Across Atlantic Ice

by Dennis J. Stanford

  Adovasio points out that the bladelet cores from the Cross Creek drainage are reminiscent of cores from northern China that are around 30,000 years old. He speculates that this Asian technology might have been a prototype that spread into North America during the Last Glacial Maximum and eventually gave rise to the Miller Complex. If this speculation is correct, there should be sites containing the proto-technology that become progressively younger as they approach Meadowcroft along the 10,000-or-so-mile trail. So far none have been discovered.

  The final verdict on the earliest cultural levels at Meadowcroft remains disputed, and many critics have required (perhaps unreasonably) that other sites with Miller Complex artifacts and equally early radiocarbon assays be found before they accept the early dates at Meadowcroft.8 In our opinion the demand for a second site to validate the antiquity of the Miller Complex was satisfied by the discovery of similar artifacts at the Cactus Hill Site in southeastern Virginia. This site contains a relatively unmixed stratigraphic sequence of Mid-Atlantic cultures from circa 22,000 years ago to historic times in the seventeenth century. Of particular interest are the earliest levels, which are situated below the Clovis occupation level.

  FIGURE 4.2.

  Meadowcroft (a–d) and Krajacic (e–g) artifacts from the Miller Complex: (a) obverse, reverse, and cross section of projectile point; (b) obverse and reverse sides of retouched flake tool; (c) retouched bladelet; (d) obverse, reverse, and side view of bladelet; (e–f) obverse, reverse, and side view of biface fragments; (g) bladelet core.

  Two research teams have investigated this location. Joe McAvoy and his team from the Nottoway River Survey excavated several areas, and Mike Johnson and volunteers from the Archaeological Society of Virginia excavated other parts of the site. McAvoy’s detailed report of the ongoing interdisciplinary investigations is the basis for our inferences about the significance of Cactus Hill.9

  Cactus Hill is on the east bank of the Nottoway River in Sussex County, Virginia, some 13 miles east of the Fall Zone, a limestone escarpment that defines the eastern edge of the Virginia Piedmont. During the early occupation of the site, the river meandered its way across 180 miles of the Atlantic Coastal Plain until it drained into Albemarle Sound on the ice age coast of North Carolina. A series of prehistoric campsites were situated on an east-west-trending sandy ridge deposited on the first terrace of the river at roughly 80 feet above sea level.

  The ridge was formed by wind-blown sand from riverine deposits and by sediments deposited during the flooding of a backwater channel adjacent to the site. A flood-scoured wetland basin south and west of the sand ridge was a source of water and a diverse array of game animals and edible plants, seeds, and fruits. Quartzite cobbles deposited along the banks of the Nottoway and alluvial point bars provided an ample source of stone raw material, which enhanced the lure of this well-situated campsite.

  Sand began to accumulate on the terrace around 22,600 years ago, and the dune underwent a series of building, stable, and deflationary phases, reflecting periodic local and regional environmental perturbations. Plant life was established and soils developed when the dune was stable, whereas drier and windy conditions caused deflation. Sometime between 20,100 and 22,600 years ago, people began to use the sand ridge as a temporary campsite. The sand apparently ceased to accumulate either during or just after the earliest occupations, creating a stable surface that allowed soil to develop. Later there was a new round of dune building and subsequent deflation. The deflation removed the upper part of the buried soil, leaving only slight traces of its former existence. Eventually, sand once again began to build up over the crest of the dune. How much sand may have accumulated during this episode is unknown, but when Clovis hunters established a camp on the ridge about 12,900 years ago, only three to six inches of sand separated their living floor from the earlier occupation some 7,000–8,000 years older. However, faint traces of the former soil were identified between the two occupation zones, clearly indicating that the two occupation levels represented separate and independent camping events. After Clovis times there was a relatively constant accretion of sand until the modern ground surface was established.10

  Since depositional and preservation conditions eliminated obvious stratigraphic boundaries and only a thin layer of sand separates the Clovis and underlying artifacts, it may be that a natural process, through which animals, plants, or even people churned up the sand, caused some of the Clovis artifacts to work their way down into a lower stratum and thus gave a faulty indication of very early Clovis-like materials. If this were true, however, one would expect the stone materials in the lower levels to include the types used in the upper levels, and this is not the case. The flintknappers of the early occupation primarily flaked the local quartzite, whereas the Clovis craftsmen worked a wide variety of cherts, jaspers, and other more exotic stones. Hence, we agree with the site investigators that trickle-down redeposition was not a significant factor and that the occupation levels are not mixed.

