Here you'll find a complete listing of all of the main project's associated figures and tables.  For figures with less than desirable graphics at an easily downloadable size, there is a link to a higher resolution version below.  Choose whichever matches your internet speed and/or patience level.

Table 1

Comparison of the ages before present (BP) of the Last Glacial maximum (LGM) black mat formation events at Lake King in Trans-Pecos Texas, paleosol formation events in the dunes of the Dempsey Divide, and peaks of stream discharge into Lake Estancia near Albuquerque. The three records combined indicate a 400 year rainfall cycle was operating in the Southern Plains region between 26,000 and 17,000 years ago.

Table 2

Radiocarbon dates on which the  Dempsey Divide Late Holocene pluvials have been defined. A legend of the locality/site abbreviations is provided at the end of the table.

Table 3

Rainfall record for the region encompassing the Dempsey Divide study area since 1893. The sources of this data are detailed at the end of the table.

Figure 1

High Resolution

Three dimensional (orthographic) map of the Southern Plains, looking to the northwest. The area encompassed ranges from Guadalupe Peak on the southwest (lower left), to the Sangre de Cristo Mountains north of Santa Fe on the northwest, to north-central Oklahoma on the northeast, to Waco on the southeast.

Figure 2

High Resolution

A closer view in 3-D of the terrain surrounding the Dempsey Divide study area in the Texas panhandle and the western half of Oklahoma. Note how the interfluvial divides between the rivers of western Oklahoma lead up onto the crest of the High Plains. These divides would have formed natural highways for prehistoric east-west travel, being less dissected and less heavily vegetated than the river bottoms

Figure 3

Surface geology of the Dempsey Divide study area in northern Beckham and southern Roger Mills counties, Oklahoma. Note that a peninsula of basal Ogallala Formation occupies the crest of the divide. Archaeological survey along the northern edge of this outcrop has detected a striking concentration of prehistoric campsites. From Thurmond 1991b: Figure 4.

Figure 4

Distribution of recorded prehistoric campsites in the Thurmond Ranch archeological survey area. Note the concentration of large campsites along the boundary between the Ogallala and Permian redbed outcrops, and the rapid decline in campsite size and density as one proceeds northward into the Permian outcrop area. Areas outside the Thurmond Ranch boundary have not been surveyed for the presence of archaeological sites.

Figure 5

Benthic d¹⁸O record from Deep Sea Drilling Project Site 607 (41°N, 33°W) in the North Atlantic, plotted to the paleomagnetic time scale. The time line reads from right to left. This is a proxy for sea surface temperature. The inset at top is an expanded view of the record for the last 15,000 years. Note that pronounced cooling began around 2.9 Ma (million years ago). The glacial cycle began at 2.4 Ma, and operated on a 41,000 year cycle until about 700,000 years ago, when it transitioned to a 100,000 year cycle. Each 100,000 year glacial cycle terminates in a 10-15,000 year long "interglacial" of warmer and more stable climate. We are living in such an interglacial now, which began about 10,000 years ago. From Raymo 1994: Figure 1.  

Figure 6

High Resolution

Terrain of the Canadian River valley in the vicinity of the Hajny Mammoth locality, which is in Section 16-17N-15W, Dewey County, Oklahoma. The view is to the northeast. The alluvial terrace system of the Canadian is clearly visible as a sequence of progressively higher benches above the modern river. Each terrace represents an interglacial period of climatic stability in the 100,000 year glacial cycle of the last 600,000 years. The downcutting events that separate these terraces occurred in the wetter, gyrating climate of the full glacials.

Figure 7

Closer view of the terrain around the Hajny locality, looking to the northeast. The skeletal remains of three mammoths were found around an ancient spring outlet (Wyckoff et al. 1992).

Figure 8

High Resolution

Reverse view from that of Figures 6 and 7, looking to the southwest from the valley of the Cimarron River to the south side of the Canadian River valley, opposite the Hajny locality. The Putnam volcanic ash (Section 24-16N-16W) at the top of the Canadian River terrace system has been fission-track dated and identified as deriving from the Lava Creek B eruption at Yellowstone of 610,000 years ago (Ward 1991a-b).

