramics never made up much of the Chucalissa or Walls phase assemblages, in contrast to developments to the south, where slipping remained an important technique from the Middle Mississippi through Protohistoric periods. One of the most striking aspects of the later Walls phase, which differentiates it from other Mississippi Valley complexes, is the richness of incised/engraved decorative techniques and the variation in vessel form. Overall, the Ellis assemblage has only general affinities with the Walls phase sites, although some particular modes (interior lip bevel and flat base on Bell pastes) are shared. Other characteristics of the Boxtown-Walls phase area are summarized by Smith (1990). There are a few small Walls components without mounds, some small sites, "camps", located in the upland creek valleys east of Memphis. Triangular and Nodena points are found up to 20 miles east of the bluffs (Smith 1990:143). Most of the main centers are spaced 1 to 3 miles apart in DeSoto County, Mississippi. Boxtown houses at Chucalissa were oriented north-south and were of square wall trench construction. There were two buildings on the latest summit of the main mound at Chucalissa. One was a 15 m (50 foot) square building with four large internal cypress posts, a floor 20 cm below the surface, and some very large storage pits containing trash, nut hulls, corn, and beans inside it. There was also a smaller building with a floor sunk three foot below the mound surface. Lower construction stages revealed in excavating trash-filled pits on the summit were about 1.5 foot thick (Smith 1990:144). These structures sound similar to the ones encountered in the Ellis mound. The site also has small house mounds with 18 to 22 foot square buildings and associated dumps on their sides. Skeletons from the site show high incidence of diet-related osteoporosis. The Kent Phase. Phillips' (1970) original definition of the Kent phase has been modified by the work of House (1987, 1993). The ceramic complex of the Kent phase was initially described as 10-20% decorated, with Barton Incised vars. Barton, Kent, and Togo a little more common than Parkin Punctated vars. Parkin and Castille., along with smaller amounts of Old Town Red and Carson Red on Buff. Subglobular jars occur in a broad size range, sometimes having strap handles or lugs. Bell paste bottles and bowls are a consistent minority. The lithic complex includes pebble cores, flake tools, polished pebble celts or chisels, and Mill Creek hoes or spades. The Kent I (1350-1450) and Kent II (1450-1550) phase sites are mostly in Lee and St. Francis Counties, north and east of Crowleys Ridge, in the lower St. Francis basin. The Kent site itself is very near the Mississippi River levee on a historic cut-off. Chronometric dating indicates that Kent is contemporary with Ellis (House 1993). Refining Phillips' characterization, the Kent I ceramic complex is now seen to have Barton Incised and Parkin Punctated, some with strap handles, Old Town Red flaring rim bowls, and Bell Plain. Triangular strap and appliqué handles and the Memphis rim mode (interior bevel on sharply outflaring rim) date to the Kent II subphase. Interestingly, the Kent phase around 1500 includes two groups of sites with different frequencies of Bell Plain, one group near the river having 20% or more Bell and another group to the west with the proportion of Bell being as low as 5% (House 1991). In comparison, the Ellis assemblage also has more Barton than Parkin. There are some sherds that could be counted as Barton Incised var. Kent, but that are here called Campbell, as the parallel lines perpendicular to the rim begin below the lip, rather than on the body as is more proper to Kent. The Ellis punctated sherds are all counted as Parkin Punctated var. Parkin, but two examples showing orientation could be called Castile. Bell bottles and bowls do occur in fairly substantial numbers, in contrast to Phillips' (1970) characterization of the Old Town phase as largely lacking Bell Plain, and red slipped materials are present at Ellis in both mound fill (presumably redeposited domestic debris) and in the mound occupation contexts, as is the case in the Kent phase sites where slipped vessels occur in both mortuary and midden samples. The Ellis assemblage does not include the Kent II handle types, but the Memphis rim mode does occur on a wide-mouth bottle or small, vertical jar neck, indicating that the late rim mode may originate slightly earlier than the handle modes. The small gravel core and expedient flake tool technology is very much in evidence at Ellis, although mixed with an unknown amount of Woodland material. No chisel fragments were recovered, but this common artifact class is to be expected on any Late Mississippi period site in the area. Mill Creek, Dover, or other imported cherts such as are typical of Kent phase sites are also present at Ellis. The Parchman and Quitman Phases. Some of the Parchman phase sites have been mentioned in the introduction. I once suggested that the Parchman phase as originally defined by Phillips (1970) should be modified to exclude the Coldwater River sites with low frequencies of Bell Plain (Starr 1984). In consideration of House's (1993) finding that Bell Plain frequency varied markedly between eastern and western Kent phase sites, this strict interpretation of the Parchman phase may not be necessary. Alternately, the difference may be temporal, with the Coldwater drainage Quitman phase sites lacking the post-1450 components that presumably would include higher proportions of Bell Plain (contra Brain 1988: Figure 196, who would place these sites in his seventeenth century version of the Parchman phase). At any rate, many of the large Sunflower River headwater sites have components contemporary with and later than Ellis. The "core area" of the Parchman phase includes the Carson, Parchman, and Saloman mound groups east of Ellis. While mound construction at Ellis may have ended around A.D. 1500, there are multiple indications (radiocarbon dates, bundle burials, glass beads, vessel form) that occupation in the Parchman phase continued into the Protohistoric period (A.D. 1542-ca. 1700). One site of debatable Kent-Parchman phase assignment in Tunica County, Mississippi, has yielded very late (ca. A.D. 1700) radiocarbon determinations for mound summit occupation (Buchner 1996). The Parchman phase settlement pattern (admittedly a synchronic mess) that has been reconstructed seems to resemble that documented for the Kent phase. Both areas have a similar variety of site types (farmstead clusters, hamlets, non-mound villages, small centers with one or more mounds, and more than one center with multiple large mounds), indicating parallel developmental histories. Both have a dense concentration of mound groups near the Mississippi and also include numerous sites on the bayous well away from the main channels of the Mississippi, St. Francis, and Sunflower Rivers. Recall that Wilsford (22CO516), the site with the unusual platform structures, dates A.D. 1200-1600 (in all likelihood ca. A.D. 1425), contemporary with Ellis and, as is the case at Ellis, Mississippi Plain, Bell/Addis Plain, Barton Incised, and slipped ceramics make up most of the small collection (Connaway 1984). While the Nodena Red and White bottle with carafe neck, carinated body, and vertical panels has no close correlate at Ellis, the Barton jar with gently curving profile, square lip, and lug closely resembles the Ellis material, as does the carinated profile and rounded lip of the small Addis Plain (as Bell Plain, but not necessarily having shell) flaring rim bowl (Connaway 1984:114). Another of the Coahoma County sites, Clover Hill (22CO625) which has produced radiocarbon dates indicating contemporaneity with the later end of the Ellis component as well as occupation after the apparent cessation of mound construction at Ellis, has produced ceramics broadly comparable to those thus far recovered at Ellis: deep or helmet bowls and wide-mouth, short-neck bottles or "jugs", but also Avenue Polychrome with horizontal bands (Connaway 1981:48). Overall, the assemblage seems to me to date a little later than that at Ellis, as the radiocarbon dating suggests. The Hushpuckena-Oliver Phase. This phase, also defined in Phillips (1970), likewise includes a range of site types, with no obvious paramount center. It is geographically the most extensive phase of the region, with sites in Coahoma, Bolivar, Sunflower, and Tallahatchie Counties, Mississippi. Likewise, it spans some four centuries of the Middle, Late, and Protohistoric Mississippi periods. While the contemporaneity of Ellis and the Oliver site cannot yet be demonstrated (radiocarbon samples from the site have not been processed) there are good typological reasons to believe that the sites were occupied simultaneously. Ellis site jar profiles are typically of the gently curving "Hushpuckena" form rather than the angular "Oliver" form as defined by John Belmont (1961). This Hushpukena form has been shown to be stratigraphically earlier in the Oliver mound construction sequence, probably dating to the late Middle or early Late Mississippi interval (i.e. ca. 1300-1450). Another characteristic shared by the two sites is the type Winterville Incised, occurring in the guilloche motif (Starr 1992). Two examples recovered from a trash-filled pit at Oliver occur on the neck of a large jar and the on the shoulder of a very large jar. Both of these jars have vertical necks meeting the body at a sharp angle (Oliver vessel form), while the large Winterville Incised jar from Ellis has the decoration on the incurving part of a curving wall vessel and the smaller jar has it on the slightly restricted neck (Hushpuckena vessel form). Both sites have scalloped and nicked flaring rim bowls. Another late vessel form seen at Oliver is the helmet bowl. The form varies from nearly cylindrical to somewhat outflaring, with bases somewhat flattened. At Oliver, the helmet bowl is a fairly small cooking vessel, much smaller than the largest jars (Starr 1992). I could not confidently identify any of the Ellis sherds as helmet bowls, however, there are some slightly restricted jars and deep flaring rim bowl sherds that approach this poorly-defined vessel type. Powell Bayou (22SU516), one of the southern Hushpuckena phase sites, had some salvage work done by an early Mississippi State University (MSU) field school, and I had the opportunity to write up the site while a student there (Starr 1991). Three radiocarbon dates from the lower mound stages indicate construction contemporary with the early episodes at Ellis. The mounds were perhaps comparable in size, although both were heavily impacted before investigations began. Jar and flaring rim bowl profiles from Ellis are similar to those from Powell Bayou, and there are a number of other similarities. The larger Powell Bayou collections have more types represented than the Ellis collection does. Pouncey Ridge Pinched, seemingly an extreme variant of aligned Parkin Punctated var. Castille is common at Powell Bayou, based on a number of examples in the MSU and Cottonlandia Museum collections, while the most similar material at Ellis have aligned, widely spaced sets of punctations on jar necks. The Powell Bayou collection has examples with rows of similar nail marks on jar necks but these are spaced pinches (the critical difference being whether a single nail was jabbed into the wet clay or whether it was pinched between thumb and finger). Several modes (horizontal lines, rows of punctations or nodes forming the lower boundary of the incised area; line-filled triangles, cross-hatching, plats of vertical and oblique lines) occurring with the type Barton Incised are shared by the two sites. Rim modifications of bowls are likewise similar: red filming, scalloping, nicks and tooled punctations. Painted materials are well represented in the Powell Bayou collection. Triangular points and pebble celts were found. Structures on the mound, rebuilt many times on a single stage, were predominantly very large wall trench constructions, although one had been built with individually set posts. A third excavated Hushpuckena mound with comparable materials, but lacking radiocarbon dates, is the Buford site, east of Oliver and Powell Bayou in Tallahatchie County (Marshall 1988). The site also has Early Mississippi period components which are quite rare in the central Yazoo basin. The burned structures encountered in the mound apron or midden ridge trench were split-cane-wattle, untempered-daub walled buildings, set in wall trenches in the Middle Mississippian and supported by singly set posts later. Middle Mississippi period examples had sunken floors and exterior wall banks or bases such as are proposed for the Stage III/IV structures at Ellis. Marshall (1988) illustrates Barton Incised sherds showing the crosshatch and bounded slanting parallel line motifs seen at Ellis and records an example of var. Campbell. Old Town Red, Nodena Red and White, and Winterville Incised (sharply outflaring jar neck with squared lip) recovered from Buford also indicate fifteenth century occupation. The Quapaw phase. Although Arkansas archaeologists now prefer the term "Menard complex", I will follow the older terminology in this discussion for the sake of consistency. I do not imply (or refute) a Quapaw ethnic affiliation for the Late Mississippian sites near the mouth of the Arkansas River. A whole vessel collection from two undated, prehistoric pothunted sites, Poor (3AR3) and Massey (3AR1), a few km north of the Menard-Hodges site on the same prairie/terrace edge, has been studied by House (1991). The sites probably date 1400-1650, based on the low incidence of bundle burials, lack of European artifacts, and presence of Caddoan types. The collection is a "formally heterogeneous lot" (House 1991:19), with the most common form being the flaring helmet bowl, followed by the wide-mouth bottle or jug. Materials similar to those of Ellis and the other previously discussed sites include a Nodena Red and White bottle with vertical panels, simple hemispherical bowls, scallop rim bowls, and notched flaring rim bowls. While the assemblage includes Bell effigies, there are a number of vessels that have no correlates to the north. Overall, however, the collection has similarities to the Mississippian ceramic complexes to the north, "Old Town, Kent, and Walls, in approximately that order, on the basis of specific shared modes of vessel form and decorative design" (House 1991:20). This order is of course only logical from a geographical perspective. The utility of these collections is limited by the lack of midden materials and the apparently permeable cultural boundaries around the mouth of the Arkansas. The Winterville Site. This site, one of the largest mound groups in the Lower Valley, lies just south of the fifty mile radius around the Ellis site and a little south of the mouth of the Arkansas, but consideration of contemporary sites would not be complete without mention of Winterville, as at least the later part of its occupation is contemporary with the initial Mississippian occupation of Ellis. Numerous radiocarbon determinations indicate a long occupation encompassing several sequential phases (Brain 1989:106). Winterville, near the modern channel of the Mississippi River north of Greenville, has large and small mounds with typical construction stages and burned structures. While the site is characterized as a Coles Creek/Plaquemine locality accepting Cahokian influences (Brain 1989:117), most of the occupation is fully Mississippian and dates to the thirteenth and fourteenth centuries. The construction of mounds and the site's paramount status seem to have ended with the ca. 1400 burning of mound-top structures, although occupation of Winterville continued until about 1500 in conjunction with a number of smaller, younger ceremonial centers to the south along Deer Creek (Brain 1989:110). Its only regional rival in volume of earthworks was the contemporary Carson complex, in Coahoma County, across from Old Town Lake, and the Lake George site in the extreme southern Yazoo basin. Architecture recorded at Winterville is typically Mississippian: wall trenches, split cane wattle, and grass thatch on horizontal rafters. Brain (1988:63) describes exterior finishing coats of a 2 to 3 mm thick mud plaster. With daub fragments averaging 5 cm thick and both sides of the wall being plastered the walls were 20 to 30 cm thick. While he believes that the Winterville people were "remarkably consistent in adherence to architectural rules", the excavations on the summit of one of the largest mounds also showed lines of 20 cm diameter singly set posts (Brain 1988:64, 51). Burials were both bundled bones and interred whole corpses; one individual showed marked bowing and swelling of the tibias typical of rickets, a nutritional deficiency (Brain 1988:69) and trepenomal syndrome. Artifacts from the Early Mississippian levels included clay elbow pipes, clay "ear plugs" (to hold feathers or other ornaments in large piercings), pebble celts, and conical bone and antler arrow points, none of which occur in the small Ellis sample. Brain (1989:138-140) discusses varieties of Barton Incised from Winterville, including Barton, which has sharp-cut, careless, hatched or crosshatched patterns. The southern varieties Estill and Arcola have not been recognized at Ellis. Barton Incised is assigned to the Yazoo 2 ceramic set, a "catchall [representing] an initial hybridization of ceramic traditions" (Brain 1989:85) during the Winterville I phase. Few Yazoo 2 treatments other than Barton Incised and notched and cut-out "plate" rim are common to the Winterville and Ellis collections. Most of the red slipped materials at Winterville appear to date to the Early Mississippian component, but there is some Nodena Red and White, and late in the sequence, some Bell pastes. Of the five varieties of Winterville Incised from the Winterville site, only var. Winterville is represented in the small Ellis collection. Of the ten specimens illustrated by Brain (1989:161) none duplicates the guilloche represented at Ellis and Oliver, but the schematic drawing of the Yazoo 3 set in Figure 61 (Brain 1989:84) shows it clearly. Yazoo 3 "exemplifies the basic (i.e. most universally ramified) Mississippian expression of decorative intent and ware" (Brain 1989:85) and dates to the Winterville II phase, ca. A.D. 1350-1400. Overall, the similarity between the Ellis and Winterville assemblages is only general. As at Chucalissa, fifty miles north of Ellis, specific whole-vessel or mode combination analogues to the Ellis material are hard to identify. Decorated types such as Barton Incised and Winterville Incised are common to both sites, but in minority types and varieties they differ markedly. The Greenville and Holly Bluff (fine grog tempered Bell-equivalents) sets are missing on sites to the north, although there is a tendency for fine shell, fine shell and grog, and fine grog pastes to intergrade throughout the region. Ellis lacks the steeply carinated bowls similar to Caddoan forms and many of the incised and engraved types used at Winterville. The Yazoo 3 and 4 sets which should be contemporary with the later part of the Ellis occupation include the gently curving punctated and incised jars, but the varieties present at the two sites differ and painted ceramics are scarce in the Late Mississippian assemblages at Winterville. We have, therefore, a nearly 100 mile stretch of highly similar fifteenth century material culture. Only at the far north and south (here characterized by Chucalissa and Winterville) do ceramic type frequencies, vessel form, and other characteristics diverge to any great degree. Triangular points, pebble celts, and imported hoes are the mainstays of the stone tool industries. Mounds, arranged around rectangular plazas, accumulated rapidly. Mound and village architecture seem to have originally been founded in wall trenches, but single set post construction gained in representation as time went on. This is in general agreement with the belief gleaned from the A.D. 1542 DeSoto accounts that there was no sharp linguistic break between the provinces of Quizquiz, Casqui, Pacaha, and Aquixo (Hoffman 1990:214). Morse and Morse (1983:308-310) have suggested that these political provinces can be identified with the Walls, Parkin, Nodena, and Kent/Old Town ceramic phases, respectively.