  Of major importance are two areas where charcoal concentrations indicated possible hearth features. The youngest, represented by a scatter of white pine charcoal, produced a date of 15,070±70 RCYBP (18,450 years ago). Seven quartzite flakes and three blades were found associated with this charcoal scatter. A second hearth-like feature somewhat deeper than the first was dated to 16,670±730 RCYBP (19,800 years ago). Another dating method, called Optically Stimulated Luminescence (OSL), produced similar results supporting the accuracy of the original dates.

  Burned phytoliths (silicate particles formed in plant cells) were found in these features and are not part of the general background noise in the sediments.11 Their presence favors the identification as fire hearths rather than simply scatters of charcoal flecks. Additionally, geochemical tests revealed high traces of phosphorus in the strata where the artifact clusters were found, while phosphorus traces were low in the non-cultural strata. The high phosphorus levels are attributable to concentrated biological refuse resulting from human activities. All of these tests support the integrity of the site.

  Three distinct artifact clusters were also encountered, two associated with the hearths. The upper cluster contained two bifacial projectiles (figure 4.3a–b), a retouched biface flake (figure 4.3c), and five small bladelets averaging 29 millimeters long, 16 wide, and 3.6 thick (figure 4.3d–e). The middle cluster consisted of two concentrations of nine quartzite blades (figure 4.3f) associated with the 18,000-year-old hearth feature. These blades are larger than those from the upper cluster, averaging 66 millimeters long, 28 wide, and 7.8 thick. The third cluster contained six incomplete blades or broken tools made on blades. These overlap the range of sizes of Clovis blades, but they tend to be shorter while maintaining the width-thickness dimensions of Clovis blades.12 There is also a scraper plane (figure 4.3g) and a flake core (4.3h).

  As in stratum IIa at Meadowcroft, the artifact assemblage from these lowest levels at Cactus Hill is sparse. The waste flakes and most of the blades and unifacial tools are made almost exclusively from the local quartzite, whereas the two projectile points are made from a metavolcanic rhyolite from a more distant source.

  Mike Johnson independently conducted research in another locality at the Cactus Hill Site.13 While working there he became familiar with the various quartzites present in the gravels of the nearby Nottoway River. These were the same materials used by the earliest inhabitants of the Cactus Hill Site. Being curious about the flaking qualities of this material and an accomplished knapper, he focused on learning all he could about how this quartzite flakes and especially if it has any apparent limitations. While at first glance it looks coarse, tough, and intractable, with proper flaking techniques and tools it can be a reasonable material. This is attested to by the large number of well-made quartzite projectile points and knives from various time periods in the area.

  On the other hand, these quartzites pose some interesting challenges to a knapper. They are limited in size, with those 15 by 10 by 5 centimeters bein
g at the large end. The exterior surfaces are water worn and smooth. The shapes are uniformly rounded ovoids. These characteristics make the cobbles difficult to break into. Once knapping starts, flakes tend to break off with step terminations, especially as they get thinner. The stone is coarse and hard but also quite brittle. Johnson has developed a low-energy antler flaking technique that seems to work best, showing that this material may be flaked in a controlled manner.

  FIGURE 4.3.

  Cactus Hill artifacts: (a–b) obverse, reverse, and sections of projectile points; (c) retouched flake; (d–e) bladelets, (f) blade; (g) scraper plane; (h) flake core.

  These quartzites can be used to make decent bifacial tools like knives and projectile points, but it might be difficult to use the cobbles for the production of large blades. It is, however, quite easy to produce smaller bladelets from the quartzites. Even if the Cactus Hill flintknappers were accustomed to making large blades, it is unlikely that they would have made them from the local quartzite. The difficulties of flaking these materials may be the reason that a large blade technology is not present at Cactus Hill. Significantly, the same sorts of quartzites are the main raw material available across much of the Vasco-Cantabrian region of northern Spain, where large blade technology is also rare. The large blades and cores found in that region were made from non-local cherts and flints.

