Evidence of Holocene sea levels higher than the present level have often been reported from the Northwest Pacific. Eustatic interpretations have been propounded, but age and level of the maximum transgression vary with each new analysis. In this investigation, after an inventory of approximately 250 items of data, some of which are new, a tentative synthesis transcending local scale is advanced. The highest levels are reported from Taiwan, where they often reach several tens of meters in altitude. In the Ryûkyûs and in the main islands of the Japanese Archipelago, evidence of recent emergence is found along most of the coasts. Elevation increases towards the oceanic trenches, but former sea levels at above 6 m and even higher may also be recognized along the coasts of the Sea of Japan. In a few areas, such as in the Niigata Plain, marks of Holocene sea levels higher than at present are lacking. On the other hand, in other basins regarded as subsiding, such as those in the Nôbi and the Kantô plains, evidence of recent emergence is quite frequent. In many places, marks of several sea levels indicate that a step-by-step uplift has occurred. All the investigated insular arcs, therefore, seem to be situated in epeirogenic areas formed by several more or less large blocks affected by relative movements. The blocks are larger in the Outer Zone of Southwestern Japan; in the Inner Zone, an intricate network of fault lines marks the boundaries of many smaller independent blocks. During great earthquakes, relative movements of uplift, subsidence, tilting, or undulation occur in one or several blocks, depending on the position of the epicentres. Subsidence, however, must often be simply of a temporary nature, because a long-term uplift trend seems to prevail in most regions, even if it occurs at different rates. This interpretation may explain the cause of the great variety of ages and elevations of the former sea levels (with the oldest ages corresponding to the highest elevations) and the great number of indicators of step-by-step sea-level change. The inference, drawn by several authors, that the Holocene sea level in the Northwest Pacific was higher than at present, is, therefore, reasonable on a local scale, but does not define an eustatic sea level.

It has been suggested that the Laurentide Ice Sheet originated with extensive perennial snow cover, and that the snow cover affected climate so as to aid ice-sheet development. In this study, a large increase in extent of October 1st snow cover in the Canadian Arctic from 1967–1970 to 1971–1975 is compared to changes in October means of other climate variables. Over the area of snow-cover expansion, mean surface air temperature decreased by up to 3°C, mean 500-mbar height was lowered by over 60 m, and precipitation was increased by up to a factor of two. These effects, if applied to the entire summer, together with the temperature change computed by Shaw and Donn for a Northern Hemisphere summer insolation minimum (the Milankovich effect), can account for glacierization of the Central Canadian Arctic.

Paleoecological and paleogeographical evidence is used to mold a framework from which the basic parameters of the late Quaternary glacial-age climate of tropical Australasia can be inferred. The theory of physical circulations, a knowledge of present tropical circulation patterns, and a study of anomalous and extreme events in the present era are used to reemphasize the view of a less pluvial tropical and subtropical zone at that time. Cooler sea-surface temperature, cooler trades, and the effect of the then exposed land areas are indicated as instrumental in producing drier conditions. Tropical areas west of Cape York Peninsula and Torres Strait were subject to fewer tropical disturbances and were similar to the present tropical savannah of the northern interior of Australia. Such effects would exist even without shifts in major climatic zones, although they are shown to be consistent with an equatorward shift of the westerlies brought about by the increased pole to equator temperature gradient. Paleoenvironmental evidence from the New Guinea Highlands and southeastern Australia suggests that their climates were anomalous. Substantial data of the glacial period in New Guinea show snow lines to be 1000 to 1500 m lower than at present which matches a 6 to 8°C lowering of temperature in highland New Guinea. The deep-sea cores of the CLIMAP Project suggest a mere 2°C cooling of the surrounding tropical oceans. It is shown that it is highly unlikely that an upper-level decrease in temperature of 6 to 8°C could be maintained while the surface cools by only 2°C. It is suggested that either the temperature of the tropical oceans of the western Pacific were overestimated by CLIMAP or that cold air incursions from higher latitudes (for which some analogs exist today) were sufficiently frequent to allow the maintenance of a snow line well below the freezing level of the ancient ambient tropical atmosphere. It is shown that in southeastern Australia considerable evidence of aridity cannot be explained by merely displacing the westerlies more equatorward. To account for the aridity, a new circulation pattern is proposed. Noting that there is CLIMAP evidence of preferred equatorward extension of sea ice, a pattern is postulated that displays only small seasonal change and is characterized by an enhanced Indian Ocean trough, marked ridging at eastern Australian longitudes, and a further trough in the western Tasman. Such a basic flow is consistent with (i) a low rainfall over southeastern Australia, (ii) frequent cold outbreak conditions favorable for the maintenance of the New Guinea glaciers, and (iii) considerable precipitation to nourish the ice caps of Tasmania and the Australian and New Zeland Alps.