THE OTHER MISSISSIPPIAN COMPONENTS IN PHILLIPS COUNTY

We will now return our attention to Phillips County to recount the extent of information available on the 50-odd other recorded Mississippian components. The evidence from some of these sites is minimal and I will leave it to the reader to decide if a Mississippian component is actually present. The sites are arranged in a north-to-south order by geographic units (Figure 19). Three sites at the mouth of the St. Francis, including two Phillips (1970) included in the Kent phase will be discussed first, followed by the Crowleys Ridge and Wisconsin terrace sites, the Old Town Lake area sites, sites in the White River backswamp, and finally, the southernmost Bee Bayou sites traditionally associated with the Quapaw phase. I will give the history of the research; describe anything known about the site setting, layout, and materials; and offer any comment I feel is warranted about the chronological placement of each site. For many sites the evidence is quite scant, and for some of the most important sites, the record is confused and indicative of continual site degradation. In most cases, I am unable to make an estimation of Mississippian site function based on the evidence at hand, but I feel it is worthwhile to compile all available evidence, as many of these sites have probably been destroyed since they were recorded around 1970. Very few professional revisits have been undertaken, and then generally only on the report of looting or landlevelling. The following discussion is based largely on LMS and AAS site files. Many of the sites discussed here were reported by John Leuken, an amateur archaeologist. A few were recorded or revisited during cultural resources inventories by private contractors. Other agencies have carried out very minor amounts of assessment and conservation. I believe that many of the recorded sites in the county could be found to have Mississippian components, as there are many reported sites that only have "pottery", "arrows", daub, or mounds reported from them. I have selected to describe as Mississippian components any site with Mississippian diagnostics, chiefly shell tempered pottery. In a few cases lithics, site plan, daub, or report of pothunting were used to assign a Mississippian component. Some of the sites were large, the result of many occupations, but have only produced a small amount of Mississippian material. This is particularly true on the older terrace surface, where some Archaic-Woodland scatters appear to have had Mississippian farmsteads or linear hamlets overlying them. Phillips County appears to have as spectacular Marksville and Cole Creek period manifestation as Coahoma County across the river. Several sites will be mentioned which are of uncertain location. Mostly, they are presumed destroyed. People surveying in the vicinity of these sites should be particularly alert for fragmentary deposits. The Smithsonian Institution surveyors dug into two mounds, Barney and Rodgers, probably between Modoc and the Mississippi River levee. A site may have been removed for levee fill at Ferguson Landing. Mississippian occupation of the city of Helena may have been more extensive than the ephemeral component recognized in the Woodland mounds at the end of the ridge, as an early nineteenth century geologist seems to have unidentified a daubed palisade there and there is a persistent local tradition that the Stephenson collection came from in or near the city, perhaps at what was later Ft. Curtis, rather than, say, the Old Town Lake earthworks or the Fitzhugh mound. The St. Francis River Sites Three sites have been recorded along the base of Crowleys Ridge at the mouth of the St. Francis River. The first two recorded, both damaged by high water and levee building, were included in the initial formulation of the Kent phase (Phillips 1970). The third site was reported by a convicted looter as part of his court-ordered restitution. In response to the Smithsonian Institution's Circular 316, mounds at the mouth of the St. Francis River were examined by C.H. Boyd, according to abstracted replies (Mason 1880). Also in the general vicinity, on Big Prairie, later known as Prairie Point, Plum Orchard Point, and Quarles Island, remnants of which form an island below the mouth of the St. Francis, a "large Indian burial ground" was found in the early-to-mid nineteenth century (Kirkman 1964:4). This may or may not be the same site mentioned as "Indian graves" on an estate 5 miles from Helena that John Quarles bought from John Patterson before 1852 (Griffin 1969:8). The latter site, an Indian burial ground and mounds, is referred to as being five miles below Helena in two newspaper articles, without date or publisher, in the Phillips County Library clippings file. The mention of prairie and a plum thicket makes this sound like a recently abandoned village. It is noted in the histories of early Phillips County that Patterson, a son of one of the earliest White occupants of the county, grew up in an Indian village and could speak an Indian language as a boy (Phillips County Historical Quarterly 1962:1; Kirkman 1964:4) Moore (3PH7/14-N-1). The Moore site, a large late prehistoric site with a cemetery, lies on the river bank near the mouth of the St. Francis River. Phillips et al. (1951:50) describe Moore as a large village site with a small mound. This heavily impacted village suffered a great deal of pothunting up through the 1980s. It has also been damaged by the action of the St. Francis River. There has been much discussion about how to manage the Federally-owned site, but the main objective is to keep looters off. Hoffman's (1975) assesment of the Smithsonian Institution collections from Arkansas notes "small fragments of gray buff ware-slight sand temper…,[other] sandy pastes…, some red paint…, yellow brown flint…[and] chunks of burned red clay with grass and reed impressions." The Smithsonian site definition, perhaps based on Boyd's aforementioned correspondence, appears to have included the Steagal site on the opposite bank. Phillips and Motz of the LMS recorded the Moore site in 1940 as an oval mound, 100 by 60 foot and 4 foot tall, covered with daub and surrounded by a village site with Woodland and Mississippian materials. Their collection included 199 Neeleys Ferry Plain, 35 Old Town Red, 19 Bell Plain, 17 Parkin Punctated, 13 Barton Incised, 4 Nodena Red and White, 3 Carson Red on Buff, 1 Hollywood White, and 1 unclassified shell tempered punctated and/or incised. U.S. Forest Service (USFS) archaeologist Earl Neller evaluated the site after efforts had been made to eliminate looting of the site, including convicting three pothunters. He counted 151 potholes, probably from looting during the winter of 1977-1978, some beginning to silt in and others deep enough to show site stratigraphy. He noted 8 to 10 inches of recent alluvium covering the top of the prehistoric deposits and a house floor "at considerable depth in one of the potholes" (Neller 1978). According to Neller's estimate, the Moore site measures at least 300 by 200 foot and has deposits as deep as 2 foot below the surface. Robert Ray (1978), USFS, estimated the site to measure 200 by 75 m along the bank. He noted charcoal and bone at the north end of the site, and charcoal, bone, and sherds at the south end. According to Ray, there may be a burned floor at the center of the site. He also notes the possibility that the site is partially buried by post-Mississippi period alluvium and that it may be deeply stratified by periodic flooding. The AAS site file extension shows that about the same time, John House of the AAS visited the site to make a map and evaluate the damage done by looters. He noted that Phillip's mound apparently is the low rise, 25 m in diameter and .5 m tall, on the west edge of the ridge along the riverbank. The site boundary was defined based on several hundred potholes along 200 to 300 m of high ground, most of which were less than .5 m deep. In contrast to the rich deposits reported by the LMS, House noted sparse material in the backdirt and little evidence of intensive occupation except daub exposed where a road crosses the mound. As the land is administered by the USFS, a Federal agency, a small amount of material was salvaged from potholes for purposes of evidence after a looter was arrested, eventually to be convicted. The small quantity of material used in this trial is stored in at the Ozark-St. Francis National Forest headquarters in Russellville. Materials collected included four tested cobbles, an early stage preform, and three decortication flakes, all of local Crowleys Ridge gravel. Small daub fragments included sandy material with vegetable temper imprints and clayier material without evident cut grass temper. All the daub fragments were small and eroded, lacking surfaces or other diagnostic traits. Three large, well-preserved deer vertebrae were collected. The USFS sherd collection consists of 13 Mississippi Plain, 4 Old Town Red, 2 Bell Plain, 1 Carson Red on Buff, 1 Avenue Polychrome, and 1 Parkin Punctated. The Mississippi Plain sherds included some with adhering charred material, one of which was a rim sherd from a deep or helmet bowl. The slipped materials include coarse paste red sherds, two of which are rims with sooting as well as examples of fine, heterogeneous paste polished sherds. The two sherds classified as Bell Plain are a body sherd from a bowl inflection point, with some fairly coarse grog in the paste, and a polished, more typical, Bell body sherd. The small collection at the University of Arkansas, Pine Bluff, (UAPB) has 8 sherds of Mississippi Plain, 1 of Barton Incised, 6 of Old Town Red, and 1 of Owens Punctated (Figures 20 and 21). The Barton sherd, from a large jar, has steep, wet-cut line filled triangles while the Owens sherd has triangular fields of nail punctations zoned with sets of incised lines. Both appear to be from curving shoulder jars. Two of the red sherds are from small flaring rim bowls. There is also a mussel shell effigy (spoon?). A Kent phase attribution for the site seem reasonable based on geography and the small amount of available cultural material (310 sherds). The component appears to be roughly contemporary with and probably somewhat later than that of Ellis, that is, extending into the sixteenth century. Much of the material (flaring rim bowls, inflected Bell bowls, deep cooking bowl/jars, curving-shoulder jars, and painted bottles) coincides with that from the dated fifteenth century sites, but there are also somewhat later types such as Carson Red on Buff, Avenue Polychrome, and Owens Punctated. Steagall (3PH8/14-N-2). This site was largely eroded away when it was recorded by the LMS, and supposedly very little is left of the Steagall site today. Phillips et al. (1951:50) describe Steagall as a village site. Impacts are believed to be chiefly scouring and levee construction, but there has probably also been alluviation, suggesting the potential for buried deposits. It lay downstream, on the opposite bank from Moore. Hoffman's (1975) annotation of the Smithsonian Instirution catalog cards on file at the AAS, Fayetteville, appears to note 3PH8 as one part of a village site on both sides of the river at the end of the St. Francis levee, indicating that they included it with Moore. The Smithsonian collection lists only small fragments of yellowish brown flint, chunks of burned red clay with grass and reed impressions, and gray buff ware with a slightly sandy texture. Based on a few shreds, this village site with thin daub and other evidence of occupation, was assigned to the Middle Mississippi period. Based on Motz and Phillips small sherd collection (36 Neeleys Ferry Plain, 1 Barton Incised, and 1 Parkin Punctated) and position at the mouth of the St. Francis, Phillips (1970) assigned this site to the Late Mississippi period Kent phase, a highly tenuous attribution. Phillips et al. (1951:Figure 95) show a very small Barton Incised jar from "north of Helena Crossing, mouth of the St. Francis", giving the site number as 14-N-6 (Helena Crossing Mounds), but it seem likely this pot comes from Moore or Steagall. The LMS site card notes that they believed Steagall and Moore were being referred to in Thomas (1894:233) when he describes a site where pottery and human bones were being eroded away on the lower portion of the St. Francis "especially opposite Phillips Bayou…in such a state of preservation as to indicate that they, as well as many found on the west side above the bayou, pertain to a comparatively modern period….some of the earliest settlers [state] that when first occupied by the whites it was not an unusual thing to plow up fragments of bark boxes or coffins, together with bones and pottery." However, the early report may refer instead to some southern Lee County site No collections from the site were encountered in this research. It should not be assumed that the Steagal site has been completely destroyed. I find no firm basis for assigning it to the Late Mississippi period, as Barton Incised and Parkin Punctated could just as easily date to the later Middle Mississippi or Protohistoric intervals. County Line (3PH233/3LE139). As the name implies, this small site lies partially in Lee County, on a low, 20 m wide spur descending from Crowleys Ridge to the St. Francis River on the St. Francis unit of the Ozark National Forest. USFS archaeologist Gary Knudsen (personal communication 1997) believes the location to be the site of a single late prehistoric house, based on the small size of the ridge top. It was reported in 1978 to Knudsen and Ann Early of the AAS by a convicted pothunter who as part of his sentence was required to show USFS other sites he was familiar with. He stated that he had taken pottery, presumably from a grave, from a depth of three foot. The site had been damaged by other looters, and some reconstructible vessel fragments were retrieved by USFS personnel during a brief damage assessment. Most of the impact appeared to be small test holes, but there were many of these, both on the level area and upslope. The extent of pothunting does not agree well with the assessment of County Line as an isolated house. The upslope area was further assessed by shovel testing by Michael Pfeiffer preliminary to a timber sale in 1990. Some artifacts were exposed in eroded areas of Memphis-Natchez association loess soils. These are described as three pieces of pottery with incising and no apparent temper and four with smooth surfaces and a small amount of grog. The vessel fragments, a Barton Incised var. Kent/Kent Incised jar, an Old Town Red or Carson Red on Buff bowl, and a small globular bottle, are curated in Russellville (Figure 22). The coarse shell tempered Kent Incised jar has a number of sixteenth century attributes: a sharply outflared lip (Memphis rim mode, John House personal communication 1997), eight appliqué arcaded "handles" with fine-line incising, and sharp body-neck junction (Oliver jar form, after Belmont 1961). The Kent Incising is, by definition, vertical lines on the body of the jar. In this case they extend slightly below the point of maximum diameter. The jar has a flattened globular or squat body about 22 to 24 cm in diameter; its combination with these other diagnostic traits indicates that this form also dates to the 1500s at the earliest. The Old Town Red/Carson Red on Buff deep helmet bowl has a squared lip, slight interior rim thickening, unmarked inflection point, and coarse paste. It is about 20 cm in diameter. The slip is eroded, but the red covers the interior and exterior of the rim. The interior body sherds have some red, but it is impossible to tell what, if any, pattern existed. The plain bottle has a squared, outflaring lip, very short neck, and fairly fine Mississippi Plain var. Neeleys Ferry paste. The body was apparently globular and about 15 cm in diameter. The site appears to be later than Moore, i.e. Protohistoric, although the County Line vessels offer a much better potential for comparative dating than the sherds from Moore. The Crowleys Ridge and Wisconsin Terrace Sites Fourteen sites will be discussed here. The Hall site (3PH12) is an ambiguous, probably early, mound on one of the main tributaries of lower Big Creek. There is a cluster of six sites around Marvell (3PH6, 39, 54, 115, 116, and 117), most presumably farmsteads and hamlets but including one village with mounds. Sites 3PH5, 3PH79, and 3PH148 lie in a similar setting to the east and 3PH51 and 3PH81 are at the western base of Crowleys Ridge, opposite the mouth of the St. Francis River. Based on findings in other parts of northeastern Arkansas, these sites should date to the Middle Mississippi period, as the braided stream surfaces of the Western Lowlands were abandoned ca. A.D. 1400 (Morse and Morse 1983:283). However, the presence of late types such as Bell Plain and Old Town Red on some of these small sites may be an indication that occupation of the old terraces continued later in this more southerly region. The Marksville-Hopewell Helena Crossing Mounds (3PH11) at the end of Crowleys Ridge also appears to have had a minor Mississippian component, as does the nearby Bowie Village site (14-N-4). David Dale Owens' (1860) geological study describes a site four miles west of Helena, at the terminus of Crowleys Ridge, where eroded embankments "built of sun-dried clay mixed with the stems and leaves of the cane… the clay and stems of cane appear[ing] to have been mixed together and molded into a wall, somewhat after the manner of pise", refuting the local impression that the walls were made of fashioned brick. The wall enclosed a number of small mounds and the northern boundary of the enclosure was made by the hills. Owens may have been referring to the Helena Crossing site. There is some indication that the city of Helena may have erased a mound group. The Judge Marshall Stephenson collection, partly preserved in the Lemley collection at the Gilcreast Museum in Tulsa, Oklahoma, and partly retained in Helena and displayed in the public library, according to many local sources came from mounds in the city of Helena and within an hour's walk around it. There is repeated reference in the Phillips County Historical Society materials, perhaps just local legend, that the Ft. Curtis site, now destroyed, was the site of one of the principal villages invaded by DeSoto. However, none of these make mention of pots or graves being found at this Union army fortification, which has itself now been destroyed. Turner (3PH5/14-M-1). This small village reportedly had a one and one half foot high, plowed house mound on the bank of a small tributary of Big Creek when Phillips and Motz of the LMS visited it in 1940. The mound was eroded with considerable daub showing, with only two small shell tempered sherds and a blade being found but saved. Phillips et al. (1951:50) describe the site as a small village site with mounds. No collection was available for this study. The location deserves revisiting. R.L. Broom (3PH6/14-M-3). This large (ca. 5 acre) village site with mounds includes Mt. Meriah Cemetery where the recent graves showed pottery (Phillips et al. 1951:50). The largest mound, which had been leveled by 1940, was reportedly 10 foot high and contained skeletons with pottery vessels, some of which were reported to Phillips and Motz to have been dug up by an Indian for the St. Louis World Fair. Phillips et al. (1951:313) give the mound dimensions as 100 x 10', 36 x 2', and 45 x 2', in an irregular arrangement. The site card gives the dimensions as 12 m diameter and ½ m tall and 15 m diameter and ½ m tall, also noting that more burials and pots had been found when the farmer's father leveled the large mound and that material was said by the farmer to be moderately abundant on the site. However, surface collecting conditions appear to have been poor and the only Mississippian materials collected by the LMS, as shown on their tabulation form, were 6 Neeleys Ferry Plain sherds. The site card notes that two shell tempered sherds were found in the road near Mound A, with the rest of the collection being clay tempered. John Leuken next reported the site in 1969, noting that many others had also collected this large Woodland site. The UAPB/Leuken collection (69-192, 90-501) has one sherd each of Mississippi Plain and Bell Plain, three arrow fragments (Figure 23) and a celt. The presence of human teeth, deer bone, and mussel shell indicates faunal preservation in parts of the deposit. The site has Early Archaic through Coles Creek period components, in addition to the ephemeral amount of Mississippian material collected by Phillips and Leuken. The gray chert sub-triangular (similar to Madison var. Russell/Nodena var. Banks) arrow point and celt fragment may be attributable to the undifferentiated Late Woodland period occupation(s), however it seem likely that Broom was a large Mississippian village site as well, based on the potted mound. Although the site has been heavily damaged, it is worthy of a revisit. Helena Crossing Mounds (3PH11/16-N-6). This cluster of conical mounds lay at the south end of Crowleys Ridge, overlooking what is now the foot of the Mississippi River bridge. The LMS recorded the site in 1940, noting that local people did not realize that this was a mound group. Some "wattle" was noted between Mounds B and D, and some on Mound B, but not much. Sherds were scarce at the site, but did include some minor amount of shell tempered (7 Neeleys Ferry Plain and 3 Old Town Red). Two of the Helena Crossing mounds (B and C) were salvaged by James Ford, who notes no additional shell tempered material in his report, however, most of the site had been destroyed by that time and he excavated Mound B, the likely locus of the Mississippian component, with a bulldozer. He did note a small amount of human bone near the surface of Mound B, these may have been intrusive later burials (Ford 1963:41). John House (personal communication 1997) has recently examined the Helena Crossing materials and noted no shell-tempered materials. The location continues to suffer degradation, and it appears to be nearly completely destroyed. The nature of the Mississippian component will probably never be ascertained. Hall (3PH12/15-L-3). The LMS recorded this large rectangular platform mound (Phillips et al. 1951:51). The AAS site file extension shows that in 1978 John House of the AAS visited the Hall Mound, finding nothing on the surface, but making a map of this 4 m high, rectangular flat-topped mound on a wooded terrace edge. A few old pot holes were evident. The corners of the mound are aligned with the cardinal directions, indicating an early date (John House personal communication 1997). The dimensions are 30 m