  The Cactus Hill projectile points are small, sub-triangular lanceolate bifaces with slightly indented bases. They are relatively thin and have slight to no basal grinding. Although McAvoy describes the point technology as being highly controlled percussion, based on our knapping experience, we consider much of the final flaking to be selective pressure. It is probable, as McAvoy suggests, that these points have been reworked from larger broken lanceolate weapon tips, much like the Miller point from Meadowcroft.

  If Meadowcroft and Cactus Hill are not proof enough of a pre-Clovis occupation of eastern North America, a site found actively eroding from pre-Clovis deposits should clinch the argument. Darrin Lowery, while a graduate student in the Department of Geography at the University of Delaware, discovered Miles Point while surveying for archaeological evidence along the Eastern Shore of the Chesapeake Bay.14 At the time of occupation the site was on the Miles River, a second order tributary of the Susquehanna, but today the intersection of these rivers is submerged under Chesapeake Bay. The presence of several freshwater springs at the site likely attracted early hunters to the locality.

  Lowery recognized the significance of a large quartzite boulder protruding from an LGM loess (wind-lain sediment derived from glacial deposits) exposed in a wave-cut vertical bank. He and the geologist Dan Wagner cleaned the boulder’s profile to assess its depositional situation. Their efforts were rewarded by the discovery that the boulder was not only in situ, in its original undisturbed position, but had been used by a flintknapper as an anvil. The boulder had impact depression scars on its upper surface, and two quartzite bladelets (figure 4.4a–b), an exhausted bi-polar bladelet core (figure 4.4c), and a hammerstone were arrayed around it. A projectile point (figure 4.4d), a biface preform (figure 4.4e), a burinated biface fragment (figure 4.4f), another quartzite blade and several retouched flake tools (figure 4.4g), a polyhedral bladelet core (figure 4.4h), a decortication flake (figure 4.4i), and a flake core (figure 4.4j) were recovered during subsequent trips to the site for mapping and collecting soil and dating samples.

  The artifacts are made of raw materials that are locally available in exposed gravel deposits. The cherts found in these gravel deposits were washed downriver from bedrock sources in Pennsylvania, but unfortunately as they tumbled downstream they were battered into fist-sized cobbles, limiting the potential size of the artifacts to be manufactured. Early flintknappers would likely have explored upriver to seek larger cobbles, perhaps ultimately discovering their bedrock sources. Such expeditions up the Susquehanna would have taken the paleo-explorers into Pennsylvania and New York. Perhaps they even carried cherts from newly found sources to Meadowcroft Rockshelter.

  The occupation level at Miles Point is about 10 centimeters below a paleosol (an ancient soil) that developed in the LGM loess (figure 4.5). The soil, tentatively termed the Tilghman Paleosol, is a horizon marker—a distinctive deposit found over a wide area that can be recognized as a stratigraphic horizon having a single geological source—with calibrated radiocarbon dates ranging between 16,300 and greater than 24,000 years old at four localities throughout the region.15 It may be equivalent to a climatic warming phase known as the Alerød/Bølling interstadial in Europe. It was during this climatic episode that the productivity of grasses and other plants increased, causing soils to become highly organic and readily recognizable by their dark color. This warming phase was rapidly truncated around 12,000 years ago by a cold, dry, and windy period known as the Younger Dryas (YD), when windstorms scoured and deposited loess over the paleosols. Many Clovis sites on the Delmarva Peninsula were destroyed by YD deflation, and their artifacts were redeposited, along with lag gravels, on the interface surface between the truncated Tilghman soil and YD deposits. It is unclear why the dates from the Tilghman soil are older than the Allerød/Bølling dates.16

  FIGURE 4.4.

  Miles Point artifacts: (a–b) bi-polar blades; (c) bi-polar core; (d) projectile point; (e) biface preform; (f) biface fragment used as a burin, with arrows indicating multiple burin spall removals; (g) retouched flake tool; (h) polyhedral bladelet core; (i) decortication flake; (j) large polyhedral flake core.

  FIGURE 4.5.

  Miles Point stratigraphy showing in situ quartzite anvil. (Photo by Darrin Lowery)

  Cactus Hill is roughly 80 miles across the Chesapeake Bay from the Eastern Shore, and it is highly likely that the paleosol separating the early occupation from the Clovis occupation at that site was formed at the same time and under the same environmental conditions as the Tilghman Paleosol.