The Gaap Escarpment is a dolomite cuesta demarcating the southeast margin of the Kalahari. Since Miocene-Pliocene times, thick masses of lime tufa have repeatedly accumulated at several points along this escarpment, and four regional sequences are described. These allow discrimination of six major depositional complexes, commonly characterized by basal cryoclastic breccias or coarse conglomerates that reflect frost shattering and torrential runoff, followed by sheets and lobes of tufa generated in an environment substantially wetter than today. A chronostratigraphy for the last 30,000 yr is provided by 14C dating, with direct or indirect correlations to the Vaal River sequence. The regional stratigraphy as well as faunal dating indicate an early Pleistocene age for Australopithecus africanus at Taung. Repeated episodes of protracted cold or wetter climate or both begin in terminal Miocene times, and the last Pleistocene cold-moist interval began after 35,000 yr B.P. and ended 14,000 yr B.P. Early and late Holocene times were mainly wetter, whereas the middle Holocene was drier than today. The paleoclimatic sequence differs from that of the southern and southwestern Cape or that of East Africa, but close parallels are evident throughout the lower Vaal Basin and the southern Kalahari. The tufa cycles provide a unique, 5,000,000-yr record of climatic variation in the Kalahari summer-rainfall belt that can be related to complex anomalies of the general atmospheric circulation.

Marine ice sheets are grounded on land which was below sea level before it became depressed under the ice-sheet load. They are inherently unstable and, because of bedrock topography after depression, the collapse of a marine ice sheet may be very rapid. In this paper equations are derived that can be used to make a quantitative estimate of the maximum size of a marine ice sheet and of when and how rapidly retreat would take place under prescribed conditions. Ice-sheet growth is favored by falling sea level and uplift of the seabed. In most cases the buttressing effect of a partially grounded ice shelf is a prerequisite for maximum growth out to the edge of the continental shelf. Collapse is triggered most easily by eustatic rise in sea level, but it is possible that the ice sheet may self-destruct by depressing the edge of the continental shelf so that sea depth is increased at the equilibrium grounding line.

Application of the equations to a hypothetical “Ross Ice Sheet” that 18,000 yr ago may have covered the present-day Ross Ice Shelf indicates that, if the ice sheet existed, it probably extended to a line of sills parallel to the edge of the Ross Sea continental shelf. By allowing world sea level to rise from its late-Wisconsin minimum it was possible to calculate retreat rates for individual ice streams that drained the “Ross Ice Sheet.” For all the models tested, retreat began soon after sea level began to rise (∼15,000 yr B.P.). The first 100 km of retreat took between 1500 and 2500 yr but then retreat rates rapidly accelerated to between 0.5 and 25 km yr−1, depending on whether an ice shelf was present or not, with corresponding ice velocities across the grounding line of 4 to 70 km yr−1. All models indicate that most of the present-day Ross Ice Shelf was free of grounded ice by about 7000 yr B.P. As the ice streams retreated floating ice shelves may have formed between promontories of slowly collapsing stagnant ice left behind by the rapidly retreating ice streams. If ice shelves did not form during retreat then the analysis indicates that most of the West Antarctic Ice Sheet would have collapsed by 9000 yr B.P. Thus, the present-day Ross Ice Shelf (and probably the Ronne Ice Shelf) serves to stabilize the West Antarctic Ice Sheet, which would collapse very rapidly if the ice shelves were removed. This provides support for the suggestion that the 6-m sea-level high during the Sangamon Interglacial was caused by collapse of the West Antarctic Ice Sheet after climatic warming had sufficiently weakened the ice shelves. Since the West Antarctic Ice Sheet still exists it seems likely that ice shelves did form during Holocene retreat. Their effect was to slow and, finally, to halt retreat. The models that best fit available data require a rather low shear stress between the ice shelf and its sides, and this implies that rapid shear in this region encouraged the formation of a band of ice with a preferred crystal fabric, as appears to be happening today in the floating portions of fast bounded glaciers.