square at base and 18 on the top, with the height being 3.5 to 5 m. The site may be a Coles Creek period Plum Bayou culture or Early-to-Middle Mississippi period vacant ceremonial center. It deserves further investigation, in particular, shovel testing or test pit excavation to recover a material culture sample. Shadden (3PH39). This three acre Late Archaic-Late Woodland camp midden on Little Cypress Bayou also produced a single Mississippi Plain body sherd when collected by John Leuken (69-119, 90-505), who also reported a "mound" on the north end of the site. The hundred or more points from the site included two unusual lithics, neither unequivocally attributable to the Mississippian component: an end scraper with polish over the flake scars made on a steeply retouched yellow chert flake (this artifact more likely dates from early in the Archaic period as it does not closely resemble the Protohistoric endscraper of the Oliver lithic complex) and a white chert or novaculite small lanceolate biface, possibly an arrow made on a retouched flake. Shadden also has other Early Archaic materials and small Coles Creek period arrows. The ambiguity of the Mississippian component and the difficulty of finding ephemeral Mississippian farmsteads overlying earlier middens makes this site a low priority for research. Otey Miller (3PH51). This site consisted of two loci, 200 yards apart near the foot of Crowleys Ridge. The larger location lay over a small lake and the other was on a small rise downslope. The area was new ground when Leuken reported it in 1969. Four sherds of leached Mississippi Plain and two possible arrow point preforms are curated at UAPB (90-523). The site may be particularly significant if it had a lithic procurement function during the Mississippi period, however, it could have been a farmstead with an only secondary emphasis on gravel chert. Max Hirsh (3PH54). This site was also new ground when John Leuken recorded it in 1969. This small site near Little Cypress Creek produced Woodland points and two leached Mississippi Plain sherds, one probably from a jar shoulder (69-199, 90-505). Herman Hall (3PH79). This scatter of daub and clay tempered sherds lies on a series of rises or ridges along an old channel scar lake. In 1969, John Leuken reported the site as large and prolific, but only obtained a small collection (69-753). Four sherds of Mississippi Plain, one indicative of a jar, and one of Bell Plain var. Nickel were recovered, along with Big Sandy and San Patrice cluster points. The Bell Plain probably indicates a fairly late component. Melio Farm (3PH81). This small site on the west edge of Crowleys Ridge lay at a former spring. There was a large amount of lithic material around a deep dry stream bed. Points, metates, and two leached shell tempered body sherds were collected by John Leuken around 1970 (70-13, 90-554). The setting is similar to that of Otey Miller. Louisiana Purchase Marker A, B, and C (3PH115, 116, and 117). A total of nine sherds of Mississippi Plain is reported from this series of farmsteads. Site A was a small scatter on a sloping small ridge overlooking Cypress Creek. An arrowhead is reported from the site, along with at least 5 sherds. Site B, similarly a small scatter on a ridge sloping towards Cypress Creek produced points and sherds, including one with shell temper. Site C, similarly located, produced "scrapers, chips, sherds" including angular, leached, possible shell tempered sherds. It seems from a CRM perspective that this archaeological manifestation should have been considered potentially significant, and it should be given priority for revisit, if it still exists. An Early or Middle Mississippi component is to be expected, if the pattern of isolated houses being early holds up in the Phillips County area. Abandoned Railroad (3PH148). This site lies on an east-west trending high terrace near an old creek run. In the earliest CRM venture in Phillips County, Jeff Fleniken (1974) reported it as chiefly Archaic and Woodland, but five sherds of Mississippi Plain, one of Bell Plain, and one of Old Town Red (coarse shell tempered) were collected (74-162). A minor Middle or Late Mississippian occupation is indicated by the small collection. Bowie Village (14-N-4). This large Marksville and Baytown period site produced a Parkin Punctated sherd and an unclassified shell tempered sherd when it was visited by the LMS in 1940. The site lies in the bottom below the south end of Crowleys Ridge, adjacent to the Missouri Pacific railroad. It was two foot thick then, despite sheet erosion from long cultivation. 14-N-4 was tested by Ford at the same time as the Helena Crossing Mounds, but not reported in that volume (Ford 1963:6). He spells it Bouie, stating that he dug a test trench which demonstrated that the Bowie site was of a slightly later date than the Helena Crossing Mounds. The Stage 2 channel that lay at the foot of the Ridge passed by Bowie also (Ford 1963:8). This site appears to have had an ephemeral Mississippian occupation overlying a larger early midden. Backswamp and Big Creek Sites There are 13 probable Mississippian sites in this region. Seven components (3PH13, 45, 56, 60, 152, and 240) have been identified in the backswamp between the modern Mississippi River meander belt and the White River and Big Creek. All are from fairly tenuous site descriptions, but it is noteworthy that three have rough discoidals, perhaps more representative of Cole Creek period than Mississippian culture occupation. 3PH112 lies near the alluvium east of Old Town Lake and 3PH128 lies on a bayou that is a crevasse draining south out of Old Town Lake. Thomas' (1894) Barney and Rodger's sites also lay in the lower-lying region south of Old Town Lake now known as Modoc. There are three sites (3PH17, 41, and 239) near Big Creek, which runs southeast to the White River along the southern edge of the Wisconsin terrace. The Woodland culture conical burial mounds in Elaine may have also had a slight Mississippian component, like Helena Crossing, as the LMS reported a Barton Incised sherd from 15-M-2 and a Parkin Punctated sherd from 15-M-3. Barney (15-M-5). Thomas (1894:235-236) describes a compound, two-tier, oval mound with a surrounding ditch where "patches or beds of clay burnt to a brick red" were uncovered. The platform was 15 foot high and the mound atop it an additional 20 foot high, with the ditch being 10 to 15 foot deep and 50 to 75 foot wide. The cultural affiliation is uncertain and Hoffman's (1975) notes on the Smithsonian Institution documents state that the site as destroyed by levee construction ca. 1891-1900 and that bones, pottery and flint were found, but not saved. The site was reportedly five miles east (this is unlikely, as it places it in Mississippi) of Old Town and an equal distance from the Mississippi River and Beaver Bayou, with the Rogers site lying a mile south of Barney. The LMS assigned the site a number, apparently without visiting it, although they do give a legal description location, which also falls in Mississippi. Rodgers. Thomas (1894:236-237) describes a group of house mounds surrounding a 247 by 200 foot flat mound 20 foot high with a 50 foot diameter, 5 foot high house mound on its summit. There was a recently abandoned house and barn on the mound and "a heavy fire-bed was found immediately below the surface of the upper mound". The next smaller mound also had a bed of burned daub, which was hacked through, revealing "charcoal, ashes and flakes of mortar…retaining the casts of the stems of grass and canes. Two foot below this was another fire-bed." These may be the mounds near Modoc, caved into the river, mentioned in the catalogue of the Smithsonian Institution collections (Hoffman 1975). Tinsley Mound (3PH13/15-M-1). This site was recorded in by the LMS in 1940 as a village site with three small (50-150 foot diameter, 4 foot high) house mounds (Phillips et al. 1951:51). The plaza length is given as 200', oriented east of a 4' high mound, with other possible mounds 50 foot in diameter and 1 to 4 foot high, sparse daub and moderately abundant refuse (Phillips et al. 1951:322). Tinsley appeared to be predominantly Woodland with a small percentage of shell tempered sherds, but it has been crossed by a levee and drainage ditch, with an unknown amount of site destruction resulting. The site plan sounds more like a Mississippian culture site than a Late Woodland one. No modern collection was available for this study, but the LMS reported "briquettes" (daub) and a burial exposed by plowing in Mound A. They obtained a fairly sizable sherd collection (398 Neeleys Ferry Plain, 5 Parkin Punctated, 3 Barton Incised, 3 Old Town Red, 2 shell tempered cordmarked, and 1 Kent Incised). The possible Cahokia Cordmarked is important in indicating a Early Mississippi period component, in contrast to the Late Mississippi type Kent Incised. Given the location, the site should be easily re-identifiable. Vogel (3PH17/15-M-5). Phillips et al. (1951:51) describe the Vogel site as low mounds in woods, with no surface visibility. In 1940, Phillips and Griffin were informed that John Lining, a storekeeper at Wabash, had pottery dug from one of the mounds. The site was crossed by the new White River levee. Mound A was 200' in diameter and 3-4' high, Mound B was 250 by 400' and 3-4' high, and Mound C was not visited. They note that the mounds gave little indication of artificial origin, besides pots being taken from them, and that they seemed "too large". As it too lies along a levee, it should be relatively easy to relocate. The UAPB collection (69-198) has only Archaic and Woodland materials. Graumann & Heskett (3PH41). This 40 by 60 m scatter followed a Dubbs silt loam or sandy loam ridge that had not yet been leveled when it was recorded by John Leuken in 1969. The area was high ground south of Old Town Lake with a barn and small cemetery on it; tools, points, and sherds were found (69-193, 77-968, 90-515). Leuken returned to guide a cultural resources survey party to the site (Dorwin and Stewart 1977). They concluded that the site was Baytown or Coles Creek in origin, however, among the Archaic and Woodland materials at UAPB are a rounded-corner triangular point fragment (Madison var. Russell/Nodena var. Banks) of pink chert and a small flake with hoe polish that may be attributable to a Mississippian component (see Figure 23). Willow Tree (3PH45). John Leuken reported this site near the White River levee as lying on a ridge above a dry slough. The site extended for about a quarter of a mile along this ridge and was partially new ground in 1969. The tenant, Dickson, reported that he had dug 65 burials and sold all the pots around 1968. Perhaps one was sold to John Leuken, as there is a photograph (AAS Negative 892513) of an apparent ""Tippets Bean Pot" from the portion of the Leuken collection not deposited with the state. Although the handle is missing, it has the characteristic placement location below the rim, fine brown paste, and cylindrical sides and flat bottom. It is about 12 cm in diameter. The UAPB collection has Woodland ceramics (Indian Bay Stamped, Larto Red, fancy Mulberry Creek Cordmarked rim), bone and shell, quartz, and five shell tempered sherds (one from a small, thin, inslanting jar), a Madison var. Russell arrow point, and an igneous discoidal (69-194, see Figure 23). The Mississippian component is presumably fairly early given the thin Mississippi Plain and the discoidal. This interpretation is supported by the "beanpot" attributed to the site, as it is a classic 1200s marker in the American Bottom. Hardesty (3PH56). The Hardesty site was reported in 1969 as an isolated, cultivated house mound or natural ridge with a large amount of daub with cane impressions. No additional information is available. The location should be revisited to determine if any cultural deposits remain. Montgomery #2 (3PH60). This site at a modern headquarters had a conical mound that had been destroyed for an airstrip by 1969. The reporter, Billy Bebee, noted "I could cry." Most of the material was Woodland (Mulberry Creek Cordmarked, Mazique Incised), but 16 sherds of Mississippi Plain and 1 Barton Incised sherd were also recovered (69-215). The typical Barton body sherd from a curving shouldered jar has a lower bounding line and line filled triangles. Plain sherds indicated jars and helmet bowls. Some of the bone, shell, cut antler, and human remains in the collection may also be attributable to the Mississippian occupation. Happy/F.F. Kitchen Farm (3PH112). This 1970 Leuken collection (70-414, 90-568) has one Mississippi Plain sherd and three fine grog tempered sherds (probably Coles Creek period Baytown Plain rather than Bell Plain) along with a triangular point fragment (see Figure 23). The site was a conical or house mound, a rise coinciding with a tenant house scatter, on a ridge in the land sloping away from Old Town Lake. Material was concentrated on the rise, in a 50 by 100 yard area. Quartz, points, flakes, and a drill were also reported but are not curated at UAPB. J.D. (3PH128). This tentative Early Mississippi component might be better called Archaic or Woodland; stone was the main material found. The location lies on a small, cultivated rise in the ridge and swale area northwest of Old Town Lake. A quartzite or other metamorphic imperfect discoidal and small, rectangular, green and brown metamorphic celt (90-576) probably date to the Coles Creek period around A.D. 750 rather than to the Mississippi period. The only other materials collected here by John Leuken were "nutting stones" and chert. Snowden/Howe Mound (3PH152). This somewhat dubious Mississippian component designation is based on the report of a chunkey stone in a site report made by a student, Kylene Reece, in 1973. Several other large stone tools were also reported (mortars, anvils, hammers). The site was in beans and had been cultivated for years and leveled, so it was hard to distinguish from the surroundings, but it had once had a 4 foot elevation. The location is vague, but may be worth attempting to re-locate. Bridge Replacement Borrow Pit (3PH240). An Arkansas Highways and Transportation Department cultural resources survey reported this very late Woodland or Early Mississippian small farmstead scatter as following a slight, partially sand-covered, ridge (Williamson 1989). David Williamson recommended moving the borrow area to avoid the site. The subsoil was gray clay, with a ditch lying along one side of the site. Few artifacts were visible in a 200 by 40 m area, but a tight concentration of daub with cane impressions (N=15), a stemmed dart point, fire cracked rock and other lithics (N=9), and a very concentrated area of "grit and shell tempered" pottery (N=7) were collected (88-810). Trenton-McCluskey (3PH239). This site identification is made based on Ian Brown's (1978) description of the J.O. Wheeler vessel collection from an allegedly extensive (5 acre) aboriginal cemetery that produced Late Mississippian materials. Three pots were recorded: 1) a Mississippi Plain bottle with tall, straight neck with flaring, tapered lip and collar at the base on a globular-to-high-shouldered body; 2) a small Mississippi Plain hemispherical bowl with slightly flaring rim with wide-spaced notches; and 3) an Old Town Red bottle with pedestal base, flattened or almost carinated body, and short, wide neck (Figure 24). John Leuken visited a location he believed to be Wheeler's site and collected 11 Mississippi Plain sherds, an Avenue Polychrome sherd, and Woodland materials (90-584). Although this pothunted village seems to be a high priority for investigation, no further visits have been made by professionals. Around Old Town Lake This area comprises the prime recent alluvium in the county, and it is today largely in cotton production. At least seven areas with evidence of occupation have been defined north and east of Old Town Lake (3PH18, 19, 38, 119, 160, 255/256, and 257). These include the Ellis Mounds site. There are ten sites reported along and inside this large, recent oxbow lake (3PH20/21/153, 40, 42, 44, 46, 48, 87, 88, 89, and 133). The most significant is the Old Town Works (3PH20/21/153) at the southern cut-off point of the lake. Many of these are large sites, some with earthworks, and many of them have been pothunted. This appears to have been the most significant area of Late Mississippi period occupation in the county. Owens (1860:415) describes Sugar-tree Ridge as a "considerable district which lies in an elbow formed by Long lake" and "elevated a few foot above the overflow of the Mississippi". Mounds and traces of fortifications were found on this land, along with skeletons, pottery, arrowheads, and stone axes. The degree of destruction of the Mississippian sites along the river can be seen in Hoffman's (1975) index of the Smithsonian Institution collections, which mentions two mounds off Westover Landing (part of the Helena Harbor project area, Childress et al. 1995), 200 foot into the river ca. 1900-1932. Fitzhugh (3PH18/15-N-1). In 1883, Edward Palmer reported a Goach Mound, six miles below Helena, with a recent cemetery on it (Thomas 1891). This may be the Fitzhugh site. Griffin, Motz, and Phillips of the LMS recorded Fitzhugh as a village with small mounds, heavily damaged by a levee breach, stating there was very little left to describe in 1940 (Phillips 1951:51). The plaza of this large ceremonial center appeared to be 550' long, oriented south-southeast of a stepped rectangular mound 20' tall facing another square mound about 9' tall, with daub abundant but other refuse sparse (Phillips et al. 1951:327). The LMS obtained a small surface collection (15 Neeleys Ferry Plain and 1 each of Barton Incised, Parkin Punctated, Ranch Incised, Old Town Red, and Nodena Red and White), but did note about a foot of occupation debris in the bank of a wash. The site was initially considered representative of the Middle Mississippi culture, and was considered a Late Mississippi period Old Town phase site by Phillips (1970), but has been entered in the AAS files as Protohistoric. The reference to Ranch Incised and Nodena Red and White supports a Late Mississippi period occupation at least. The AAS apparently does not have a collection from this site, but James Harcourt (1991) attempted to relocate the site during the Helena Harbor survey. The LMS- plotted location of the site was found to be a large lake or borrow pond, and the AAS crew dug shovel tests at a 10 m interval in a 20 acre pasture around the lake. All the shovel tests were negative, and after surface inspection of surrounding fields with low visibility, Harcourt concluded that the site had been destroyed, as the pond was not shown on the 1939 Farrell 15' quadrangle (the edition probably used by the LMS). No work was done east of the levee. Although intensive survey failed to relocate Fitzhugh, future COE efforts should take the potential for residual deposits into cognizance. R.M. Hornor (3PH38). This site is described as a 200 to 300 m long daub and sherd scatter along a ridge over Old Town Creek. It had recently been the site of a series of field hand houses when it was reported by Leuken in 1968, the second year it was in cultivation. At that time, some burned daub, pottery, stone, and small celts were exposed on the surface (68-275, 90-512). The Hornor site, apparently contemporary with and slightly later than Ellis (based on the ceramics), is very interesting because it has produced the largest available collection of Late Mississippian lithics in the area. A number of pots have probably been dug here, too (John House, personal communication 1997). Hornor produced many Nodena arrow points and preforms; polished pebble celts and fragments, but also larger igneous-metamorphic celts or axes and celts made of petrified wood; and daub fragments (Figure 25). The Late Mississipp/Protohistoric period materials include large bipointed blades and an example of the large triangular blade associated with the "Oliver Lithic Complex". Leuken made a large collection of ceramics as well: 77 Mississippi Plain, 22 Bell Plain, 6 Barton Incised, 3 Parkin Punctated, 10 Old Town Red, 2 Nodena/Avenue, and 1 Wallace Incised. This is one of the highest percentages of Bell Plain in the county. Four red rim appear to be from small flaring bowls (Figure 25). One Mississippi Plain rim has wide, deep nicks and another indicates a wide-mouthed bottle. The Old Town Red sherds include a large, battered rim adorno, probably with the conical, dented-edged head of the "corn god" (see Figure 21 h). There is also a much smaller, very simple animal rim effigy. These simple rim-rider effigies appear to be quite late, based on their presence at Oliver and Menard and the Wallace Incised also indicates late occupation, as it is often associated with the Oliver lithic complex, although it is rare at the Oliver site itself. R.M. Horner is evidently an important village site for studying the latest Mississippian occupation of the area north of Old Town Lake. Sears (3PH119). This site, a half mile south of Helena Crossing, lay on a slight rise along a dry slough that forms the uppermost extension of Old Town Lake. The site was 75 m long, but cut by a highway and partially in pecans when reported by Leuken in 1972. He made at least three collections from the site and notes several other individuals from around Helena who had collected points, celts, and sherds there. The Leuken collection includes large, flat based Mulberry Creek jars, scalloped rim Baytown Plain bowls, and Larto Red thickened rim bowls, but one shell tempered sherd, daub, and a petrified wood celt were also recovered (72-137, 90-571). There is also a large polished celt intermediate in shape between the classic Baytown adz of Coahoma County and the long, narrow standard Mississippian pebble celt or chisel. This unusual artifact indicates that Sears could have a ca. 1000 component, making it a very interesting location for the study of emergent Mississippian society. 3PH160. This small (70 by 35 m) mostly Archaic and/or Woodland site lies on a ridge of Sharkey silty clay along a bayou. The surrounding area had been ditched. It has two eroded sherds that appear to have had sparse coarse shell temper (77-979, 77-980). The materials are from the Leuken collection. HPA 13, 14, and 15 (3PH255, 256, and 257). These three small occurrences of shell tempered pottery were recorded by HPA during the course of the Helena Harbor planning survey. All have grog tempered ceramics as well. HPA 13, a low circular knoll, produced four pieces of Mississippi Plain and two of Parkin Punctated, grog tempered pottery, a core, a cobble, and some daub. HPA 14, on a low ridge on the land side of the levee, produced 16 sherds of Mississippi Plain and 14 unclassified decorated sherds. HPA 15 produced a single shell tempered sherd. The AAS revisited the area of these finds, recommending mitigation or avoidance for 3PH256 and monitoring for 3PH255, which were considered potentially one site based on their proximity to each other. The AAS recorded both areas as 400x300 m and concurred with HPA's site boundaries for 3PH257. 