Figure 9

Olson Dune #1 atop the crest of the Dempsey Divide in Section 21-12N-24W, Roger Mills County, Oklahoma. The positions and radiocarbon ages of dated soils within the dune are shown. View to the northwest.

Figure 10

View to the northeast of the upper Brokenleg Creek basin in Sections 4/5-12N-24W and Sections 33/34-13N-24W, on the west side of the Thurmond Ranch archaeological survey area in southern Roger Mills County, Oklahoma. The canyon has eroded down into redbeds of the Permian Elk City Sandstone and Doxey Shale formations. The white (gleyed) sediments at upper left mark the position of the Brokenleg Creek valley floor during the Late Pleistocene. See Figure 11 for a cross-section of these sediments

Figure 11

Brokenleg Bend Exposure #1, Section 4-12N-24W, Roger Mills County, Oklahoma. View is to the northeast. The white banded sediments at lower right record a spring-fed perennial pond spanning the Last Glacial Maximum. The profile is transected by an erosional unconformity that dips to the left. Above the oxidized units associated with this event is a series of marsh soils largely dating 11,500 to 9,000 years ago. This profile stands some 70' (21 m) in elevation above the modern canyon floor

Figure 12

Sergeant Major Creek Exposure #1, Section 35-13N-24W, near the center of the Thurmond Ranch. This is a laminated stream deposit of sediments derived almost entirely from the Ogallala Formation, the modern outcrop edge of which lies 1.25 mi (2 km) to the south. The virtual absence of Permian redbed clastics in this deposit indicates that the locality was within the outcrop boundary of the Ogallala Fm. 17,000-15,000 years ago. In other words, the Ogallala boundary has eroded over a mile to the south in the last 15,000 years. Like Brokenleg Bend #1, this profile stands some 70' (21 m) in elevation above the modern creek.

Figure 13

Oxygen isotope curve for the last 40,000 years from the GISP-2 ice core in central Greenland. This is a proxy for average annual air temperatures over Greenland, as indicated on the right side of the graph. The timeline reads from right to left. Note that the temperature swings before 10,000 years ago were orders of magnitude greater. Before the current interglacial, average annual temperatures in the North Atlantic region were gyrating by as much as 30°F, and these changes were happening in no more than a decade (the sample resolution of the record). The heavy black bars delimit the Bolling-Allerod global climatic warming of ca. 15,500 years ago. Compare to the dates from Brokenleg Bend #1 and Sergeant Major Creek #1 in Figures 11 and 12. From Maher and Mickelson 1996: Figure 8.

Figure 14

Correlation of Dempsey Divide habitation events with the Holocene fossil shrew record from Hall's Cave on the Edwards Plateau of central Texas. These two species have opposite moisture requirements, and their alternating dominance over time records long-term fluctuations in effective precipitation on the Southern Plains. Note that each of the major occupations of the Dempsey Divide coincides with a period of greater effective precipitation as indicated by the Hall's Cave record. The long drought of the Altithermal stands out above all else. Pond and marsh deposits accumulated in the upland valleys of the Dempsey Divide from Late Pleistocene times until about 9,000-8,000 years ago. Deep canyons were eroded in these valleys during the Altithermal, and began rapidly refilling with dry sediment around 3300 years ago. Adapted from Toomey et al. 1993: Figure 6.

Figure 15

Selected Late Paleoindian (10,000-8,000 BP) projectile point types of the Southern Plains. From Thurmond 1990: Figure 7. Specimens referable to the types Plainview and Meserve (which are really just resharpened Plainviews) are the most common within the Dempsey Divide study area.

Figure 16

The effects of resharpening on projectile points of the type Calf Creek. This type may mark an incursion from the east onto the Southern Plains during a brief climatic amelioration around 5000-4000 BP, in the middle of the Altithermal.