Rebound of the seabed after the ice sheet had retreated to an equilibrium position would allow the ice sheet to advance once more. This may be taking place today since analysis of data from the Ross Ice Shelf indicates that the southeast corner is probably growing thicker with time, and if this persists then large areas of ice shelf must become grounded. This would restrict drainage from West Antarctic ice streams which would tend to thicken and advance their grounding lines into the ice shelf.

Pumiceous tephra, resulting from multiple eruptions of Glacier Peak volcano in late-glacial time, mantles much of the landscape in the eastern North Cascade Range and extends eastward beyond the Columbia River as a thinner discontinuous deposit. Within about 25 km of the source, the tephra is divisible into as many as nine layers, distinguishable in the field on the basis of color, grain size, thickness, and stratigraphic position. Three principal layers, designated G (oldest), M, and B, are separated from one another by thinner, finer layers. Layer G has been found as far east as Montana and southern Alberta, whereas layer B has been identified as far as western Wyoming. By contrast, layer M trends nearly south, paralleling the crest of the Cascade Range. Available 14C dates indicate that the tephra complex was probably deposited between about 12,750 and 11,250 years ago. Glacier Peak tephra overlies moraines and associated outwash east of the Cascade Crest that were deposited about 14,000 years ago. Unreworked tephra occurs within several kilometers of many valley heads implying that major valley glaciers had nearly disappeared by the time of the initial tephra fall. Distribution of tephra indicates that the southern margin of the Cordilleran Ice Sheet had retreated at least 80 km north of its terminal moraine on the Waterville Plateau by the time layer G was deposited. Late-glacial moraines of the Rat Creek advance lie within the fallout area of layer M but lack the tephra on their surface implying that they were built subsequent to the eruption of this unit. Moraines of the Hyak advance at Snoqualmie Pass, which are correlated with the Rat Creek moraines farther north, were constructed prior to 11,000 14C years ago. The late-glacial advance along the Cascade Crest, therefore, apparently culminated between about 12,000 and 11,000 14C years ago and was broadly in phase with the Sumas readvance of the Cordilleran Ice Sheet in the Fraser Lowland which occurred between about 11,800 and 11,400 14C years ago.

Polygonal patterned ground and associated sediment-filled wedges occur in thin-bedded shale in central New Jersey. During a dry growing season, numerous areas of polygonal ground were visible owing to differential growth of vegetation over the sediment-filled fractures. Construction sites where material was removed from the surface prior to excavation also revealed areas of polygonal ground. Measurements of the patterns show networks of polygons with diameters ranging from about 3 m to over 30 m, with an average of approximately 20 m. The wedges examined in vertical exposures range in depth from 25 to 260 cm (average, 125 cm) and in width from 10 to 240 cm (average, 50 cm). The infilling material of the wedges is mostly a sandy loam, although this material is not homogeneous throughout the wedges. The presence of ventifacts and frosted sand grains within the wedges indicates eolian activity during the formation of these features. The polygonal ground and wedge structures exhibit several characteristics similar to those of ice-wedge casts, and available evidence suggests they may have originated as ice-wedge polygons during the last glacial maximum.

Pumice layers of set S from Mount St. Helens can be correlated with certain ash beds associated with young flood deposits of the channeled scabland. The correlation points to an age of about 13,000 14C yr B.P. for the last major flood to have crossed the scabland. Until recently, the last major episode of flooding was thought to be closer to 20,000 yr B.P., an age inferred chiefly from the relation of the flood to glacial events of the northern Rocky Mountains. Several investigations within the last few years have suggested that the last major flood occurred well after 20,000 yr B.P. Tentative correlations of ash beds of the scabland with set S pumice layers, the relations of flood and glacial events along the northwestern margin of the Columbia Plateau, and a radiocarbon date from the Snake River drainage southeast of the plateau all indicate an age much younger than 20,000 yr. The postulated age of about 13,000 yr B.P. is further supported by a radiocarbon date in the Columbia River valley downstream from the scabland tract. Basal peat from a bog on the Portland delta of Bretz, which is a downvalley deposit of the last major scabland flood, has been dated as 13,080 ± 300 yr B.P. (W-3404).