3PH256, which was tested by HPA, had artifacts to 70 cmbs. 3PH257, tested by HPA with 119 shovel tests, 5 posthole digger tests, and 2 1x1 m squares, also had artifacts (shell, sherds, and bone) up to 70 cm deep. Old Town (3PH20/15-N-3). The remains of a large Indian town "on Old-town Bayou, 8 miles below Helena" on a horse-shoe lake, consisting of "mounds, embankments, and bricks of antique appearance and form", supposed to be the site of Capaha were noted early in the nineteenth century (Monette 1846:50). Local legend has associated the Old Town and Helena localities with the events recorded in the DeSoto entrada accounts ever since this time, although modern interpretations often place the crossing to the north. Before 1868, Jerome Pillow, owner of a plantation on Long Lake Bayou, but then resident in Maury County, Tennessee, had given a third of his pottery collection to J.H. Devareax of Cleveland. In 1872, Devareax (1873) deposited his collection at the Smithsonian Institution and included a description of the conditions the materials were obtained under. In constructing a levee across Old Town and Long Lake bayous, hundreds of skeletons, in "a state of perfect preservation" and in "sitting posture" were exhumed from depths of three to ten foot, being covered with three to five foot diameter trees (Devareax 1973:435). The sketch map of the area of the finds shows the Pillow Plantation and, southwest of a mistakenly-formed Old Town Lake, an apparent large earthwork to the river side of the levee. Pillow recounted that the laborers [probably slaves] did not want to disturb the bones and "were persistent in destroying the pottery to prevent its removal. Their aim was to conceal in the dump or embankment both bones and pottery" (Devareax 1973:436). One he did obtain was a half-gallon vessel, shaped like an animal. The remainder of the collection apparently stayed in Tennessee. Thomas (1894:234-237) reports on the Old Town Works in some detail (Figure 26). The larger part of the site as it existed at that time (recorded as Buie by the LMS) had a fortification line, then incorporated as part of the levee system, surrounding a large pyramidal mound (Mound 3) and smaller circular mounds and "numerous saucer-shaped house sites [containing] the usual fire-bed, charcoal and fragments of pottery". Mound 3 measured 96 by 86', with a 4' platform supporting an 8' tall second tier 50 by 60', where "nothing of interest was revealed". The northern part of the site had already been largely destroyed for levee fill (as reported by Pillow) but it included a circular enclosure and another length of earthen wall 4 to 5' high and twice as wide, extending from the old river bank 400' to the head of the lake outlet, ending at a 15 by 30' rectangular enclosure near the outlet of Old Town Lake. Thomas reports a local tradition that these last small works were built by the remnant of the DeSoto entrada as they built their barges to go to Mexico, an attribution that would make the Old Town area that of the province of Animoya rather than Pacaha, as Monette (1846) believed. Mound 2, which had already been mostly removed for fill, seemed to have been 600' long, 200' wide, and about 8' high, containing from one to three layers of burials from which "several hundred" pots had been removed. Excavation in the mound showed it to be made of the surrounding soil, "with the usual admixture of fire-beds, charcoal and ashes." It was from the full-length, cypress coffin graves in this mound that the hooded human effigy bottle illustrated by Thomas (1894) was recovered. Old Town had already been largely destroyed by borrowing, pothunting, and alluviation by the time of the BAE visit and it has only deteriorated in the subsequent century. The main, southern enclosure contained a platform mound (Mound 3) with construction stages and a sequence of burned structures and smaller mounds typical of walled Mississippian towns. The Smithsonian excavator noted "all appearances indicate that this immense mound [Mound 2] was originally for a long time, not a cemetery but a huge platform of dwelling sites, but for some unknown cause, at a comparatively recent period converted into a broad burial mound" where people were buried full length on their backs (Thomas 1894). This sounds very much like what occurred at the contact period Oliver site, where the mound-top interments are attributed to the initial periods of epidemics believed to have depopulated the region between the time of the De Soto entrada and the establishment of regular contact by the French. Smithsonian Institution records further detail the locality as having four mounds "built onto old plantation owner's levee near Oldtown Ldg. 3 on one side Oldtown Bay 1 on the other" (Hoffman 1975). The Smithsonian houses a collection from the site and notes that relics found by Gen. Pillow in 1855 are now in Nashville, Tennessee. At the time of the Smithsonian report (ca. 1880), all that was left of the village of Old Town was a saloon and dwelling on Mound 1 and a house in the small enclosure. Material listed was "Plain ware thick Gray brown paste some shell many soot blackened some red decoration-incised short grooves; straight lines diagonal curves-large punctation one elaborate pattern White flint & crystal points" (Hoffman 1975). However, Hoffman (1975) has reported that the Smithsonian Institution National Museum of Natural History Arkansas Collection from Phillips County is confused but that they have a collection of Mississippi Plain, Barton Incised, Parkin Punctated, and Old Town Red donated by the Phillips County Museum in 1934, from a site 14 miles south of Helena Crossing, probably referring to this same site. Dale Kirkman (1964:1), a Phillips County historian interested in the DeSoto route, recorded much of the same information about the Indian remains at Old Town landing, 17 miles below Helena. She states that many of the mounds there were used for levee fill to close off Old Town Bayou before the Civil War, that the Nashville War Memorial Building collection was sent by Gen. Gideon Pillow and others, and that at least part of the Judge M.L. Stephenson collection was obtained from the site, with the help of "an old Indian guide". Griffin (1969:8) gives the date of Pillow's discoveries as prior to 1878, during levee construction, and adds that besides the Nashville, Smithsonian, and Stephenson collections from Old Town, pots from the site can be found in museum across the country. Citing "Monette's History" and "Mrs. H.C. Rightor" Kirkman noted that the remains of this large town were found in 1846 but as late as 1898 parts of its walls were still standing. It was further disturbed in 1933, when levee contractors dug into a burial ground and ancient town, recovering 70 or 80 skulls from under an old levee. Phillips et al. (1951:332) describe Old Town as a St. Francis type town of short duration, with a plaza of unknown length oriented southeast from a 12 foot tall square or stepped mound towards another mound about 4 foot high, with abundant daub and other refuse surrounded by a wall. This description is apparently based on Thomas (1894) rather than anything visible in 1940. The LMS surveyors believed the site to be almost completely destroyed by borrowing, with the remnant being covered by alluvium. In some places, these recent deposits were 6 foot thick. The borrow operations had left a few areas standing where the two to three foot thick village deposits were not completely removed. They suspected that their informant, Orville Buie, the light tender at Old Town, was potting the remnant. Their surface collection, mostly from outwash around the bases of the borrow pits, consisted of 437 Neeleys Ferry Plain, 36 Parkin Punctated, 14 Barton Incised, 12 Old Town Red, 3 Bell Plain, 2 Carson Red on Buff, 2 unclassified shell tempered incised and/or punctated, 1 Tyronza Punctated, and 1 Nodena Red and White. The AAS site file extension shows that in 1975 John Connaway of the Mississippi Archaeological Survey wrote to the AAS informing them of J.O. Wheeler's looting activities at Old Town, noting that COE ditch clearing had revealed midden and that about 50 pots had been removed. Burney McClurkan of the AAS Pine Bluff station contacted the COE and arranged to have some 7 to 10' of fill deposited around the looted areas in a futile attempt to halt the depredations. This area, a borrow pit between the mainline levee and the older spur still had areas of occupational debris eroding around its edges a few years later (Dorwin and Stewart 1977:19). When the site was next visited in 1977, a badly potted area in an overgrown field was noted among the irregular remnants left by COE construction (Dorwin and Stewart 1977). Material available at the time of their study, like Mississippian Plain, Barton Incised, Ranch Incised, Parkin Punctated, and Old Town Red were considered indicative of Late Mississippian and Protohistoric occupation (74-144; 77-969,975,976). The area was recorded as Newellton and Robinsonville soils, with the original topography indeterminate. A select collection was made, mostly from one potted burial. The pothunting appeared to have occurred within the past year. Probing continued and Dorwin and Stewart (1977:19) noted that "the field resembles, in some ways, an old battlefield, complete with cratered holes, broken relics of human manufacture and even bone fragments. At least one person has brought in a backhoe to excavate for burials." Brown (1978:16-17) recorded four pot-hunted vessels alleged to be from Old Town in the J.O. Wheeler collection: an Avenue Polychrome bottle, two Avenue Polychrome teapots, and a Rhodes Incised/Carson Red on Buff "corn god" effigy (Figure 27). The Avenue bottle is unusual in having a wide mouth with rounded lip and short, recurved neck. It has four sets of interlocking scrolls on a flattened bulbous body with flattened base. One large teapot with interlocking scrolls and spiral-wrapped spout has the neck broken off. The other has horizontal bands, knob opposite the spout, and short spool neck. It is very similar to the Wheeler collection teapot from McCluskey. The very large effigy bowl combines three symbolic motifs: swastika spirals, festoons, and the mythological being with the conical, dented-edge head. Like most effigy bowls, it has an ovate orifice with the head and arc-shaped lug (with incising) on the long sides. The Rhodes incising (swastika) is on the exterior and Carson slipping (festoons) on the interior. Hathcock (1988:239, 277) illustrates two red-and-white standing animal or mythological creature effigy teapots attributed to the Old Town Site. One is of the type commonly referred to as "dogs", but the other has a quite different head. The UAPB Old Town collection (74-144, 145) has 234 Mississippi Plain, 3 Bell Plain, 7 Barton Incised (parallel and crosshatch lines), 6 Parkin Punctated (at least two on the upper part of the vessel), and 5 Nodena/Avenue slipped (Figures 28 and 29). Many of the Mississippi Plain sherds come from a single thin vessel; thin paste (var. Coker) is thought of as an Early Mississippian diagnostic. The 10 Mississippi Plain rims indicate deep, nicked rim bowls; a flaring rim bowl with sharp inflection and wide spaced, shallow cut-outs; jars; and indeterminate bowl/jars similar to helmet bowls. There is some bone in the collection, including a few fragments that may be human. Pieces of daub showing split cane lathe and grass temper were collected, along with a few Woodland sherds and lithics. As at Carson-Stovall-Montgomery across the river, the Old Town Works included multiple areas of earthwork construction, possibly indicative of temporally or ethnically distinct occupations. The Old Town site appears to have been occupied over a long part of the Mississippian period, especially in the 1600s roughly contemporary with the late occupation of the Parchman phase sites, as Beaudoin (1952) suggested. Buie (3PH21/15-N-4). This LMS site corresponds to the southern town of the Old Town Works of Thomas (1894). Dorwin and Stewart (1977) point out that while Cyrus Thomas appeared to have called the whole area "Old Town Works", the LMS had called only the east part "Old Town" and the western part, across the levee "Buie". The LMS reported Buie as a Baytown and Mississippi period village site (Phillips et al. 1951:51). Materials were exposed where a drainage ditch cut through an old plantation levee. A feature exposed in the ditchbank was also reported by the LMS. Their site form describes this feature as two superimposed fire basins about 20" deep and of the same diameter, along the west edge of a pit 6' wide and 4' deep which had a hard floor with some evidence of burning. The two sides of the pit did not seem to meet at the base and the pit appeared to have been filled with river-deposited material. This unusual feature is hard to interpret. The LMS collection consisted of 168 Neeleys Ferry Plain, 10 Parkin Punctated, 6 Barton Incised, 3 Old Town Red, 2 Carson Red on Buff, and 2 Nodena Red and White. Brown (1978:17-21) recorded 13 pot-hunted vessels believed to be from the Buie site. It was the largest collection Wheeler displayed from any of the Phillips County sites and they appear to be those collected at the time Connaway reported looting after CoE construction. The Buie vessels were seven bottles, including a square bottle and a tripartite bottle; a "dog" effigy teapot, a double frog effigy, two bowls, and two fine-paste jars (see Figure 27). Four vessels were said to come from a single lot: a carinated, flaring rim bottle with vertical alternating red and white panels; the small compound Mississippi Plain frog effigy; and the Old Town Red square and tripod bottles, which have flared, beveled lips. There are three bulbous body bottles with long, flaring necks: one Mississippi Plain, one Old Town Red, and one Avenue Polychrome (heads with forked eyes and open mouths separated by maces). The four-legged, spout-tailed Avenue Polychrome creature (of the form commonly called dogs) has spirals on the body and spout, and a carved face with forked eye and carved teeth very like the faces of panther head-snake tail bowls. A Mississippi Plain bowl with notched rim and a Bell Plain flat-based, high-shouldered jar with short, straight neck and round lip were found together in another grave. Finally, a bottle or jar with short, wide neck had incised imbrications on the exterior (cf. Walls Engraved or Carter Engraved). The Hatchcock (1988:76, 106) compendium of pothunted "Quapaw" pottery illustrates two compound bottles attributed to the site. One appears to be the same red tripartite bottle illustrated by Brown while the other is a coarse shell tempered double bottle with rows of nodes like those seen on squash or conch effigies. A very similar red tripartite bottle was recovered from the Oliver site. Overall, it appears that the Buie portion of the site dates quite late in the Mississippian sequence, i.e. Protohistoric, like the northern town, although there are some vessels indicative of fifteenth century occupation and the incidence of the Memphis rim mode may indicate sixteenth century occupation as well. 3PH153. This 40 m diameter outlier of the Old Town site was reported by Leuken to Judith Stewart during a cultural resources survey (Dorwin and Stewart 1977). The location was a low rise, possibly with a 10 m diameter mound, on Commerce silt loam under cotton cultivation. The site boundary was later enlarged upon Roy Cochran and Ross Dinwiddie's revisit (Klinger and Cochran 1983) to include three loci in a 50 by 80 m area around the Old Town berm. HPA screened a 40x30x50 cm shovel test, apparently negative, and they observed the extent of the site outside their project area without making collections. In the small collection from the proposed filling area, there were very few lithics, some daub, bone, and historic material. Locus 1, surface 10 m west of berm toe, produced a bottle neck fragment, two Mississippi Plain sherds, and two pieces of daub. Locus 2, 13 m to the south, produced a Mississippi Plain sherd. Locus 3, 15 m north of Locus 1, produced two Mississippi Plain and a grit tempered rim sherd. UAPB has 38 Mississippi Plain, 1 Old Town Red (medium paste), 1 Barton Incised (crosshatched, square lip), 1 Parkin Punctate, and 1 Owens Punctated (triangular fields of punctation on large jar rim, separated by pairs of lines, see Figure 21 i), and 1 Winterville Incised sherd (77-809, 970 or 82-1317). Two plain rim appear to be from short necked, wide mouthed bottles. Hugo (15-M-6) . This large village scatter along Old Town Lake was reported by Philip Phillips and Stephen William in 1953 after Wiley Wilcox, a collector from Memphis, had reported finding Poverty Point objects. The LMS surveyors found that recent bulldozing in the narrow space between the berm of an old levee and the lake had exposed a half-meter-thick stained level. The site had been visited by pothunters, judging from large sherds with fresh breaks associated with human bone. The scatter had extended along the lakebank for a ¼ mile or more, with all Mississippian material on the south end and some admixture of Woodland to the north. William made a collection, but the loacation of this site is uncertain. Smithsonian #11. Hoffman's (1975) annotation of the Smithsonian Institution collections and catalog describes the provenience of the Phillips County #11 collection as mounds on the Key Ward place built into the levee near Ward Bayou with refuse pits filled with bone, sherds and charcoal exposed in a road cut. Hoffman (1975) presents a catalogue of these materials from the west side of Old Town Lake as "Flint-yellow brown, mottled gray…Blue gray ware-fine shell temp-buff slip thin rim-slight notch Buff gray ware-fine shell temp. medium-thick large pot rim Chunks of burned clay Sun dried chunk like effigy?" The location is uncertain. Buford Culp/Shirttail Ridge (3PH40). This ten acre scatter south of Old Town Lake lay on a long narrow ridge. No mound was in evidence when Leuken reported the site as new ground in 1968. Lueken collected 12 pieces of Mississippi Plain, representative of jars and bowls, and daub fragments with cane impressions (68-281, 90-514). There were also metamorphic celt fragments, Archaic points, and a discoidal with pits on both sides (see Figure 23 a). The site appeared to be mostly Woodland, but Leuken indicated that J.O. and Buddy Wheeler were collecting the site. The collection is too small to say anything about the date of the site. Eloise Toney (3PH42). Around 1960, a mound was leveled on this sloping land near Old Town Lake. Abundant occupation debris reported by Leuken as still covering several acres of this "stained" sandy loam field in 1969. He noted that the mound had been pushed downslope towards the lake. Materials included stone tools, grog and shell tempered pottery, and a pebble celt. Marksville Incised, Larto Red, Indian Bay Stamped, Gary points, adze/Baytown choppers, black chert and novaculite, and a possible earspool indicate Middle Woodland occupation. Shell tempered material was reported, but is not present in the present collection (69-191). Warren Spivey (3PH44). This ridge on the inside of the north end of Old Town Lake produced one fine shell tempered sherd and several Coles Creek ceramic and lithic types (69-196, 90-518). Leuken reported the locality in 1969. Moore Place (3PH46). Leuken reported this site at Little Zion Church in 1969, stating he found flakes, biface fragments, clay? tempered sherds, and a "laurel leaf shaped blade", which is not in the UAPB collection (69-197). The site lies at the north end of Old Town Lake. The possible Nodena point is the basis of assigning this site a Mississippian component. A.M. (Mac) Toney (3PH48). This several acre site on the bank of Old Town Lake lies atop a slight bluff and along a bayou draining the lake. When Leuken reported the site in 1969, there was a small mound in timber on the site, which had been dug several times, but that most of the site was in cultivation. He found "chunks of clay resembling daub…partly polished celts, bird points." A five foot high mound 15 by 30 foot on top, lying on sandy soil and producing the base of a Baytown vessel, Marksville crosshatched rim with hemiconical punctations, arrowheads, and daub, as reported by a student, Kylene Reece, in 1973, is assigned the same site number. Leuken collected 50 Mississippi Plain, 1 Bell Plain, 1 Barton Incised, 1 Parkin Punctated, 2 Old Town Red, and 1 Wallace Incised, as well as daub fragments showing cane impressions (69-201, 76-557, 90-522) from his Mac Toney site. The Mississippi Plain rims are mostly thin and from slanting to curving walled jars, with some larger, square lipped jars. The Barton sherd comes from a small globular jar with line filled triangles around the orifice restriction. The Parkin sherd shows unaligned nail marks on a similar small, thin jar. The red sherds are an eroded fine paste body sherd and a coarse shell tempered hemispherical bowl rim with nicking. The Wallace body sherd has wide spaced, deep curving lines. Leuken's collection also included crosshatched rims, dentate rocker stamped sherds, flat bases, quartz, small corner notched (Alba cluster) arrow points, and a small rectangular "greenstone" celt, all indicative of strong Marksville and Coles Creek period components. Wabash. Wabash was recorded by Brown (1978:2) as "just north of the central part of Old Town Lake…east of Wabash" on the Toney land. It seem likely to be the same site as 3PH48. The four vessels Brown (1978:25-26) recorded were all bottles (Figure 30). There are two Avenue Polychrome bottles, one with an elongated body and long converging neck, with vertical panels, and one with a round body and converging neck, with alternating waves around the body and striped neck. The latter has negative painted cross-in-circle motifs in the white spaces filling the troughs of the waves. The wave motif is like that sometimes seen on Owens Punctated. The long converging necks, vertical panels, and negative painting all seem indicative of occupation in the fifteenth century. Both plain bottles, one coarse paste and one fine, have round bodies and short, wide necks. One of the Avenue bottles is also pictured in Hathcock (1988:214). Wild Duck/Hill Fontaine (3PH87). This site lay on a sandy ridge around a marshy lake when Leuken