  Test excavations by Lowery at Paw Paw Cove on Tilghman Island, Maryland produced an in situ Clovis occupation level on the upper surface of the paleosol.17 Moreover, many Clovis artifacts have been found in lag deposits on the surface of the paleosol at other localities on the Delmarva Peninsula. The stratified context of these artifacts is clear evidence of human activity in the area of the Miles River in Maryland during the LGM.

  Additional evidence supporting an Atlantic LGM occupation comes from Oyster Cove, Virginia, where a projectile point similar to the one from Miles Point was found protruding from the Tilghman Paleosol exposed in a steeply cut bank (figure 4.6a). A small blade made from a chert cobble was also recovered from the beach (figure 4.6b). A bulk soil assay of carbon collected and combined from a soil sample provided an estimate of 25,800±120 RCYBP for the paleosol. While bulk soil dates may not be as accurate as dates from hearth features, the radiocarbon assays at seven different exposures of the Tilghman Paleosol on the Eastern Shore have all clustered within 1,000 years of one another.

  FIGURE 4.6.

  Oyster Cove artifacts: (a) projectile point; (b) bladelet.

  During the spring of 2008 Lowery was conducting another shoreline archaeological survey, in Mathews County, Virginia.18 While visiting the Gywnn’s Island Museum, he noted a large Solutrean-style laurel leaf biface (figure 4.7) on exhibit with a mastodon third upper molar and tusk section. A label in the exhibit indicated that in 1970 the crew of the vessel Cinmar dredged all of these specimens at the same time while harvesting deep-sea scallops 100 kilometers off the coast of the Virginia Capes. In an interview with the vessel’s captain, Thurston Shawn, Lowery determined that the artifact and mastodon remains came from a site approximately 75 meters beneath sea level on the outer edge of the continental shelf. Shawn mentioned that several other well-preserved teeth and tusk sections, along with skull bones, were distributed as souvenirs among the crew members. Because of the unusual nature of the assemblage, he took particular note of the water depth and plotted the area on his navigation chart.

  From Shawn’s bathometric information
it is clear that the remains came from the LGM shore of the James Peninsula, adjacent to the South Lagoon of the ice age Chesapeake Bay (figure 4.8). The South Lagoon separated the mainland from James Peninsula Barrier Islands and created an excellent sheltered habitat for mastodons. It is likely that this combination of physiographic features also protected the site from erosion during the sea level rise at the end of the last glaciation.19

  Lowery arranged for the Smithsonian Institution to borrow the artifact, tooth, and tusk segment for analyses.20 The wear on the molar and the small diameter of the tusk tell us that the mastodon was a small female, about thirty years old when she died. A calibrated radiocarbon assay of collagen extracted from the tusk determined that she lived nearly 23,000 years ago. Based on LGM sea level calculations, the date is consistent with a terrestrial landscape; moreover, the site was beneath 15 meters of water by the time Clovis people occupied the Chesapeake region, indicating that this animal died long before Clovis people were in the area (figure 4.9).

  FIGURE 4.7.

  Cinmar laurel leaf biface, obverse and reverse.

  The laurel leaf is 188 millimeters long, 55 wide, and 6 in maximum thickness. It exhibits well-controlled percussion-thinning flaking on both faces. Use-wear studies demonstrate that it was a hafted knife, and non-invasive retouch along its distal margins resulted from resharpening dulled edges. The use-wear traces are consistent with those found on prehistoric knives recovered from mammoth and mastodon butchering sites, as well as those seen on stone knives used during butchering experiments conducted on modern elephants. X-ray studies indicated that the knife is made from a banded meta-rhyolite from South Mountain on the Pennsylvania-Maryland border near Emmitsburg (figure 4.8).21 The rhyolite used for this knife was of only fair flaking quality, leaving little question that the artifact was made by a highly skilled knapper. A much finer, better knapping grade of banded rhyolite was quarried at another locality, perhaps by the flintknapper who left waste flakes at the Thunderbird Clovis site near Front Royal, Virginia. This might indicate that the flintknapper who produced the Cinmar biface was unaware of the better quality stone that could be quarried elsewhere and that early Mid-Atlantic people had not yet thoroughly investigated the source area.