Lithologic, paleontologic, and chronostratigraphic investigation of 13 gravity cores indicates the following environmental evolution: a high- (mid-) arctic period with a slight influx of ice-rafted debris occurred during the early middle Weichselian followed by a mid- (high-) arctic environment with a high influx of iceberg-rafted debris during the remainder of the middle Weichselian. The continental ice sheet probably did not extend beyond the inner shelf during middle Weichselian and a minimum relative sea level was ca. −120 m. A low-arctic environment occurred during (parts of) the late Weichselian with an initial winnowing of the sediments. The Norwegian Current entered the area during this substage. A high- (mid-) boreal environment occurred during the Holocene with high winnowing activity in the early Holocene. Winnowing is still very active on the shallower banks in contrast to the deeper banks where it has ceased. Relatively high percentages of carbonate in the form of biogenic skeletal remains occur in the Holocene sediments.

A 2000-year accumulation of varved sediments from Hell's Kitchen Lake in north-central Wisconsin was analyzed for pollen, charcoal, and seeds. The varves provided an accurate time scale for the study. The pollen record indicates changes on two different time scales. Short-term changes lasting several decades appear to be superimposed on long-term changes lasting several centuries. The short-term changes are related to individual fires, and the long-term changes result from increases or decreases in the frequency of these perturbations. From 2000 to 1150 years ago the average interval between fires was about 100 years, and from 1150 to 120 years ago the interval increased to about 140 years. Evidence from pollen, seeds, and charcoal at Hell's Kitchen Lake suggests that at least two “moist” intervals occurred during the past 2000 years, one between 2000 and 1700 years ago and the other between 600 and 100 years ago. A third but minor “moist” period occurred about 1150 to 850 years ago. A pollen and seed diagram shows that these intervals are characterized by increased percentages of white pine pollen, hemlock pollen, and yellow birch seeds, and by decreased levels of charcoal. The “dry” interval of 1700 to 1150 years ago is characterized by increased percentages of paper birch seeds, oak pollen, and aspen pollen, along with high levels of charcoal. The times of climatic change indicated at Hell's Kitchen Lake are nearly synchronous with those based on studies of tree rings, soils, glacial activity, and other pollen studies from various regions of North America, but the direction of these inferred changes is not always the same. This result suggests that the long-wave pattern of the general circulation has been variable during the past 2000 years.

The reef-crest coral Acropora palmata from late Pleistocene reefs on Barbados has recorded the same global variations in oxygen isotopes as planktonic and benthonic foraminifera. Although the record of oxygen isotopes in Acropora palmata is discontinuous, it offers several advantages over the isotope records from deep-sea sediments: (1) the coral grows at water depths of less than 5 m; (2) the samples are unmixed; (3) specimens may be sampled from various elevations of paleo-sea level; and (4) aragonitic corals are suitable for 230Th/234U and He/U dating techniques. The latter advantage means that direct dating of the marine oxygen isotope record is possible. Oxygen isotope stage 5e corresponds to Barbados III, dated at 125,000 ± 6000 yr BP. Petrographic and geochemical evidence from five boreholes drilled into the south coast of Barbados indicates a major eustatic lowering (greater than 100 m below present sea level) occurred between 180,000 and 125,000 yr BP. The age and isotopic data suggest correlation of this change in sea level to Emiliani's oxygen isotope stage 6. Acropora palmata deposited at various elevations of sea level during oxygen isotope stage 6 vary by 0.11 ‰ δ18O for each 10 m of change in sea level. We further hypothesize a minimum drop of 2°C in the average temperature occurred during the regressive phase of oxygen isotope stage 6. These data indicate that temperature lowering of surface water near Barbados lagged behind a major glacial buildup during this time period. Using the δ18O vs sea level calibration herein derived, we estimate the relative height of sea stands responsible for Barbados coral reef terraces in the time range 80,000 to 220,000 yr BP.