reported it in 1970. There were two "mounds" and "firepits" on a bank along a bayou, one having an abandoned historic cemetery on it. The UAPB collection has two leached shell tempered sherds from vessel (jar?) curvatures (70-19). Anvil/Jessee (3PH88). This abandoned historic cemetery had two apparent mounds. It lay on the bank of a dry bayou when recorded by Leuken in 1970. It produced two shell tempered sherds, with the rest of the material being Woodland (70-17, 90-557). This is definitely a location worth revisiting. Pete or T.E. Wooten Farm (3PH89). This predominantly Woodland site, three miles south of Helena Crossing, produced two typical chipped and polished pebble celt polls (see Figure 23 j,k) and some amorphous and polyhedral cores probably attributable to the Mississippi period (70-20, 72-129). The setting was a rise over a small lake in a cotton field. Two Mississippi Plain (slightly restricted and flaring rim bowls), 3 Bell Plain, 3 Old Town Red (fine paste or var. Beaverdam), and 1 Walls Engraved var. Hull (a fine paste flaring rim bowl with interior imbrications) were also collected by Leuken in 1970. A large proportion of Bell Plain and Walls Engraved var. Hull were taken as markers for the Parchman phase by Phillips (1970). Occupation in the 1400s or 1500s is expected. Old Key Plantation (3PH133). This site lay on a ridge or slight bluff over a dry slough when it was reported by Leuken in 1972. The site was in cultivation and he collected 2 shell tempered sherds, clay tempered sherds, lithics, and reported that Allan Keesee of Helena had a "highly polished stone figure of a man" from the site. There is also a basal fragment of a thin, probably very old point that has been reused as a pebble celt, resulting in a sharp chisel edge (72-116, 90-579). This is probably a hamlet or farmstead. The Southern Sites There is a cluster of five or six sites on the high, loamy land along Bee Bayou, a crevasse between the Mississippi and the lowermost course of the White River, and a final, largely destroyed site near the Desha County line. The area includes two of the only excavated Quapaw phase sites, Avenue (Moore 1911) and Dupree (McGimey 1965). The reported occurrences are 3PH1, 3/65/186, 64, 113, and 126. The first of these, Avenue, seems to have been assigned multiple numbers. In a history of the Cary Brothers Construction Company, which built levees in the mules-and-slips days, we find that around 1890 or 1900, an Indian burial ground was used for fill in the Ferguson's Landing locality, in extreme southern Phillips County. Cary (1980:1) notes that many jugs were found and that his father, the contractor, kept a couple of water jugs, 6 to 8 inches tall and about 5 inches in diameter. These sound like the typically small bottles that often accompany Mississippian burials. The actual construction date may have been 1912-13, as this is the date of a crevasse at Ferguson's Landing as shown on the 1954 Mellwood 15" quadrangle. Alternately, the redeposited site materials may have been destroyed by this crevasse. The University of Arkansas, Fayetteville, museum collection has a Ramey Incised jar or bottle with high, sharp shoulders, standing rim, and flat base. The lip is missing. The incising, on the upper body, above the shoulder is done with broad, shallow, u-shaped lines. The motif has multiple sets of curving and rectilinear lines. The provenience is given as the Cash site, "about 2 miles from Dupree". It appears to have a fine grog tempered paste, but it has been coated with glue and/or shellac. It was donated in 1972 by H. R. McPherson. This vessel indicates Cahokia-horizon occupation ca. 1100-1200. There are several other indications of Early Mississippi period occupation in this locality. Dupree (3PH1/16-L-6). Phillips et al. (1951:52) describe Dupree as a large village site with large rectangular platform mound and 40 small mounds in a loose, oval plaza arrangement (Figure 31). The site was considered to be a St. Francis type town of short duration, having a 600 by 250 foot long plaza oriented to the east of Mound A, a 10 foot tall, 100 foot square mound with a 20 foot diameter, 3 foot high, conical or house mound atop it. A pothole in the house mound on the Mound A summit showed a thick daub lens. Daub and refuse were abundant around the numerous house mounds (Phillips et al. 1951:334), which typically measured about 30 foot in diameter and 3 foot high, with dumps of black midden soil, ash, sherds, and bone near the presumed doors on the plaza side of the house mounds. One, Mound D, at the south end of the site, was 100 foot in diameter and 6 foot high. The LMS site form adds some additional information: the site had not been in cultivation long in 1940 and borrow pits were still observable behind the house mounds. Their sketch map shows the sides of Mound A to be oriented with the cardinal directions. The LMS collection included 708 Neeleys Ferry Plain, 53 Wallace Incised, 43 Old Town Red, 6 Barton Incised, 4 unclassified shell tempered incised and/or punctated, 3 Parkin Punctated, 2 Rhodes Incised, 2 Nodena Red and White, 1 Kent Incised, and 1 Oliver Incised. This 13 acre village site on an island between two forks of Bee Bayou was land leveled in 1964. In the winter of 1963-1964, the University of Arkansas Museum conducted salvage excavations at Dupree, making it one of the only professionally excavated sites in Phillips County (Moselage 1965, McGimey 1965). The main mound had been pushed down around 1953, with bones and whole pots being found. The site still had a loose ring of small house mounds with associated middens and artifacts indicative, apparently, of Early Mississippi through Protohistoric occupation. The fact that the AAS site file extension has House's 1983 report of pots being dug there may indicate that portions of the site still remain, although the site was reportedly completely leveled. Moselage (1965) reported on two days of testing in gumbo soil that uncovered a house in one of the small mounds opposite where Mound A had been. A 15 by 18' staggered or double row individually set post house with a central fired area of prepared clay floor is reminiscent of the structure Ford (1961) reported from Menard. The structure sat on layered midden soil and there was a small trash pit near the north wall; another quite large one, possibly a borrow, on the north side; an elbow pipe in the east wall line, and a folded (partly decomposed or bundled) burial southeast of the building. Structural debris included daub with split cane and grass impressions, charred cane on the floor, and the ubiquitous dirt dauber nests. Artifacts from the floor were 24 Mississippi Plain, 2 Bell Plain, 1 Caddoan? Plain, 1 Wallace Incised, and 2 Barton Incised sherds as well as fragments of a sandstone mortar and a sandstone whetstone and a few other stones. The UAPB collection (63-47, 64-53, 69-200, 73-90) has a very small amount of Late Woodland cordmarked, fabric marked, check stamped and brushed pottery, but the main component of the site appears to be Late Mississippian/Protohistoric. It also has a ca. 1200 "Cahokia horizon" component, as evidenced by a Burlington chert microdrill (Figure 32, i) and a L'eau Noir Incised sherd (Figure 33, l; also illustrated in House 1983). This Plaquemine ceramic type has its center of distribution in the northeast Louisiana Bayou Bartholomew-Macon Bayou-Tensas River area, but was recovered from an early 1200s feature at the Boydell site (John House, personal communication 1995). The paste of the L'Eau Noir sherd bears obvious parallels to later Caddoan pottery. House (1989:4) describes it as having angular grog and shell up to .5 mm in diameter with dry incising . The site apparently lacks European materials and Caddoan sherds are very rare. C.R. McGimey thought in 1965 that Caddoan sherds indicated very late (ca. 1700-1800) Quapaw occupation. However, in his tabulation of several thousand sherds from the site the amount of a Caddo-affiliated material is quite low: 2 Means Engraved and, possibly, 3 "brushed", which could as well be Plaqumine culture-derived Grace Brushed, a type more strongly associated with the Tunican peoples of southeastern Arkansas and northeastern Louisiana.. He also stated that the connection with the Little Prairie (Menard complex) sites was geopolitical: "I take the moat and the 'hidden' location of the site some miles back from the main river to be indicative of defensive maneuvering. I also take the presence of the backwater [Bee Bayou] passage to Menard Mounds to be along the same lines" (McGimey 1965:5). However, he later restated this somewhat, saying Dupree was one of the largest Quapaw sites, either "earlier in time than Menard-Hodges" or "at least partly contemporaneous with Menard-Hodges and other sites around Arkansas Post" (McGimey 1989:41-42). Hathcock (1988:30) has illustrated a late-looking pot-hunted vessel attributed to the Dupree site. It is a small jar with a complex curvillinear and punctated pattern and sharply inflected short neck. A large (N=1765) collection from the 1963-1964 investigations is stored at UAPB (see Figures 33 and 34). It includes 1550 Mississippi Plain, 12 Bell Plain, 18 Barton Incised, 10 Parkin Punctated, 107 Old Town Red, 10 Avenue/Nodena, 49 Wallace Incised, 1 L'eau Noir Incised, and 8 other small decorated sherds (Leland and Winterville incised and Owens Punctated). Variants of the helmet bowl as most widely interpreted (deep flaring rim bowls, indeterminate bowl-jars, shallow flaring rim bowls, and short-rimmed hemispherical bowls) make up a large part of the vessel assemblage. Mississippi paste occurs with such modifications as red slipping, rim effigies, lugs, rim nicking, and appliqué ropes (Haynes Bluff rim mode) on bowls and jars. Wallace Incised commonly co-occurs with large indented rim nicking and lower boundary lines. Combinations of punctation and incision (Owens Punctated) are more common than either Barton Incised or Parkin Punctated alone. Fine shell tempered paste sherds with red or red and white slip appear to come from bottles. Other ceramic artifacts include discs (counters?), shell effigies (spoons?), and pipe bowls. A large lithic collection was also obtained. Presumably this is the same site as that the "Dupee" collections at the Smithsonian Institution came from. The description of the mounds' location "in an open glade between two branches of Bee Bayou" matches that of 3PH1; and one of the mounds was reported as 15' high, dug into, perhaps by Dillinger, and by a later White camp (Hoffman 1975). All of the other mounds were under cultivation. The collection of sherds and daub is apparently small, but perhaps worthy of reexamination, if it still exists. Avenue (3PH3/16-M-1). The Avenue site on the R.B. Goalder and A. Martin plantations was dug by C.B. Moore (1911), and the results of this excavation have already been mentioned as part of the historic background of the formulation of the Old Town and Quapaw phases. Moore's site consisted of several loci and the exact location of his excavation is now uncertain; although pot-hunters have dug a site they believe to be Moore's site; the LMS has visited a village site with small mounds they also believed to be Moore's Avenue site; and ASS personnel have attempted but failed to find a site in the location given by Moore, finding a similar site (3PH65/186) nearby. Moore (1911) described the Avenue site as a mound on a high ridge with buildings on it. The mound had been spread by six years of cultivation. Adjacent ridges with occupation debris give the impression that the location was a sequence of point bars. Moore spent three days at the site with seven men digging the tenacious clay; fortunately, from his perspective, the graves were two foot deep or shallower. These were 39 bundle burials, 16 adult inhumations, 3 infant/child burials, an isolated skull, and 3 already disturbed graves. Other than pottery, the only artifacts recovered from the graves were a pebble chisel, long bone pins or awls, a perforated flat pebble, and ground red iron oxide. Seventy-five vessels were recovered and are described in Moore's (1911:401-405) report as thin, often yellow ware. Not all were described or pictured (Figure 35). Of the 17 painted vessels, 13 were red and 4 were red-and-white. They included an animal effigy teapot, with the animal, called a turtle by Moore, on its back with legs and tail-spout extended (Vessel 20). This apparent opossum effigy closely resembles one from Humber down to the details, but lacks the short rim of the Humber specimen. A red vessel with tail but head missing; a bottle with a constricted then flared opening and polychrome scroll (Vessel 32); a red bottle with similar neck (Vessel 68); a bottle with six vertical panels of red and white paint and a flared lip (Vessel 27); a red bowl with predatory bird head and tail (Vessel 43); and an unusual pedastled raccoon head bowl with modled snout, eyes and ears, red on the mask, back of head and interior and white below (Vessel 8). Other pots described but not illustrated include two bowls with "slight, rude, trailed decoration"; a small bowl with rough single festooned lines incised around the exterior with "rude imprints of a pointed tool"; a vessel with six small lobes; and a vessel with small pointed knobs near the rim. Moore's collection deserves further documentation. At some subsequent time, the Smithsonian Institution obtained collections attributed to the Avenue Mounds near Mellwood, Bee Bayou, and the Mississippi River which had been "disturbed by Moore" (Hoffman 1975). Hoffman's index describes a mound and elevated area near the settlement of Avenue, with the mound in full view of the road. The mound, occupied by several buildings, was much spread. The site was on the R.B. Goalder plantation, with a ridge on the same plantation about a quarter of a mile to the southwest measuring 120 by 240 foot and 2 to 6 foot high. A second ridge 125 by 95 foot long and 3 foot high lay to the west, with an almost contiguous ridge 85 by 70 foot long and 2 foot high. The description of three adjacent ridges matches Moore's (1911) description. The materials are a few types of sherds and light yellowish brown and dark blackish brown flint. This collection, if it is still intact, would be valuable in evaluating the Avenue occupation. In 1940, Phillips, Ford, and Griffin of the LMS reported 16-M-1 as two house mounds half a mile west of Avenue Landing. They note that there was Mississippian material on the mound dug by Moore, a mound with a house on it 100 foot west of the mound dug by Moore, and a small house mound south of this second mound. There was a vary large barn owned by Mr. Solomon on a mound which had been higher before being plowed. That the LMS did indeed visit Moore's Avenue site is supported by the fact that their informant had been present at Moore's dig. The LMS collection consisted of 411 Neeleys Ferry Plain, 8 unclassified shell tempered incised and/or punctated, 5 Barton Incised, 3 Parkin Punctated, 3 Old Town Red, and one sherd each of Ranch Incised, clay tempered Avenue Polychrome, Wallace Incised, Owens Punctated, and Leland Incised. In 1963, McGimey located a cultivated site, formerly having a historic house on it, which he believed corresponded to Moore's Avenue site. He collected only five sherds (63-49), now at UAF. McGimey (1989:42) believes that Avenue, which like Dupree apparently lacks European materials or any significant quantity of Caddo pottery, dated "as late as the latest occupation at Menard-Hodges", a seemingly contradictory belief, as menard has European trade goods. The UAPB collection (72-123, 73-73, 90-499) marked "3PH3" has in it 288 sherds Mississippi Plain, 2 Bell Plain, 2 Parkin Punctated, 7 Old Town Red, 3 Nodena/Avenue, 21 Wallace Incised, and 3 unclassified sherds (Figures 36 and 37). There are an additional 5 sherds marked as coming from Avenue in the UAF collections (63-49): a moderately sandy coarse shell tempered plain, a fine grog and shell plain, a coarse grog and shell plain, a medium-sized shell plain, and Mazique Incised (a grog-tempered Plaquemine type similar to Barton Incised). Mississippi Plain rim modifications identified in the collections include broad tooled nicks, nail punctations, beads, notches, lugs, and cut-outs. One of the two Bell Plain sherds is from a bottle body-neck junction. Broad line incised sherds are the main decorated type, with several motifs being represented: line filled triangles, alternating line filled and punctation filled triangles, and punctated and incised lower boundaries. The sherds are fairly thick, with square lips and sharp inflections. The unclassified sherds are 1) a body sherd showing broad, curving incising on a neck over narrower incising of descending triangles on the body; 2) an irregularly shaped (lobed?) body sherd with very narrow line filled triangles on a neck over tooled punctations on the body; and 3) combined curvilinear and rectilinear incised with hemiconical punctations on a moderately fine paste body sherd. The last two could perhaps be considered Owens Punctated. Other artifacts include sherd discs; a tan chert triangular arrow fragment; pebble celt fragments and a larger celt made on a cobble; a quartzite discoidal; a small ground and partially drilled quartz crystal; and daub fragments with cane impressions and exterior surfaces. The faunal remains may include some human bone. A more recent search for this site failed to locate one in Moore's given location, but a site that matches the description (3PH186) was found on high ground in an adjacent section. 3PH186. This site, possibly one of the uninvestigated ridges identified by Moore as part of the Avenue site, is at least very near the location he gives. The AAS site file extension shows that in 1978, when Leslie Abernathy and John House attempted to find Moore's Avenue site, they concluded that their 3PH186 corresponded to Bebee's 1969 report of 3PH65, the Chip Franklin site. In 1978, the site was still farmed by Chip Franklin of Mosby. In an area of minor relief, this site lies at the interface of recent sandy (Commerce-Robinsonville-Crevasse association) soils with siltier, lower levee and overbank (Newelton-Sharkey-Tunica) soils. It is described as a large, low density scatter of Late Mississippi period materials, a small village or hamlet with graves on the highest land in the section, as shown by 2 potholes beside the road crossing the site. Augering revealed a 35 cm thick, long farmed midden remnant. The 50 by 80 m area grab collection (78-570) has Woodland material (novaculite Gary point, Mulberry Creek Cordmarked) as well as Mississippian. It produced 48 sherds of Mississippi Plain, 2 Barton Incised, 2 Parkin Punctated (aligned sets perpendicular to a jar rim), 1 Old Town Red, and 1 L'eau Noir. The only vessel form identified are jars and deep hemispherical bowls. The primary archaeological component appears to be "post 1400" (House 1989:4) However, one sherd recovered from the site, the L'eau Noir Incised body sherd with fine grog and shell paste, indicates occupation earlier in the Mississippi period (House 1989). This sherd has dry incising with excised elements at the corners of interlocking frets. Some finer grog tempered material may be attributable to a Late Woodland occupation. Chip Franklin (3PH65). This site lay on a stream bank knoll, with material on the lower part of the knoll. Billy Bebee reported the site in 1969 as having beans so thick as to prevent much investigation. The AAS site file extension shows that in 1978, when Abernathy and House tried to find Moore's Avenue site, they concluded that their 3PH186 corresponded with Bebee's 3PH65, although the two sites are mapped some distance apart. Some of the ceramics collected by Bebee are clay tempered, so some of the lithics in the collection are probably Woodland (69-220). UAPB curates 101 Mississippi Plain, 1 Parkin Punctated, and 2 Wallace Incised sherds, and a fragment of a sherd disk, along with a small amount of grassy daub and bone. The Wallace Incised sherds are from a curving jar or helmet bowl and have large imbrications (cf. Ranch Incised/Winterville Incised var. Ranch). Eight Mississippi Plain rims indicate incurving jars and deep, thickened rim bowls. Bush (3PH64). A half mile east of Dupree, this scatter lay 8 foot above the bed of a dry part of the Bee Bayou network. The site was in cotton when it was reported in 1969. The reporter, Billy Bebee noted that this was not the only similar setting in the area, and that other rises along these creeks might be expected to have sites. The UAPB collection (69-219) has 17 Mississippi Plain and 1 Wallace Incised.. The Wallace Incised appears to be on a jar neck and one thin Mississippi Plain sherd represents a small incurving rim jar. The daub was grassy. Woodland ceramics were also recovered. Cottonwood (3PH113). In 1972, Leuken characterized this site as "Toltec-like" (i.e. probably related to the Plum Bayou culture), and returned to make at least three collections. This probable low, conical mound and 100 m long scatter of Baytown and Coles Creek period pottery on Newellton-Sharkey-Tunica association soil lies in the bend of a slough on the Quaternary terraces cut by Bee Bayou. A single shell tempered sherd and celt are the only indications of Mississippian occupation. This may be the same site as 3PH2, a small Woodland mound a quarter mile north of Avenue Landing also visited by the LMS. Percy Martin (3PH126). This location, only 3/10 of a mile across Bee Bayou from Avenue also had a mound. The scatter lay over several acres, with plentiful sherds and bone. The large surface collection from this site at UAPB (72-124, 73-75, 90-576) has 158 Mississippi Plain, 1 Bell Plain, 4 Barton Incised, 3 Parkin Punctated, 3 Old Town Red, 1 Wallace Incised, and 3 other decorated (1 with circular punctations, 1 small incised rim, 1 Mound Place Incised). The 23 Mississippi Plain rims include broad, shallow nicked rim bowls, deep nicked rim bowls, and curving shouldered jars (Figure 38). The collection includes Barton Incised-like (Mazique Incised) and Parkin Punctated-like (Evansville Punctated) grog tempered sherds, a mixed coarse shell and grog tempered sherd, and fine grog tempered plain sherds all presumably indicative of terminal Woodland-initial Mississippian culture occupation. The Percy Martin collection also has grassy daub, shell, and bone. There was very little stone, but a small polished celt was reported. There are also four pieces of white chert, possibly Burlington/Crescent, indicative of Early Mississippi period occupation. Mound Place Incised and a cut, partly drilled-out, conical antler tine arrow point are also indicative of Early-to-Middle Mississippi period occupation.