Figure 17

34RM507, West Profile #2, Section 29-13N-23W, Roger Mills County, Oklahoma. Radiocarbon dates on charcoal in this laminated canyon fill indicate rapid refilling of the canyons on the Dempsey Divide at the end of the Altithermal, from about BC 1300 to AD 100. Distinct bands of high charcoal density in the profile, presumably reflecting periods of frequent range fires, suggest periods of greater rainfall at BC 500-300 and BC 900-700. This may seem counter-intuitive, but coming out of the Altithermal, it probably took considerable increases in rainfall just to generate sufficient biomass for a fire to carry. We infer that a first relatively robust post-Altithermal pluvial around BC 1300-1100 initiated the accretion of sediment in Higgins Canyon at 34RM507. The earliest occupation at this site dates AD 30-45.

Figure 18

A broader view of West Profile #2 at 34RM507. Note the small gully filled with dark sediment at the top right of the profile, underneath the people screening; the slip-off slope leading down to the left; and the deep creek channel at left, also filled with dark sediment. The latter is actually two sequential channel cut-and-fills, one atop (and dissecting) the other. The fill in the gully and the two creek channels is organic enriched soil containing a jumble of rocks and artifacts. These fills appear to have been stripped off upslope and/or upstream and redeposited in individual catastrophic events, presumably as a result of torrential rainfall. The soil in the gully radiocarbon dates to AD 420. The fill in the lower creek channel dates AD 785, and that in the upper channel dates AD 1035. These are not the dates of the erosional events, but the age of the material that was washed into them. Again, the massive structure, uniform character, and chaotic orientation of the artifacts and rocks contained within these fills make it clear that they were indeed washed in all at once, and not formed in place. The three erosional events thus date some time before AD 420, 785 and 1035. The first date falls within the Herring Creek pluvial (AD 400-600), and the last two near the boundaries of the Higgins Creek pluvial (AD 775-1000). We may be seeing the results of torrential rainfall events at or near the ends of these two pluvials, stripping off and redepositing soils formed during each. The slip-off slope indicates a longer term and larger scale canyon incision event, possible during the AD 100-400 Herring Creek interpluvial.

Figure 19

Corner-notched dart point, type Marcos, of silicified wood, and lunate stone of Kiowa Formation fossiliferous shale from the Late Archaic burial at 34RM208C, Section 5-12N-24W, Roger Mills County, Oklahoma (Thurmond 1991a). We believe these unusual boatstones are the cultural markers of the people who reinhabited the Caprock Escarpment borderlands some time after BC 50.

Figure 20

Chipped stone tools from 34RM446, a typical Late Archaic Twilla phase campsite and chipped stone tool workshop in the Thurmond Ranch archaeological survey area, Section 33-13N-24W. The dart points from 34RM446 are virtually identical to those which have been recovered from Late Archaic bison kills in northwestern Texas and western Oklahoma (cf. Bement and Buehler 1994, Hughes 1973).

Figure 21

High Resolution

Hypothesized path of the Hopewellian Late Archaic immigrants to the Caprock Escarpment borderlands, presumably from south-central or southeastern Kansas. After Thurmond 1991a: Figure 5

Figure 22

Chipped stone tools from 34RM773, a typical Late Woodland Beaver Dam phase campsite in the Thurmond Ranch archaeological survey area, Section 2-12N-24W. Moderately thick-walled, coarse-paste cordmarked pottery is frequently found on Beaver Dam phase sites in association with corner-notched arrow points of the styles shown here. See Thurmond 1997.

Figure 23

Relative positions of the Dempsey Divide/Thurmond Ranch and Quartermaster Creek study areas in western Oklahoma. We believe the people who inhabited the Dempsey Divide as broad-spectrum hunter-gatherers during the first millennium AD adopted a Plains Villager lifestyle of horticulture and bison hunting around the AD 1000 end of the Higgins Creek pluvial and relocated to the Quartermaster Creek watershed. See Thurmond 1991b and Figures 24-25 on this website.

Figure 24

Relative intensity of habitation of the Dempsey Divide and Quartermaster Creek study areas. Compare the Late Archaic/Woodland and Late Prehistoric bars. Prehistoric exploitation of the Dempsey Divide plummeted at the same time it skyrocketed on Quartermaster Creek. After Thurmond 1991b: Figure 19.