The varved sediments of the Santa Barbara Basin off southern California, offer a unique opportunity to study the changes in oceanographic conditions of this nearshore area during the last 8000 yr. Quantitative analysis of Radiolaria found in recent surface sediment samples from the eastern North Pacific allows the identification of four “assemblages” which can be related to the physical oceanography of the California Current. Two assemblages are associated with the southerly flowing California Current, one with the main stream of the current (California Current Assemblage) and the other with the offshore flow along northern California (Central Assemblage). The two other assemblages are associated with the subtropical region of the eastern North Pacific (Subtropical Assemblage) and one found mostly off the coast of Baja California (Baja Assemblage). Analysis of the Radiolaria found in the varved sediments of a core from the Santa Barbara Basin give an 8000-yr continuous record of these four assemblages. The California and Baja Assemblages show only minor fluctuations in their importance in the sediments of the Santa Barbara Basin. The California Assemblage, however, shows a steady increase during this time period. Prior to 5400 yr B.P. the Radiolaria were predominately subtropical in character, whereas after 5400 yr B.P. the Central Assemblage becomes more important. Since 5400 yr B.P. most of the changes in the radiolarian fauna consist of fluctuations in the importance of these two assemblages. Past sea-surface temperatures for the month of February were calculated using the transfer function technique of J. Imbrie and N. G. Kipp (1971, In “The Late Cenozore Glacial Ages” L. K. Turekian, Ed.), (Chap. 5, Yale Univ. Press, New Haven, Conn.). The time series of paleotemperature estimates show major changes in the average February temperature of Santa Barbara Basin waters. The range of estimated temperatures (12°C) exceeds that of the historical observations of February temperatures in the Santa Barbara Basin but does not exceed the observed range for the California Current region. The intervals from 800 to 1800 yr B.P. 3600 to 3800 yr B.P. and 5400 to the end of the record appear to have been generally warmer than today. Comparison of the Holocene record of alpine glacial advances with the radiolarian assemblage and paleotemperature time series shows that the initiations of advances was coincident with a decrease in sea-surface temperatures and an increase in the importance of the Central Assemblage in the Santa Barbara Basin. The terminations of these advances were not marked by any consistent characteristic in the Santa Barbara Basin time series.

The Cutting 10 site at Elandsfontein (Hopefield), southwestern Cape Province, South Africa, is one of only three sites in southern Africa where bones and Acheulean artifacts have been documented together in primary or near primary context and the only such site where the association is in the open air. The Cutting 10 bones belong to a minimum of 15 mammalian species, which together suggest both a mid-Pleistocene age for the site and a vegetational setting in which grasses played a much more important role than they did historically. The absence of evidence for hearths, windbreaks, and other features, the composition of the artifact assemblage, the relatively sparse scatter of bones and artifacts, the relatively high bone to artifact ratio, the species composition of the fauna, and the numbers of different kinds of skeletal elements represented all indicate that the Cutting 10 occurrence was more a butchering site than a campsite. Although people are probably responsible for most of the bones, some of them probably come from animals that were killed by large carnivores or that died naturally.

The high-level Quaternary shorelines of the Mediterranean are now in a fragmentary condition, but on some parts of the coast it has been possible to reconstruct one or more over a distance of many kilometers. A few of these reconstructed segments of shoreline are tilted or otherwise deformed, but the great majority appear to be virtually horizontal. The horizontal segments show no clear preference for any particular level or levels, from which it is concluded that all have been vertically displaced by earth movements.

Camelops have been recorded in a number of Paleo-Indian sites that lack evidence of past procurement methods. Recently, two occurrences of Camelops remains have been recorded in Paleo-Indian animal kills in Wyoming. One kill situation was in a Hell Gap cultural context that produced remains of a single Camelops taken along with about 100 bison in a parabolic sand dune trap. The other Camelops was in a Clovis cultural context and deals with a single animal believed to have been taken in an arroyo trap that was used to take bison at several Paleo-Indian time periods. Identification of geomorphic features involved in these kill sites offers a basis for beginning to interpret Paleo-Indian camel procurement methods.

A core section from the Fuquene area (Cordillera Oriental, Colombia) with a total length of 60 m shows a predominantly peaty-clay zone in the upper 12 m containing 15 laminae and thin layers of fine-grained, pyroclastic material. These 15 layers consist of crystal-vitric ash derived from dacitic-andesitic magmas. Mineralogic data and refractive index (r.i.) values of the associated glass shards make it possible to distinguish a lower group of mica-rich and an upper group of pyroxene-bearing ash layers. The successive layers illustrate a progressive increase of hydration of the shards with time and with decreasing particle size. The frequency distribution curves of the r.i. values of 8–4- and 75–50-μm shards are in general rather similar. Provisional pollen data and some radiocarbon dates allow comparison of the petrographic properties of these ashes with those of some ash occurrences in a nearby palynologically investigated core section. This comparison indicates that the mica-rich ashes were produced between about 44,000 and 20,000 yr B.P. and that the production of the pyroxene-bearing ashes took place after about 20,000 yr B.P. The petrographic features of a number of ashes, collected from sections and cores at different sites of the Sabana de Bogotá, appear to be comparable with those established in the sequence of Fuquene ashes. This suggests that these petrographic characteristics may be of use in correlation studies in the Sabana de Bogotá and adjacent areas.