ASSESSING THE OLD TOWN PHASE

Here I will briefly summarize what has been learned about the Mississippian occupation of Phillips County thus far. To organize this discussion, the Old Town phase section (House 1982) of the Arkansas State Plan for the Conservation of Archeological Resources is reviewed, with comments on improvements in understanding since the State Plan was promulgated and specifics of research potential. Consideration of the ten research priorities outlined in the State Plan reveals the successes and shortcomings of the present article. Site destruction is a severe problem in Phillips County, and there are few sites originally included in the Old Town phase that have not been severely impacted. Therefore, even site remnants such as the Old Town Works/Buie site and Ellis mound should be preserved for research and as monuments to early settlers. Chronology. In evaluating the Old Town phase it must first be recognized that a single Mississippian phase is inadequate to describe the temporal variation indicated by the collections. The Old Town (ca. 1500), Quapaw I (ca. 1600), and Quapaw II (ca. 1700) sequence proposed by Brain (1988:269-277) still does not adequately take into account the full span of Mississippian occupation in the region. The Bee Bayou locality has indications of early Mississippian occupation, but no early Mississippian phase has been defined, and the period from 1250 to 1400 also seems to have a major lack of sites. The lack of middle period sites is probably more apparent than real, being due to a lack of diagnostics. The Ellis collections provide the only absolute chronology for the county, all other chronology thus far is derived from comparison with the surrounding region, not all of which is based on adequate dating. Ellis' seven radiocarbon determinations suggest that Mississippian occupation of the premound surface began in the fourteenth century and that mound building took place primarily during the fifteenth century, with mound-top buildings being burned and mounds buried on an interval of 50 years or less, with occupation extending in to the early 1500s. Therefore, if Ellis is taken as the typical Old Town site, it is a Late Mississippi (1400-1500) phase, roughly in agreement with the temporal definition proposed by Brain (1988). Earlier, (Brain et al. 1974:266-270) ca. 1540 diagnostic ceramic varieties and modes were proposed; and Brain stated that Old Town (his candidate for the principal village of Aquixo entered by De Soto in June of 1541 after crossing the Mississippi near Friars Point), Ellis, Fitzhugh, and Moore (the first town of Casqui, seen from high hills) all have diagnostics of this horizon. However, as we have seen, the collections available from the last three of these sites were quite minimal to reach such confident conclusions. Avenue and Dupree (the largest town in all of Florida, the main town of Quiguate, after Brain et al. 1974:278-279) are also seen to have De Soto horizon markers. Dupree, Brain says, had its main occupation, like Menard, as "a historic Quapaw component" (Brain 1974:278). On the other hand, McGimey (1989:36, 46) has proposed that Dupree is earlier than and Avenue later than the (still undated) main occupation of Menard-Hodges, but with both being pre-contact (McGimey 1989:41, 42). As should be apparent from the preceding discussion, there is a great shortage of absolute chronometric control. Indeed, we do not have a good artifact and stratigraphy based chronology for the area. I have made guesses as to site dates when I thought they were warranted. The Mississippian period lasted at least five centuries in the area (1100-1600). It might perhaps be better to say that late prehistory lasted nearly a millennium, from the first emergence of Mississippian traits in the Coles Creek period beginning around A.D. 750 to the final decimation of the Quapaw, Tunica and other Delta nations before A.D. 1750. From other areas, such as the Cahokia/American Bottom region, we know that it is possible to have control at the level of a century or less. Besides traditional radiocarbon dating, which requires fairly substantial pieces of charcoal, there are a number of other technologies that can be applied (accelerator mass spectrometry or AMS, thermoluminescence or TL, and oxidizable carbon ratio or OCR). Ultimately, we should seek to bring as many lines of evidence as possible to bear on the dating of sites. Geographical Extent. In Phillips County Mississippian sites have been reported from Crowleys Ridge and the Wisconsin age terrace remnants as well as from the modern alluvium of the Mississippi and White rivers. The Monroe County sites assigned to the Old Town phase in its initial formulation have not been considered here. The possible early components along Bee Bayou have been mentioned, but many of the sites on the old terrace around Marvell probably also date to the Early and Middle Mississippi periods. Late sites cluster tightly on the St. Francis River, Old Town Lake, and Bee Bayou. A few sites of the latest protohistoric periods, such as County Line, Horner, and Dupree are scattered through Phillips County, but as is to be expected, there appear to be significantly fewer of them than were occupied during the 1400s and 1500s. Evaluation of the Concept and Artifact Assemblage Definition. The third research priority outlined in the State Plan is evaluation of the Old Town phase concept and better definition of an Old Town phase artifact assemblage. After this review, the Old Town phase is still as shaky as it was thirty years ago when it was first proposed, but the evidence clearly shows that there is no lack of Mississippian occupation in the area. A major problem in evaluating exactly what we do know about Mississippian occupation in Phillips County stems from the diverse sources, extending back over more than a century, that we have to work with. The existing archaeological collections, like the site distribution results, are hard to evaluate as we know very little about the widely varied conditions and methods under which they were collected. Many are too small to be of any value other than to say that a Mississippian component is present. Phillips (1970) stated that the Old Town area sites were given a separate phase assignment based on the relative lack of Bell Plain compared to the Parchman phase sites in adjoining Coahoma County. However, we must remember that Phillips had very little access to temporal controls and that his Phillips and Monroe county sherd counts were quite small. I believe that the larger samples provided by the Ellis Mound excavations, the Wheeler vessel collection, and some of the larger Phillips County AAS collections show that finewares are not lacking in the area. Indeed, Bell Plain occurs regularly in the small collections from the west of Crowleys Ridge. At any rate, the relative presence of Bell/Addis pastes is of uncertain significance, as witnessed first by modern understanding of the development of ceramic complexes during the Mississippian period and secondly by the finding that the proportion of fine to coarse tempered ceramics can vary geographically among apparently contemporary sites (House 1991). In light of this additional research, do the Phillips County sites' ceramic elements still seem dissimilar to those of Coahoma County? I think not. A more important question seems to be: what does the wide area of great ceramic similarity in the region between the Parkin and Hushpuckena phases indicate? This is an area of about a dozen modern counties, that at the height of the Mississippi period may have had a population of tens of thousands. Overall, the ceramic content of the assemblages could use redefinition. The concept of sets as used in the later LMS work may be valuable. Also, the motifs on coarse shell tempered jars (Barton Incised, Parkin Punctated, Winterville Incised, Wallace Incised, and Owens Punctated) could use reformulation into a system along the lines of the "collegiate system" now used for Caddoan pottery. A tentative fifteenth century subphase, based on the Ellis excavations, has simple and gently curving rather than sharply defined shoulders or carinations. Jars are the most common vessel type, followed by bowls and finally, bottles. The noded strap handle may be a useful diagnostic. As in the original Parchman phase (Starr 1984), Barton Incised outnumbers Parkin Punctated by more than two-to-one. Slipped ceramics of several types are abundant. Winterville Incised may be an important minority type, as two vessel fragments of this type, both with guilloches, were recovered from mound occupation contexts. Bottles were painted red and white and were also often black stained. Red paint was also used on bowls. Flaring rim bowls, some with scalloped rim; deeper, nicked rim hemispherical bowls; and helmet bowls are typical of this and other local assemblages. The Memphis rim mode occurs in the Ellis collection, and based on Kent phase dating (House 1991), should date near the end of the construction of the Ellis mound. Settlement Pattern Variability and Site Location. When I discussed this topic in the contract report (Childress et al. 1995), I lacked the detailed distributional information for Phillips County just presented. I am now confident that there are some "lower order" sites, farmsteads or hamlets in the area Phillips parceled out to the "Old Town phase". Based on the limited data provided by late nineteenth-mid twentieth century reports of the sites along the main rivers, it appeared that Ellis was a single mound center (I now know it had two mounds), secondary in size to much larger, apparently contemporary, mound groups like the Old Town Works and Carson. Is such a hierarchical view still warranted? Consideration of settlement pattern must include a number of issues commonly found in cultural ecological studies of modern agriculturists, but which are hard to address archaeologically. How often did villages move? We know some towns lasted centuries. What prompted shifts in location? Did the exhaustion of local resources such as firewood and game, fluvial catastrophes, or weed or other pest infestation play a role, or were political reasons paramount? Under what conditions were very closely-spaced villages contemporary? What happened as abandoned village sites grew up in thickets? The Mississippian farmers seem to have had a predilection for Late Archaic-Woodland middens, perhaps because they would have been so fertile for corn, a nitrogen-dependent crop. Might not earlier Mississippian villages have come to be fields after some time of fallow? Were these locations remembered and revisited for generations, as we would say now, as the "old home place"? Did they continue to be cemeteries used by people practicing prolonged funeral rites such as bundle burial after they were no longer inhabited? Many of the other previously reported Leuken and harbor-vicinity sites probably have unrecognized Mississippian components. Much of the bank of Old Town Lake had Mississippian occupation. The evidence of ceramic typology indicates that much of it was roughly contemporaneous. The Old Town Works itself appears to have consisted of multiple adjacent townsites, as may have been the case at Carson-Stovall-Montgomery. One question that can only be answered subjectively is: how intensive were, and therefore how representative are the results of, the large surveys in the area of the Ellis site (Montsanto, industrial park, and Helena Harbor)? Without detailed examination of the project records, collections, and resurvey, evaluation of the projects enters the areas of veracity and competence. Perhaps shell tempered ceramics were misidentified or are not preserved in the plowzone on the many small Woodland sites recorded by these surveys. At present, accepting the results of these surveys at face value, it can be stated that isolated Mississippian houses are scarce in the area along the Mississippi. It would be a significant finding to say that the Old Town Lake area lacks farmsteads and that dispersed houses occur mainly on the terraces, but I can not say this with confidence given the present data. It seem that the area should at least have Middle period farmsteads. We know, based on project area size, even though older CRM reports do not typically give enough information to evaluate the level-of-effort, that the ditch surveys in the interior of the county were not adequate. At any rate, the work undertaken (Flenniken 1974) indicates that most sites lay above the project area elevations. USFS surveys have been largely limited to the upland Crowleys Ridge part of the county, and have generally recorded low frequencies of all prehistoric materials. This lack of prehistoric sites on highly erosive loess is also seen in the Bluff Hill counties of Mississippi. The confusion as to site location along Bee Bayou should be given priority for untangling. It is a relatively small area, so intensive survey could be made. We also need to predict where to look for additional Mississippian components in other areas. It is probably on the Pleistocene terraces that we need to look for additional small, presumably early, sites. In northeast Arkansas, Mississippian habitation of the higher, older, well-drained "flats", "ridges", "slashes", and "prairies" is earlier than the Ellis occupation, i.e. dating to ca. 1000-1300. Many of the small mound sites of the backwater parts of Phillips County are probably Middle Mississippian. In considering the distribution of Mississippian sites in the region, I would like to point out two areas that have been largely neglected: the Coldwater-Tallahatchie basin along the Loess Hills of Mississippi and the White River basin of east-central Arkansas. This is not because these areas lack sites. Brain (1988) attributes such Mississippi mound sites as Posey, Aldison, and Canon to a ca. 1600 Parchman phase, an attribution I contest. Based on my personal familiarity with these sites, I believe they date largely in the 1300s. I know of a number of other small mounds and hamlets among these three sites near my home in Quitman and Panola counties, Mississippi, in the Coldwater basin. A number of mound groups in the lower White River basin, such as Baytown, have Mississippian components, although the mounds themelves may be largely Middle and Late Woodland constructions. Consideration of these and other large backswamp tracts is critical to a real understanding of Mississippian settlement distribution in the region. As to the quality of documentation, the sites along the Mississippi have been highly impacted, and, at any rate, assessment of large sites is notoriously difficult, even without the extent of earthmoving associated with levee construction. Large scale site destruction is not limited to levee construction, however, and any future survey will have to take into account the extensive agricultural landforming accomplished since ca. 1970 with dirt buggies and land planes. Schultz (1968a) reported that by 1966, 15,281 acres had been leveled in Phillips County, some 15 percent of the land considered suited to this agricultural management practice. Modern agriculture has reduced or erased many of the smaller mounds and non-mound sites, but at the same time, clean tillage provides excellent survey conditions. The present move toward precision graded irrigation and reduced tillage/stale seedbed planting is daily reducing the archaeologists' chance of recording adequate information about Mississippian sites based on surface collections. Inquiry into the actual extent and location of land leveling must be undertaken to evaluate the representativeness of our sample of sites. This would also help identify areas that have not been as heavily modified by agriculture on which to concentrate future survey efforts. Community Patterning. The nature of the testing undertaken at Ellis and the degree of site destruction prevented the recovery of information concerning the size and arrangement of the site features. The sketch maps of other sites offer even less information. It can be stated, however, that the surviving mound at Ellis was about 2.5 m high and was built in at least four stages, but probably more stages, with buildings erected atop each stage. Wall trenches, postholes, compacted surfaces, sunken floors, split cane wattle, grass-tempered silty clay daub, and grass thatch were all observed in the mound remnant. I believe that substantial portions of burned structures remain at Ellis. Posthole testing and intensive survey indicated limited debris around the mound and a few scattered occurrences in the harbor and industrial park vicinity. The Ellis Mounds site apparently had a 300 foot long plaza. The smaller occurrences of shell tempered pottery in the initial Helena Harbor project and the sparseness of Mississippian material in the posthole testing indicated that the mound area may have had a small population, but this can by no means be taken as a certainty given the impact of levee construction. Because of the degree of site destruction, a full reconstruction of the sites along the levee will never be obtained, but it is likely that significant deposits remain at these sites. Some additional details on the layouts of the other Old Town phase sites can be gathered from the Smithsonian and LMS investigations, but this is at only the most general level. Size, density of occupation, and construction sequence can not be reconstructed for them, but they appear to have been fairly impressive mound groups with midden ridges, plazas between mounds, and enclosures. In contrast to the formal solar arrangement of the "St. Francis" type towns, Dupree in the south end of the county had a loose housemound arrangement around an oval plaza and at least one set-post house like the one from Menard, which closely matches Marquette's description of Quapaw cabins (see below). Use of bark slabs, especially as it is not stated if bark was the wall or roof material, is not incompatible with typical Mississippian daub, and the staggered rows of posts seem to be an ideal way of using such planks for walling, perhaps sheathing wattle and daub. The Dupree house (Moselage 1965) apparently lacked the beds at the ends seen in the one excavated by Ford (1961:154). These set post houses, in contrast to wall trench houses, appear to have a long tradition in the area. While pre-mound construction included wall trenches, the Stage III/IV structure at Ellis appears to have individual post holes, and they have been documented at Parchman, Hushpukena, and Winterville phase sites as well by around 1500. By historic times, they were also used by the Natchez (Brown 1990:230). It seem likely that wall trenches were no longer so practical when Mill Creek hoes, used as mattocks, were no longer available, so that people returned to digging postholes with dibble bars. Much has been made of the necessity of finding Iroquois-Algonkian style longhouses to identify ethnic Quapaw occupation. The presence of an alternative to the typical square wall trench house of the Mississippians in the Coahoma-Phillips-Arkansas counties area is compatible with glottochronology studies that indicate that the Quapaw language separated from its closest affines as long as 1000 years ago, allowing time for the Quapaw to be fully lower valley people without signs of significant incorporation of non-Siouan speakers (Baird 1980, Hoffman 1990:211). The associated features of individual households (refuse-filled storage/processing pits, fireplaces, borrow pits, graves, corncribs, and ramadas or brush arbors) should still be identifiable at some sites. There may also be distinct and identifiable ethnic patterns for the construction and distribution of these features. Site size, structure, depth, and feature types, artifact content, and chronometry are the critical areas to be addressed when evaluating the remaining sites according to State Plan criteria. In addition to traditional methods (surface survey, shovel testing, probing, auguring/coring, and test trenching), soil phosphate analysis, proton magnetometer, resistivity, ground-penetrating radar, and conductivity survey have some potential to document the internal structure of the remaining more-or-less intact sites. A final site feature that must be considered is fortification. Deep posts for palisades, restricted entries, and bastions along with moats or ditches should be easy enough to identify, but their presence, often assumed from site plan, ethnohistoric accounts, and ethnographic parallels, has seldom been verified in our area. The timing of wall construction around settlements is a valuable piece of information in considering the climate of hierarchical or consolidated political formations. Lithic Technology. In using shell tempered pottery as my primary definition of Mississippian, I have not concentrated on the lithic material to the extent that I would have liked to. Perhaps the most significant finding concerning the Ellis site lithic assemblage was the presence of a very small amount of exotic material (a Burlington/Crescent core trimming/platform rejuvenation flake and two utilized flakes that appear to be Dover chert, one having hoe polish). White Missouri Burlington/Crescent chert has been found at other Early Mississippian sites in the area (Barrett, Carson, Buford) but very little of this material was noted in the Phillips County review. Percy Martin (3PH126) on Bee Bayou produced four pieces of Burlington chert. The ceramics from the site seem fairly early, and the drilled antler tine is also probably a good Early-Middle Mississippian diagnostic. There are other indications of Carson and Cahokia-era occupation around Bee Bayou. Some of the rather dubious Mississippian components in the White River backswamp have discoidals, in addition to other large ground stone item. Discoidals may be a good early marker as well. Horner (3PH38) has produced numerous Mississippian lithics. There are lots of woodworking tools as well as arrows and knives. These artifacts have some resemblance to the Protohistoric Oliver lithic complex, but not all of this toolkit is represented. The collection from Horner might repay more careful study, as well as additional field visits. It may be that sites along the edge of Crowleys Ridge, such as Otey Miller (3PH51) and Melio Farm (3PH81) may have been the former locations of gravel exposures. These sites would be a good location to look for Mississippian lithic extraction and reduction sites. Quantification of the stone working and stone tool complex should emphasize, in particular 1) material sources and exchange of exotics; 2) stages of stone tool manufacture; 3) the significance of, and potential contemporaneity of, the Madison, Russell, Nodena, and other arrow points; 4) the Oliver lithic complex (Nodena points, endscrapers, large triangular knives, and pipe drills) alleged to be associated with Siouan speakers. It is hard to make much of an ethnic connection with the Protohistoric endcrapers, because they are also found among the remains of Chickasaw and Natchez towns at the same time level, and not just in the area extending from the Midwest down to the mouth of the Arkansas. Several very large Nodena-like bifaces occur in the Phillips County collections. It would be useful to have more precise context and function information on these bifaces. The Dalton project (Redfield nd) recovered numerous "arrows" and other Mississippian lithics. These collections, if still available, would be worth reviewing. Utilitarian Ceramics. The Early and Middle period base of the Old Town phase derives in part from Early Mississippian developments in the area between extreme northeastern Arkansas and Cahokia, Illinois. However, many of the surface finish types and vessel form of the initial Mississippian development did not last in the part of the Delta considered here. Cahokia Cordmarked, Ramey Incised, Varney Red, and Powell Plain made no lasting impression on local styles. Rather, traits having their origin in the local Coles Creek cultures last throughout the Mississippian development. The two most common jar decorations, Barton Incised and Parkin Punctated, have direct clay-tempered analogs in Mazique Incised and Evansville Punctated, as does the Mississippian jar form. Red filming which originates in the Tchula period and continues to be a dominant mode on shallow bowls in the Woodland and Mississippian cultures. The fifteenth century saw a wide range of ceramics being manufactured, of two different basic fabrics (coarse and fine). The fine wares are not all deposited as mortuary material. The characteristic vessels were curving-shouldered jars and flaring bowls embellished with rim nicking, red slipping, and incised interior lines. Jars sometimes had strap handles, often with sets of small nodes at the top and bottom. Barton Incised is the most common jar decoration followed by Parkin Punctated (especially towards the north) and Winterville Incised (in the south). The Winterville guilloche appears to be a particularly good marker for the late fifteenth century, as is the extreme Memphis rim mode, or flange on jar lips. Owens Punctated is not yet common, and other tool-decorated types like Leland Incised are quite rare. Fine pastes were present throughout the region, but in low or very low frequencies, as small bowls, modled effigies, and painted bottles with tall necks. An important functional category that originates late in the century is the small, generally undecorated, often heavily sooted, helmet-shaped cooking bowl/jar, which in late times was also deposited in graves. AM could be used to date the actual vessels, the kettles for cooking mush of the French accounts. If this is indeed the case chemical residue analysis should show the charred encrustation to be sugar and starch. Sixteenth century ceramics represent significant evolution. Vessel inflection points become more sharply marked and surface decoration becomes more complex, with combinations of incising and punctating and multiple decorated fields per vessel. Accompanying the Oliver vessel form is a squatter profile, particularly in jars, and increasing use of the helmet form. By 1700, white slips have disappeared, perhaps as the kaolin source areas were depopulated. McGimey (1989:46) believes that in the Quapaw phase, Parkin Punctated, Barton Incised, and Nodena Red and White gradually disappear, while the Carson Red on Buff helmet bowl, Wallace Incised, Old Town Red, teapots, coarse shell tempered pastes, and minor amounts of Caddoan ceramics increase. In saying that mortuary vessels are widely exchanged we open the discussion to the possibility of local specialty styles produced in part for export (without implying household-level craft specialization), such as red and white bottles or Bell effigy bottles. Presumably the source areas are where the form is most common, and the clay body of the far-flung examples should match the clays of the proposed locality of origin. That this is the case is supported by the fact that paste, color, decoration, and form are closely related. For example, Walls Engraved var. Hull tends to have a yellowish body with more fine grog than fine shell, and to occur on small, sharply inflected flaring rim bowl interiors. Actual vessels seem to have been moved around from the beginning of the Mississippian Period: Lower Valley Cahokia Cordmarked, Kimmwick Fabric Impressed, Varney Red, and Ramey Incised sherds tend to closely match the initially described northern types. Why would alien goods be preferred for grave goods? Are they indicative of immigrants or social ties to other areas? Did they serve as proof of one's being well-traveled or of foreign descent or were the valued for their superior artistic achievement and other symbolic content? If these mortuary item were surplus or specifically-made trade or gift item, what groups can be made of them? Was there a single distinctive type per social group which served as an identity marker? Source element studies and chemical analyses should also seek to find the origins of the white slip presumed to be kaolin and of the black varnish or lacquer believed to be of vegetable origin. Alternately, it has sometimes been suggested that this is a graphite based stain. As early as 1911, C.B. Moore (1911:485) stated that "we have been unable to obtain enough of this material to make an analysis" but also notes that it is not as heavily applied as the ferric oxide and white clay slips. The Jesuit relations may hold a clue as to the black stain of Avenue Polychrome. In 1687, Beschefer in Kebec wrote to Villermont, including a long list of American specimens being sent to Europe, such as a box of dishes and spoons, stone daggers, and seeds of three kinds of watermelons grown by the Illinois. Among them was a gourd of "copal balsam, which was brought to me from the Akansa country on the Mississippi, halfway to the sea" (Thwaites 1959c:289). This might be cedar or sweetgum resin or any number of other sap, but in all likelihood there was a considerable range of plant materials in use as glues and sealers. Subsistence. Most archaeologists would now add "Population and Health" to this category, which is addressed through biological remains (animal and human bone, pollen, charcoal, environmental reconstruction, etc.). Although pothunting of graves has been an extensive impact to the sites, I do not know of any curated skeletal populations from Phillips County (see also Hoffman 1990:216). Many of the sites have been noted to have bone and shell on the surface and a few small fragments of human bone are probably included in the surface collections. We should not expect collecting adequate floral and faunal samples to be a problem on the better-preserved of these sites. Ellis is the only site in the county that has provided any systematic samples of floral and faunal material. It is to be expected that the role of cultigens was highly significant by the Protohistoric period. It is not unexpected to find evidence of corn, along with a pattern of protein, wild plant food, and raw material procurement emphasizing the immediately local wetlands at the Ellis site. The faunal and floral materials were analyzed by Emmanuel Breitburg and Andrea Shea, respectively (Childress et al. 1995). There were plenty of fish species (especially gar, catfish, and drum), turtles, and some ducks and geese. Rabbit was the most abundant vertebrate in this likely Autiamque vicinity, where the Indians wintered the Spaniard on rabbits, like the Chickasaw had done. Also of course, there were deer, turkey, squirrel, raccoon, and opossum bones, along with beaver, woodchuck, and passenger pigeon.