Figure 25

Aerial photo of Hay Creek, one of the principal tributaries of Quartermaster Creek in northeastern Roger Mills County, Oklahoma. Alfalfa hay and wheat are grown today on the excellent farmland of the creek bottom. The native biota comprised a limited species mix, however, unsuited to exploitation by broad-spectrum foragers. Again, few Late Archaic/Woodland sites have been found here, but many Plains Village sites have been recorded (Baugh et al. 1984, Moore 1988).

Figure 26

Brokenleg Bend Exposure #3, Section 34-13N-24W, Roger Mills County, Oklahoma. This was the first exposure of Late Holocene buried soils at which we detected the 400 year rainfall cycle. Four faint, weakly melanized (darkened by organic enrichment) soils are barely visible, separated by valley fill devoid of organic matter. The organic-enriched soils formed during times of greater average annual rainfall, when the vegetation was more lush. Sediment deposition was basically continuous throughout the sequence. The time spans indicated are based on calibrated radiocarbon dates of soil organic matter from the upper and lower 5 cm of each soil.. See also Table 2 on this website.

Figure 27

Antelope Hills Exposure #1, Section 33-17N-25W, Roger Mills County, Oklahoma. The same sequence as at Brokenleg Bend #3, but much more clearly expressed. Note the diffuse lower boundary and abrupt upper boundary of each soil. These suggest that rainfall and vegetal biomass gradually increased through each of the four recorded pluvials, as during the modern one (Figure 28 on this website), and that each pluvial terminated in a drought of a severity that has not been experienced on the Southern Plains during the post-1892 Historic period.

Figure 28

Running averages of annual rainfall for the Dempsey Divide study area since the instrumental record began in 1893. Running averages smooth out the interannual variability in rainfall to detect long-term trends. The upper graph is a five year running average, in which each year is averaged with the two before it and the two after. This detects decade-scale patterns, and a roughly 11-13 year cycle is apparent, possibly driven by the 11 year solar cycle and in synch with a strong 12-13 year cycle in Atlantic ocean-atmosphere dynamics (Black et al. 1999). In the lower graph, each year is averaged with the seven before and seven after to detect century-scale trends. Note the upward progression of the record, interrupted by the severe climatic correction of the Dust Bowl and the decades which followed. The steady increase in average rainfall since 1970 has recapitulated the record of 1895-1925. See also Table 3 on this website.

Figure 29

Evidence of century-scale cycles in solar output:

A: Fourier spectrum of the variations in the d¹⁴C of Earth's atmosphere from BC 5300 to AD 1500, as observed by the La Jolla, California radiocarbon laboratory in bristlecone pine growth rings. A Fourier algorithm sorts out any inherent cycles in a long term record. The ¹⁴C content of Earth's atmosphere varies inversely over time with solar output, so this record is a proxy for cycles in solar output. Note the pronounced spectral peak at 202 years. From Suess and Linick 1990: Figure 2.

B: Fourier spectrum of the magnesium/calcium time series from Rice Lake, North Dakota for the last 2100 years. Calcium is selectively removed from this closed basin lake during periods of drought by the precipitation of calcium carbonate, increasing the Mg/Ca ratio. Change in this ratio is therefore a proxy for drought conditions. The Late Holocene Rice Lake record shows the same cyclicity as that of the Dempsey Divide; note the spectral peaks at 200 and 400 years. However, the two cycles, and thus the climates of the Northern and Southern Plains, are in antiphase. On a century scale, it is dry on the Northern Plains when it is wet on the Southern Plains, and vice-versa. From Yu and Ito 1999: Figure 3.

Figure 30

The record of changes in atmospheric concentration of the greenhouse gases carbon dioxide (CO₂) and methane (CH₄) over the last four glacial cycles from the Vostok ice core in Antarctica. This data comes from the direct assay of air bubbles sealed in glacial ice on an annual basis over the last 400,000 years. The upper line graph is carbon dioxide, the lower one methane. The middle graph is for Oxygen 18, a heavy isotope that serves as a proxy for temperature variation. The timeline reads from right to left. Note how closely temperature and greenhouse gas concentration have covaried over the last 400,000 years. Also note that the current levels of carbon dioxide at 365 ppmv and methane at 1600 ppbv are without precedent in this extraordinarily lengthy record. From Raynaud et al. 2000.