Pygmy proboscidean remains of Mammuthus exilis occur abundantly in late Quaternary deposits on the Northern Channel Islands, California. On the assumption that ancestral elephants could not have swum to the islands and must therefore have walked out, various land bridges have been hypothesized that link the northern islands to the mainland by a peninsula. Geological evidence for a land bridge, however, is lacking, and new evidence shows that elephants are excellent swimmers and skilled at crossing watergaps. The Santa Barbara Channel was narrowed to only 6 km during glacially lowered sea levels. Modern elephants swim much further, and at speeds ranging from 0.96–2.70 km/hr. Motives for California elephants to cross Pleistocene watergaps are inferred from motives that lead modern elephants in Asia and Africa to cross watergaps. These are the visual and olfactory sensing of islands and of insular food during times of drought or fire-induced food shortage. Diminutive size of M. exilis principally reflects lack of island predators, an adaption to periodic food stress in a finite forage area affected by periodic drought and fire, and an adaptation for keeping population numbers high to maintain genetic variability and to ensure survival despite accidents. A late Quaternary scenario describes the environmental setting of the Santa Barbara Channel and the conditions that led to proboscidean dispersal to the preexistent super-island Santarosae.

The dating of the Stone Age sequence in southern Africa has been considerably revised over the last decade, and one of the anomalies which has resulted is that the Middle Stone Age, now dated to beyond 30,000 B.P., does not immediately precede the Later Stone Age sensu stricto. The excavation and analysis of occupation horizons dating between the most recent Middle Stone Age assemblages and the Holocene is therefore of particular interest. Nelson Bay Cave, situated on the southern coast of South Africa, contains deposits which partly fill the “gap” between the Middle and Later Stone Ages, and the occupation horizons dating between about 18,000 and 5000 years ago are described in this paper. Changes in the habitat in the vicinity of the site caused by sea-level and vegetation changes coincident with the amelioration of temperatures at the end of the Pleistocene are clearly marked in the faunal remains at the site. Largely correlated with the faunal changes (which includes the introduction of marine resources to the cave at about 12,000 B.P.) are changes in the stone artifact assemblages. Three industries are recognized in the sequence: the Robberg, characterized by microbladelets produced from bladelet cores and a few small scrapers and backed tools; the Albany, characterized by large scrapers and an absence of backed tools; and the Wilton, characterized by a variety of Formal Tools including relatively large numbers of small scrapers and backed tools. These changes in artifact-manufacturing traditions are interpreted as signaling adjustments to changing environmental conditions. An explanation for these adjustments is not sought in a simple cause-and-effect relationship between the environment and the cultural response; artifact changes are seen instead as the result of a twofold process, with the environment acting as an external stimulus to change, and the direction of the artifact change governed by the selection of a range of possibilities offered by the technology of the Later Stone Age sensu lato that was widespread in subequatorial Africa during the last 20,000 years.

Pollen analysis of a section of lake sediments at Grassy Lake Reservoir indicates a vegetational sequence changing from tundra, to spruce-fir-pine forest, to pine forest, to tundra at the top. Pollen analysis of a section of lake sediments on Beaverdam Creek indicates a tundra vegetation at the base, followed by a brief episode of spruce-fir forest and a return to a tundra vegetation at the top. The analyses of both sections suggest a cold to cool to cold climatic sequence, interpreted as interstadial in character. However, differences suggest that they represent separate interstadials. Pinedale Till disconformably overlies the lake deposits at Grassy Lake Reservoir. The upper sediments contain wood 14C dated at >42,000 yr; the lowermost interfinger with till shown to be more than about 70,000 yr old. The deposits at Beaverdam Creek grade upward into proglacial Pinedale deposits, contain an ash that is probably about 70,000 yr old near their base, and rest comformably on gravel that grades down into lake sediments containing wood debris suggestive of an older climatic amelioration. We conclude that the warmest part of the interstadial at Grassy Lake Reservoir is probably more than 70,000 yr old, and that the warmest part of the interstadial analyzed at Beaverdam Creek is slightly younger than 70,000 yr old.