Ash, locust, bodoc, and willow were the most common trees recorded in the float samples, along with riverine and second growth species like beech, redbud, sassafras, mulberry, and cherry. If this sounds like an upland assemblage consider that in a long-term village firewood may have been most easily obtained from the River where trees caved off Crowleys Ridge would have been floating down. Of food remains, hickory, walnut, pecan, hazel, and acorn were recovered; along with domestic squash (2 seeds and 1 rind), beans (2 cotyledons), and of course, abundant corn (present in 21 samples) and persimmon seeds (90% of the identifiable seeds). Shea gives the breakdown on the corn as 22% 8-row, 34% 10-row, 42% 12-row, and 2% 14-row (Childress et al. 1995 Appendix C:5). Other roughly contemporary Delta sites also seem to have grown mostly 10 and 12 row Midwestern flint corns. Being open or wind pollinated, corn is susceptible to producing some daughter seeds dissimilar to the planted mother generation if other types of corn are grown upwind. At any rate, the Mississippian people seem to have wanted staggered crops with different kinds of corn coming off through as much of the year as possible.

Overall, environmental reconstruction is a priority for increased understanding of Mississippian adaptations. Along the active rivers, as well as on the prairies and along large lakes, we need to determine what geomorphological and biotic changes took place in the general 750-1750 interval. The important generally available sources for this include the General Land Office (GLO) surveyors notes and plats (Figure 39) from the early nineteenth century, topographic maps, and the Department of Agriculture soil survey. The specific environmental conditions around each locatable site should be examined and compared. Most do not even have soil type specified on the AAS site form. This analysis should include elevation, relief, area of pedons of specific soil types, and wetlands extent. Consideration of the role of these variable can lead to the development of a model of other location likely to contain Mississippian sites. The GLO records, with township lines surveyed 1815-1816 and section lines surveyed in the 1830s and 1840s present an impressionistic image of pre-clearing Phillips County. A number of comments can be drawn from a cursory examination of these records. Parts of the Wisconsin terrace appear to have been quite dry, making settlement on water much more critical than it in the first bottom. The most thoroughly platted township indicates prairie up to the edge of the terrace. There were many small wetlands, called slashes, on the terrace, particularly along its edge. Owens (1860:415) describes the country west of Big Creek as level land, with low ridges of reddish sandy clay or gray sand loam soil, clay flats and wet prairies. There is a need for survey in the braided stream and backswamp areas. Small Mississippian sites can be hard to find. The time between spring plowing and crop canopy establishment (March, April, and May) is the best for surface survey. Of course, such efforts are most effective with the help of interested and knowledgeable locals. In addressing the nature of the Mississippian occupation of the terraces and low-lying areas, we need to arrive at an understanding of just how environmentally different this area is from the recent loamy alluvium. The contrast between the uplands with narrow swampy bottom and small creeks and the main overflow swamp with a few high ridges is marked. Are there contrasts in the availability of persimmon and Carya (pecan, hickories, and walnut) between the two areas? GLO witness trees may hold part of the answer to this question. Crowleys ridge may have provided the main source of mast, but the White River bottom may have also had well-developed oak and hickory forests. Acorns were stored for food as late as the 1300s at Powell Bayou in the Sunflower Basin (Starr 1992). In the Old Town Lake area, we should expect a household economy dependent on fish and corn, but what about that in the Western Lowland? Is there less of an aquatic emphasis, as marked by a smaller proportion of mussels, fish, turtles, and ducks in the site refuse? There is no reason to suspect that the older alluvium was critically less productive for agriculture and edible forest products than the sandy levees along the recent channel, but a dispersed (and smaller?) population might would been much better situated for deer sustained local hunting, which should be reflected in contrasting records of diet and health preserved in skeletons. Study of Mississippian subsistence must also focus on the nature of their agriculture, the main source of the diet. We may not have the means to ascertain when a shift from swidden or riverbank planting to permanent field took place, but pollen cores could provide some information on the extent of environmental disturbance. Also of critical importance is the means of storing corn, and the way it was organized. Was the crop the exclusive private property of each household, to be stored individually in cribs outdoors or in pits, jars, gourds, or baskets in the houses, or could it be demanded in tribute by chiefs? At least by the time of DeSoto, the political leaders had a great deal of say in the redistribution of corn and other essential products. We can expect the poor nutritional status of the Coahoma County Humber Protohistoric people (Mitchell 1977) to be duplicated in the large late towns of Phillips County. Indeed, given what is known about the Late Mississippian throughout the South, it would be surprising if anemia, chronic bacterial infections, and similar stresses were not endemic. The date of the establishment of this pattern needs to be assessed. Does an over-dependence on corn originate in the Middle or Late Mississippian? Ethnographically, I think it would be difficult to find a case where a whole people willingly adopted a one-food diet. Why would people allow themselves to become so stressed? Overpopulation beyond the capacity of the local environment to provide game and a fear of neighboring enemies that keep people close to home are the most common reasons. The latter was the reason cited for lack of meat when the Quapaw and the French first made contact. Like the origin of fortification, these problem have critical bearing on the next topic, the evolution (and demise) of complex society. Explication of the Evolution of Complex Chiefdom and Nucleated Settlement. One question not of particular importance to my mind, but one that will be asked by many is: what about Phillips' (1970) Old Town phase? That is, what is the relevance of grouping sites by pottery types? More generally phrased, the question assumes greater importance: what socio-cultural phenomena do phases that are equivalent to ceramic complexes represent? Certainly there is no reason to equate them with political units like cheifdoms, as world-wide anthropology has shown that cheifdoms are often poly-ethnic, and likewise, often of short duration. It seem more likely that they represent linguistic and kinship groups which would have consisted of numerous political groups on a town level of organization, with regional political linkages perhaps not being expressed in ways that are very obvious archaeologically. In other areas of eastern Arkansas, Mississippian occupation of the Western Lowlands falls in the first half of the Mississippian period. If this is the case in Phillips County as well, it seem that the Early Mississippian is very much like Coles Creek in site distribution and site type. Consider that mound building is at least as good an indicator for the evolution of complex society as shell tempering. Does this indicate that adaptations and organization were similar with the adoption of shell tempering being the main defining difference? While the issue has not been addressed here, research into the Early Mississippian will have to take evolution from the Coles Creek base into account. The White River and Monroe County mound sites will be particularly important in this research, but the Broom, Turner, Hall, Tinsley, and Vogel sites in western Phillips County should also be considered. Johnson (1987), in noting that Carson and Buford are at the southern limit of the Burlington chert artifact distribution, has stressed shell bead production as a craft specialization which is directly indicative of the presence of a complex society, in this case directed from Cahokia. This is based on the presence of a skilled application of blade core technology to imported materials, presumed to be used to create elite paraphernalia. If this were the case, Carson should prove to have a shell-cutting area, as well as a core reduction locations. Was Carson or some other site a redistribution point of northern stones, with the blades being distributed based on the social hierarchy? While the wearing of shell beads may have been a status symbol, being engaged in their intensive manufacture may not have been. There are now a number more reported occurrences below Zebree, in the area between Barrett and Carson, but the material has not yet been reported from good excavated context in the lower valley. The Ellis find is from mound fill. Furthermore, Johnson takes the lack of this technology in the Middle Mississippian of the Yazoo basin as an indication of the development of "regional autonomy". Presumably, an inferior status, perhaps requiring bead tribute, relative to Cahokia was no longer felt as the lower valley sites began to acquire large, organized populations and old, complex towns. Shell bead production may not actually cease by 1300, and indeed, several types of shell beads were still in use in the early contact period. Of similar importance is the somewhat later importation of Mill Creek and Dover bifaces, from southern Illinois and western Tennessee, respectively. These are well-documented at the excavated sites in Coahoma County, and are to be expected in the middle and late Mississippian sites of Phillips County as well. Does the social relationship symbolized and maintained by the importation of Missouri Burlington shift to those with more practical purposes, like Mill Creek hoes? How late were hoes imported? The proposition that Ellis was a "vacant" (i.e., lightly populated) ceremonial center surrounded by farmsteads unfortunately cannot be assessed due to the damage done by the levee. However, the contemporary Wilsford small mound group also had scant non-mound occupation (Connaway 1984). Ellis radiocarbon dates cluster around 1450, the date House (1993:28) suggests for the transition from more dispersed to more nucleated settlement in the Kent phase. The Post-De Soto Era. Nineteenth century folklore strongly associates De Soto with the Helena locale. Claiborne (1880:11), in his history of Mississippi, says in this connection When I learned that Old Town Lake, below Helena, had been the channel of the Mississippi, I went to see it, with Irvingfoots "Conquest of Florida," in my hand. I found that the embers and calcined earth of blacksmith's forges, old musket barrels, fragments of saddle-trees and oxydized bullets had there been picked up, and the cruciform handle of a dagger with a cornelian in it. DeSoto built his boats at the base of a range of high hills, that ran to the water's edge. This ridge is now known as Crawley's. There is no reason to think that this was a Spanish as opposed to a French or early British site. Indeed, I have been shown a brown, prismatic blade gunflint such as are considered English in origin which the collector claim came from Old Town Lake. Between the time of initial American settlement around the turn of the nineteenth century and the time of Claiborne's late nineteenth century visit local memories could have lost track of an early settlement at Old Town. Halberds (a pole-arm with a spear-point perpendicular to a hatchet and hook, already obsolete in the sixteenth century) are also attributed in local folklore to De Soto. Around 1932, a H.E. Bushnell of Stutgartt had a halberd with a nearly petrified ash haft found by a mussel-fisher and another halberd had been found by a Minnesota soldier in Helena (Fordyce 1963:11; Dickinson 1987). A few other halberds, of various form, all of uncertain context, have been found in the southeast and attributed to De Soto's army. Correspondence with Jose Parada Carvallo, a military historian in Madrid, revealed that the Arkansas halberds seem to be "mas propias de la primera mitad del siglo XVI [they best fit the first half of the sixteenth century]" and are "de forma…antiguo (con) brazo caido [are of very old form with the hook-shaped arm]" but that "no hay que olvidar la gran libertad que a partir del mismo siglo XVI hubo in la fabricacion de dicha arma [we should not forget the great latitude they had at that time in the manufacture of these arm]" (Dickinson 1987:62). So far, if we discount these possible DeSoto entrada halberds, no evidence has appeared of Indian-European contact in Phillips County. However, local legend recounts that one of the first whites to be born in what is now Phillips County recalled in the late nineteenth century that he had been born in an Indian village five miles north of Helena, probably before 1800, perhaps in the Sylvanus Phillips colony at the mouth of the St. Francis, and had grown up speaking an Indian language (Phillips County Historical Quarterly 1962:1). Where was this village, and who were its people? Coahoma County is showing evidence of occupation in the late 1600s and early 1700s and radiocarbon dates from the West site in Tunica County, Mississippi, likewise support very late, ca.1700, mound construction (Buchner 1996). To efficiently address the question of demographic and sociopolitical change in native society, we must first ask: What are we looking for? At the seventeenth century Menard and Oliver sites, direct evidence of European contact is limited to occasional glass beads, cut copper/brass ornaments, and gunflints. When first contacted, the Quapaw complained that their better-armed enemies were keeping them from establishing relations with the Europeans. But at the same date, around 1700, guns, kettles, and axes, as well as a wide range of other imports, are fairly common archaeologically in the small populations along the Yazoo Bluffs (Brain 1988). There could well have been other people being kept entirely out of this exchange. Was Phillips County abandoned before 1700, and, if so, why? Population nucleation, in contexts a few hundred years earlier, is taken as a sign of increasing social complexity; in the historic period it is considered a response to population crashes and European-encouraged slave raiding. Many diagnostic ceramic and lithic traits in the Phillips County collections indicate sixteenth century occupation, so what role do the Phillips County Protohistoric or terminal Mississippian sites have to play in the origin of the Historic tribes? The earliest contacts with the Quapaw may have taken place in Phillips and Coahoma counties. Records of these encounters are somewhat revealing as to the material culture of the initial contact period and breathe life back into the sherds, postholes, charred corn cobs and split cane, and anemic skeletons of the late village sites. Marqette's first voyage (July 1673) found the large village of the Akamea 8 to 10 leagues below the Mitchigamea, at what they estimated to be 33 degrees, 40 minutes north latitude (a location matching the course of the lower White, but which he had no real way of ascertaining). They were met with canoes of men, with the commander holding a calumet and offering tobacco, sagamite, and cornbread, and led to the "scaffolding" (ramada) of the chief, which was "clean and carpeted with fine rush mats." Sagamite, whole corn, and dog meat were brought in large wood platters, although they had "plates of baked earth" and "great earthen jars, very well made" for cooking corn as well. Corn was stored in large cane baskets and gourds "as large as half barrels". Although the Quapaw made three corn plantings a year, they were "wretchedly provided with food as they dare not go to hunt the wild cattle" for fear of their enemies. They already had watermelons and buffalo hides. They appear to have been surrounded by enemies with guns, who prevented any commerce with the Gulf Coast, and obtained their only hatchets, knives, and beads from the nations to the east and north. The cabins were described as covered with bark, long and wide, with the beds, which are raised two foot above the ground, at the two ends (Thwaites 1959a:153-159). As Belmont (1961:10) points out, the location of "Old Kappa" given on historic maps places it around Old Town. A generation later (October 1700), after the 1698 smallpox epidemic that was the result of more extended stays by LaSalle and Tonti killed three-quarters of at least the Kappa village of the Quapaws, particularly the women and children (Belmont 1961:12), another Jesuit recorded a brief visit with the Quapaw. Kappa village lay 18 leagues below the mouth of the St. Francis and 8 leagues above the Arkansas, in a different location than that visited by Marquette. The old village was recognizable by its "outworks" but no cabins remained. Gravier, the missionary, was met by four pirogues of Akansea and was brought to land to sit with the chief, who offered the French piakimina (persimmon) loaves while the kettles of green corn sagamite and whole corn with squash boiled. When they arrived in Kappa, 40 cabins half a league from the water, they found Montigny, the previous preacher, had erected a cross "on the hill which is very steep and 40 foot high", and were met by two young men with swords, followed by the chiefs of Kappa and Tourima, 20 or 30 bowmen, and a few others with swords and 2 or 3 guns given them by an English trader who had come the year before "with a quantity of goods." The French were taken to the chief's cabin, made to sit on "a mat of canes adorned with figures" and offered green corn seasoned with dried peaches. They stayed less than a day and parted with a present of "a little lead and powder, a box of vermilion to daub his young men with, and some other trifles (Thwaites 1959b:115-135)." Rather than looking for the scarce catlinite, lead balls, iron and brass scraps, and glass beads; dog bones and recently introduced peach, plum, and watermelon seeds might be much more valuable markers of the contact period.

CONCLUSIUON: FUTURE RESEARCH Phillips County is an arbitrarily drawn unit for sampling Mississippian variation. It was selected 1) as a result of my own involvement in Eastern Arkansas research, particularly my participation in the Ellis Mound excavation and write-up, 2) in an awareness that Phillips County was an intermediate area that had been neglected in the Kent phase and lower Arkansas research, and 3) because all of the material I was seeing in eastern Arkansas seemed very similar to the material east of the Mississippi River in Coahoma County. As I have attempted to make clear, a number of important questions remain to be answered before we can confidently say that there is an adequate database for addressing important questions of historical process and evolution. One of the first is: how many sites were there, which ones remain, and what condition are they in? Secondly, is there any basis for dating these sites or breaking the Mississippian period into shorter chronological units? These issues should be addressed before attempting any further formulation of cultural phases. I say to amateurs interesting in looking for new sites, that the small sites can be hard to find even under the best visibility conditions, but that areas of high relief and loamy soil along stream are the best, but by no means only, places to look for Mississippian sites. Garrow & Associates, Inc., has recently excavated part of a ca. 1300 Mississippian house located on Sharkey clay soils near Mound City in Crittenden County, Arkansas (Childress et al. 1995). Collapsed wall materials and ceramics were very well preserved in this heavy gray soil, only a few cm under the surface of a beanfield which yielded only sparse surface sherds. Those of you resident in the area, familiar with the environment, landowners, and other collections are an invaluable resource in this research and I would enjoy hearing from anyone who would like to contribute to the research. The hey-day of pothunting was around 1970, and it began more than 100 years before that, so any information about context associated with these collections (generally only memories) is fading. Much of the material has been dispersed. Efforts should be made to document any private collections from the Old Town phase area. To CRM firm personnel and agency managers, I would like to say that small, cultivated sites producing only a few shell tempered sherds or daub fragments should not be automatically written off. These sites have a quite good potential to produce house patterns, burials, trash-filled pits, and dating samples once they have been stripped. Lack of excavated data from them constitutes a major gap in our Mississippian database for the area, from the standpoint of chronology, settlement pattern, household material culture, and domestic space planning. In addition to the Crittenden County example just cited, I would also remind you of the Panamerican Consultants, Inc., excavations at Barbee-McKnight in Coahoma County, where Mississippian houses and pits were uncovered, despite the fact that Mississippian materials have not been consistently recovered in surface collections from this deep-plowed cotton field. We probably have about as good a record as we will ever have of the mound sites along the Mississippi River, although the Sunflower Landing site (22CO713) shows that more may lie along the levees awaiting discovery (Weinstein et al. 1985) or inside them as may be the case at Ellis and Old Town. Erwin Roemer and Tad Britt (personal communication 1996) of the Vicksburg District, COE, have documented a number of cases where early levee construction simply abutted mounds and, eventually, buried them. This is also the case in Phillips County. Some have been completely eroded away with channel shifts. At any rate, around those sites considered destroyed by levee construction and flooding, engineering work should be closely monitored in an attempt to salvage any information from the site remnants. These areas include Moore/Steagall, Fitzhugh, Ellis, Old Town/Buie, Barney/Rodgers (Modoc), and Ferguson Landing along the Mississippi as well as sites buried under or included in the White River levee and spoil piles along various ditches (see Figure 2). Railroad grades may also have buried and incorporated sites. Perhaps we should have paid closer attention to the construction techniques documented in the old levees in the Helena Harbor project area. I recall seeing mule-trampled, mucky soils; individual slip-loads of dirt resembling basket loads in section; multiple berm addition episodes; and old plowzones, some perhaps attributable to cutting ground to make the application of fill dirt seal better. I am also still convinced that the Marksville materials, including an agreeable radiocarbon date, from 3PH245 were redeposited in a large, grubbed out stumphole. Finally, to students and professional archaeologists, I would like to point out that the area has great potential for research on a number of topics, only some of which have been touched upon here, which need not involve fieldwork. In fact, Phillips County is a wide-open area for Mississippian researchers. A number of existing collections, such as those of the Smithsonian Institution in Washington and the Pillow collection in Nashville, should be documented. There are two large lithic collections, Hornor and Dupree, that have enough specimens for detailed study. Many of the sherd collections are suitable for clay source, mineral pigment, and food residue analysis, as well as AM dating. Cladistic or seriation studies of the ceramic content of the sites must await larger sherd samples. There is a pressing need to find out which of the sites I have described still exist and to systematically collect samples from them. In addition to recording additional components, there is a real need, after revisiting the presently recorded sites, to actively preserve, through purchase, conservation agreement, or easement, the best of them.

BIBLIOGRAPHY Baird, David 1980 The Quapaw Indians: A History of the Downstream People. University of Oklahoma Press, Norman.

Belmont, John S. 1961 The Peabody Excavations, Coahoma County, Mississippi, 1901-1902. B.A. honors thesis, Harvard University Department of Anthropology.

Brain, Jeffery 1979 The Tunica Treasure. Peabody Museum of Archaeology and Ethnology, Harvard University, Cambridge, Massachusetts.

1984 The De Soto Entrada into the Lower Mississippi Valley. Mississippi Archaeology 19(2):48-58.

1988 Tunica Archaeology. Papers of the Peabody Museum of Archaeology and Ethnology 78. Harvard University, Cambridge.

1989 Winterville: Late Prehistoric Culture Contact in the Lower Mississippi Valley. Archaeological Report No. 23, Mississippi Department of Archives and History, Jackson.

Brain, Jeffrey, Alan Toth, and Antonio Rodriguez-Buckinham 1974 Ethnohistoric Archaeology and the De Soto Entrada into the Lower Mississippi Valley. Conference on Historic Site Archaeology Papers 7. Institute of Archaeology and anthropology, University of South Carolina, Columbia.

Brown, Ian 1978 The J.O. Wheeler Collection, West Helena, Arkansas: Pottery of the Kent and Old Town Phases of the Mississippi Period. Manuscript on file, Mississippi Department of Archives and History, Clarksdale.

1979 An Archaeological Survey of Mississippi Period Sites in Coahoma County, Mississippi. Manuscript on file, Mississippi Department of Archives and History.

1990 Historic Indians of the Lower Mississippi Valley: An Archaeologist's View. In Towns and Temples Along the Mississippi, edited by David Dye and Cheryl Anne Cox. University of Alabama Press, Tuscaloosa.

Buchner, C. Andrew. 1996 Mound A Excavations at the West Mounds Site, Tunica County, Mississippi. In Mounds, Embankments, and Ceremonialism in the Mid-South, edited by R.C. Mainfort and R. Walling. Arkansas Archeological Survey Research Series No. 46, Fayetteville.

Cary, Louis E. 1980 "Paddle Wheels to Reels." Phillips County Historical Quarterly 18(2). Helena.

Childress, M.R. 1992 Mortuary Vessels and Comparative Ceramic Analysis: An Example from the Chucalissa Site. Southeastern Archaeology 11(1):31-50.

Childress, Mitchell; Mary Evelyn Starr; Christopher Koeppel; David Crampton; and Guy Weaver 1995 Archaeological Investigations at Helena Slackwater Harbor, Phillips County Arkansas: Volume I: Survey, Testing, and Excavation of Prehistoric Sites and Components. Report submitted to the Memphis District, U.S. Army Corps of Engineers by Garrow and Assoc., Inc., Memphis.

Childress, Mitchell R., Guy G. Weaver, John L. Hopkins, and Marsha Oats 1995 Cultural Resources Investigations at Mound City Plantation, Crittenden County, Arkansas. Report submitted to the Promus Companies, Inc., by Garrow & Associates, Inc, Memphis.

Claiborne, J.F.H. 1880 Mississippi, as a Province, Territory and State. Power and Barksdale, Publishers and Printers, Jackson.

Connaway, John M. 1981 Archaeological Investigations in Mississippi, 1969-1977. Mississippi Department of Archives and History Archaeological Report No. 6, Jackson.

1984 The Wilsford Site (22-Co-516), Coahoma County, Mississippi. Mississippi Department of Archives and History Archaeological Report No. 14, Jackson.

Dickinson, Samuel 1987 -----Arkansas Archeologist 25-26:53-61.

Devereax, J.H. 1873 Catalogue of a Cabinet of Indian relics collected by J.H. Devereax of Cleveland, Ohio, from the years 1848 to 1868, Inclusive, and Presented by him to the Smithsonian Institution. Annual Report of the Board of Regents of the Smithsonian Institution. Government Printing Office.

Dorwin, John and Judith C. Stewart 1977 Long Lake Bayou Cultural Resources Survey and Reconnaissance. Report on file, Arkansas Archeological Survey, Fayetteville.

Flenniken, Jeffrey 1974 A Preliminary Archeological Field Study of the Lee-Phillips Watershed Project, Lee and Phillips Counties, Arkansas. Manuscript on file, Arkansas Archeological Survey, Fayetteville.

Ford, James A. 1961 Menard Site: The Quapaw village of Osotouy on the Arkansas River. American Museum of Natural History Anthropological Papers 48(2), New York.

1963 Hopewell Culture Burial Mounds Near Helena, Arkansas. American Museum of Natural History Anthropological Papers 50 (1), New York.

Fordyce, John R. 1963 "Where DeSoto Crossed the Mississippi." Phillips County Historical Quarterly 2(1), Helena.

Goodspeed Publishing Co. 1890 Biographical and Historical Memoirs of Eastern Arkansas. Chicago.

Griffin, Louise M. 1969 "Indian Mounds in Phillips County." Phillips County Historical Quarterly 7(2), Helena.

Harcourt, James 1991 A Cultural Resources Survey of the Proposed Slackwater Harbor Industrial Park, Helena, Phillips County, Arkansas. Manuscript on file, Arkansas Archeological Survey, Fayetteville.

Hathcock, Roy 1983 The Quapaw and Their Pottery. Hurley Press, Camden, Arkansas.

1988 Ancient Indian Pottery of the Mississippi River Valley. Walsworth Publishing Co., Marceline, Missouri.

Hoffman, Michael P. 1975 Prehistoric Archeological Material from Arkansas at the Smithsonian Institution: An inventory and assessment. Manuscript on file, Arkansas Archeological Survey, Fayetteville.

1990 The Terminal Mississippian Period in the Arkansas River Valley and Quapaw Ethnogenesis. In Towns and Temples Along the Mississippi, edited by David Dye and Cheryl Anne Cox. University of Alabama Press, Tuscaloosa.

House, John 1982 Southeast Study Unit 9: The Kent and Old Town Phases, the Late Mississippian Period. In A State Plan for the Conservation of Archeological Resources, edited by H.A. Davis. Arkansas Archeological Survey Research Series 21. Fayetteville.

1987 Kent Phase Investigations in Eastern Arkansas, 1978-1984. Mississippi Archaeology 22(1):46-60.

1989 Leau Noir Incised Pottery in East-Central Arkansas. Arkansas Archaeological Society Field Notes No. 231.

1991a The Mississippian Sequence in the Menard Locality, Eastern Arkansas. In Arkansas Before the Americans, edited by H.A. Davis. Research Series No. 40, Arkansas Archeological Survey, Fayetteville.

1991b Monitoring Mississippian Dynamics: Time, Settlement and Ceramic Variation in the Kent Phase, Eastern Arkansas. Dissertation presented to Southern Illinois University faculty.

1993 Dating the Kent Phase. Southeastern Archaeology 12:21-32.

House, J.H. and R.B. House 1987 Investigating Early Mississippi Period Occupation in the Lower St. Francis Basin, Eastern Arkansas. In The Emergent Mississippian: Proceedings of the Sixth Mid-South Archaeological Conference, edited by R.A. Marshall. Cobb Institute of Archaeology, Mississippi State University, Starkville.

Jeter, Marvin D. (editor) 1988 The Burris Site and Beyond: Archeological Survey and Testing Along a Pipeline Corridor and Excavations at a Mississippian Village. Research Report No. 27, Arkansas Archeological Survey, Fayetteville.

Johnson, J.K. 1987 Cahokia Core Technology in Mississippi: The View from the South. In The Organization of Core Technology, edited by J.K. Johnson and C.A. Morrow, pp. 187-206. Westview Press, Boulder, Colorado.

Johnson, Jay K. and Fair L. Hayes 1995 Shifting Patterns of Long-Distance Contact During the Middle Woodland Period in the Northern Yazoo Basin, Mississippi. In Native American Interactions: Multiscalar Analyses and Interpretations in the Eastern Woodlands, edited by Michael S. Nannasey and Kenneth E. Sassaman. University of Tennessee Press, Knoxville.

Kirkman, Dale 1964 "Early Days at Helena." Phillips County Historical Quarterly 3(2). Helena.

Klinger, Timothy, and Roy Cochran 1983 Old Town Berm. Report submitted to the Memphis District, U.S. Army Corps of Engineers by Historic Preservation Associates, Inc., Fayetteville.

Klinger, T.C., S.M. Imhoff, and R.J. Cochran, Jr. 1883 Brougham Lake: Archaeological Mitigation of 3CT98 along the Big Creek Enlargement and Diversion, Item 1, Crittenden County, Arkansas. Historic Preservation Associates, Inc., Fayetteville. Reports 83-7, Fayetteville. Submitted to U.S. Army Corps of Engineers, Memphis District.

Leader, Jonathan M. 1990 The Humber-McWilliam Site Brass Headdress: Preliminary Evidence for the Re-use of European Metal by Contact Period Native Americans in Mississippi. Mississippi Archaeology 25(2).

Lumb, L.C. and C.H. McNutt 1988 Chucalissa: Excavations in Units 2 and 6, 1959-1967. Occasional Papers 15, Anthropological Research Center, Memphis State University.

McGimey, C.R., III 1965 The Dupree Site in Retrospect. The Arkansas Archeologist 6(1):3-8.

1989 Quapaw Sites on the Lower Arkansas River. The Arkansas Archeologist 27 and 28.

Mainfort, R.C. 1996 Late Period Chronology in the Central Mississippi Valley: A western Tennessee perspective. Southeastern Archaeology 15(2).

Mason, Otis T. 1880 Summary of Correspondence of the Smithsonian Institution Previous to January 1, 1880, In Answer to Circular No. 316. Smithsonian Institution, Government printing office.

Marshall, Richard 1988 Preliminary Archaeological Testing Near Mound A, Buford (22TL501) Site, Tallahatchie County, Mississippi. Cottonlandia Museum, Greenwood and Cobb Institute of Archaeology, Mississippi State University, Starkville.

Mitchell, Nancy Ross 1977 Paleopathology in Archaeology: The Humber Site (22-CO-601) as a case study. Thesis submitted to the Anthropology faculty, University of Mississippi.

Monette, John W. 1846 History of the Discovery and Settlement of the Valley of the Mississippi 1. Harper and Bros., Publishers, New York.

Moore, Clarence 1911 Some Aboriginal Sites on Mississippi River. Journal of the Academy of Natural Science 14, Philadelphia.

Morse, P.A. 1981 Parkin: The 1978-1979 Archeological Investigations of a Cross County, Arkansas Site. Arkansas Archeological Survey Research Series 13.

Morse, Dan and Phylis Morse 1980 Zebree Archeological Project: Excavation, Data Interpretation, and Report on the Zebree Homestead Site, Mississippi County, Arkansas. M. On file, U.S. Corps of Engineers, Memphis District.

1983 Archeology of the Central Mississippi Valley. Academic Press, New York.

Morse D.F. and S.D. Smith 1973 Archeological Salvage During the Construction of Route 308. Arkansas Archeologist 14:36-78.

Moselage, J.H. 1965 A House Pattern from the Dupree Site. The Arkansas Archeologist 6(1):1-3.

Neller, Earl 1978 St. Francis National Forest Cultural Resources Surveys, October 24-25, 1978. Manuscript on file, Arkansas Archeological Survey, Fayetteville.

Owens, David Dale 1860 Second Report of a Geological Reconnaissance of the Middle and Southern Counties, Arkansas made during the Years of 1859 and 1860. C. Sherman and Son, Printers, Philadelphia.

Peabody, C.S. 1904 Explorations of Mounds, Coahoma County, Mississippi. Papers of the Peabody Museum of Archaeology and Ethnology 3(2). Harvard University, Cambridge.

Phillips County Historical Quarterly 1962 "John Patterson". Phillips County Historical Quarterly 1 (1). Phillips County Historical Society, Helena.

Phillips, Philip 1970 Archaeological Survey in the Lower Yazoo Basin, Mississippi, 1947-1955, 2 vols. Papers of the Peabody Museum of Archaeology and Ethnology 60.

Phillips, P., J.A. Ford, and J.B. Griffin 1951 Archaeological Survey in the Lower Mississippi Alluvial Valley, 1940-1947. Papers of the Peabody Museum of Archaeology and Ethnology 25.

Potts, T.D. 1976 Archaeological Survey of the Will Neill Watershed. Manuscript on file, Mississippi Department of Archives and History.

Ray, Robert 1978 Archaeological Investigations in the St. Francis Ranger District of the Ozark-St. Francis National Forest. Manuscript on file, Arkansas Archeological Survey, Fayetteville.

Redfield, Alden nd The Dalton Project, 1961-1962. Manuscript on file, Arkansas Archeological Survey.

Reynolds, John Hugh 1906 Aboriginal and Indian Remains. The Arkansas Historical Association (1), Fayetteville.

Santeford, Lawrence, and John House 1983 Cultural Resource Overview and Field Survey of the Proposed Helena Chemical Plant Project Area, Phillips County, Arkansas. Manuscript on file, Arkansas Archeological Survey, Fayetteville.

Scholtz, J.A. 1968a Archaeological Sites and Land Leveling in Eastern Arkansas. Arkansas Archeologist 9(1-2). 1968b The Dumond Site (3AR40), Arkansas County, Arkansas. Arkansas Archeologist 9(1-2).

1991 Investigations at the Roland Site, 3AR30, Arkansas County, Arkansas. Arkansas Archeologist 30.

Smith, Gerald P. 1990 The Walls Phase and Its Neighbors. In Towns and Temples Along the Mississippi, edited by David Dye and Cheryl Anne Cox. University of Alabama Press, Tuscaloosa.

Smith, Marvin T. 1987 Ceramics from the French Site (19 P 10). In The Emergent Mississippian: Proceedings of the Sixth Mid-South Archaeological Conference, June 6-9, 1985, pp. 32-37, edited by Richard A. Marshall. Cobb Institute of Archaeology, Mississippi State University, Starkville.

Starr, Mary Evelyn 1984 The Parchman Phase in the Northern Yazoo Basin: A Preliminary Analysis. Appendix to The Wilsford Site (22-Co-516), Coahoma County, Mississippi by John M. Connaway, Mississippi Department of Archives and History Archaeological Report No. 14, Jackson.

1991 Powell Bayou: Part I. Mississippi Archaeology 26(1):1-38.

1992 Preliminary Report on Ceramic Vessels from the 1991 Oliver Salvage. Mississippi Archaeology 27(2):40-55.

1994 Archaeological Investigations in the Central Yazoo Basin: The Oliver, Dockery, and Powell Bayou sites. Report submitted to the faculty of the Department of Anthropology, University of Memphis.

Tesar, Louis Daniel 1975 The Contonlandia Humber-McWilliam Project. Manuscript on file, Mississippi Department of Archives and History, Clarksdale.

Thomas, Cyrus 1891 Catalogue of Prehistoric Works East of the Rock Mountains. Smithsonian Institution Bureau of Ethnology. Government Printing Office. 1894 Report on the Mound Explorations of the Bureau of Ethnology. Annual Report 12, Smithsonian Institution, Bureau of American Ethnology, Washington.

Thwaites, Reuben Gold (editor) 1959a The Jesuit Relations and Allied Documents: Travels and Explorations of the Jesuit Missionaries in New France, 1610-1791 (59). Pageant Book Co., New York.

1959b The Jesuit Relations and Allied Documents: Travels and Explorations of the Jesuit Missionaries in New France, 1610-1791 (65). Pageant Book Co., New York. 1959c The Jesuit Relations and Allied Documents: Travels and Explorations of the Jesuit Missionaries in New France, 1610-1791 (67). Pageant Book Co., New York.

Weinstein, Richard 1985 Some new thoughts on the De Soto expedition through western Mississippi. Mississippi Archaeology 20(2):2-24.

Weinstein, Richard; Susan deFrance; and David Kelly 1985 Cultural Resources Survey in the Vicinity of Sunflower Landing: Investigations Related to the Rena Lara Landside Berm, Item L-628, Coahoma County, Mississippi. Report submitted to the Vicksburg District, U.S. Army Corps of Engineers by Coasal Environments, Inc., Baton Rouge.

William, S. 1980 Armorel: A Very Late Phase in the Lower Mississippi Valley. Southeastern Archaeoloical Conference Bulletin 22:105-110.

• Home • Up • About Starr • Papers • Resources •

• Walls Lithics • CMV Architecture • CMV Ceramics • Phillips Co • Smithsonian •

•  •

• Home • About M.E.Starr • Vita • Services • Campfire Tales • Papers • Mississippian • Walls Lithics • CMV Architecture • CMV Ceramics • Phillips Co • Smithsonian • Historic • Gulf Ordance • Mish • Chancey • Stoneware • AR Potters • MS Potters • Ft. Desha • Historic Lithics • Historic Ceramics • Resources • Glossary • Reading List • Journal Abstracts • Delta Counties • Environment • Physical • Biological • Neolithic • Farming • Tourism • Links •

Contact: Mary Evelyn Starr

  Box 39, Sledge MS 38670

  Phone (662) 444-5254

E-mail me at 

Keep up with me on the web and your social network
Visit:
http://360.yahoo.com/mestarr (Starr's Science and History Blog)
http://blog.myspace.com//maryestarr (weather, crops, garden, family, books, bands)

http://deltaarchaeologyannex.blogspot.com (google technorati blog)

And

mary e starr at:

www.hi5.com   ¨  www.facebook.com   ¨  www.everyonesconnected.com   ¨  www.multiply.com

Better a cruel truth than a comfortable delusion. There is a kind of poetry in simple fact.
The world is older and bigger than we are. This is a hard truth for some people to swallow.
Edward Abby Vox Clamantis in Deserto 1989

Got questions or comments about this web site?

Please send mail to

Hosted by: Tech-IT-Out

Copyright © 2003 - 2009 Delta Archaeology
Last modified: 26 March 2009