A general feedback between volcanism and climate at times of transition in the Quaternary climate record is suggested, exemplified by events accompanying the Toba eruption (∼74,000 yr ago), the largest known late Quaternary explosive volcanic eruption. The Toba paroxysm occurred during the δ18O stage 5a-4 transition, a period of rapid ice growth and falling global sea level, which may have been a factor in creating stresses that triggered the volcanic event. Toba is estimated to have produced between 1015 and 1016 g of fine ash and sulfur gases lofted in co-ignimbrite ash clouds to heights of at least 32 ± 5 km, which may have led to dense stratospheric dust and sulfuric acid aerosol clouds. These conditions could have created a brief, dramatic cooling or "volcanic winter," followed by estimated annual Northern Hemisphere surface-temperature decreases of ∼3° to 5°C caused by the longer-lived aerosols. Summer temperature decreases of ⩾10°C at high northern latitudes, adjacent to regions already covered by snow and ice, might have increased snow cover and sea-ice extent, accelerating the global cooling already in progress. Evidence for such climate-volcanic feedback, following Milankovitch periodicities, is found at several climatic transitions.

Chronological correlations established at different time scales among the lake-level fluctuations in the Jura and French Subalpine ranges, glacier movements in the Swiss and Austrian Alps, and the atmospheric 14C record during the last 7 millennia show coincidences between lake-level rises, glacier advances, and high 14C production and vice versa. These correspondences suggest that the short-term 14C variations may be an empirical indicator of Holocene palaeoclimates and argue for possible origins of Holocene climatic oscillations: (1) The varying solar activity refers to secular climatic oscillations and to major climatic deteriorations showing a ca. 2300-yr periodicity. (2) A question is raised about a relationship between the earth's magnetic field and climate. First, the weak-strength periods of the earth's dipole magnetic field (between 3800 and ca. 2500 B.C.) coincide with higher climate variability, and vice versa. Second, the ca. 2300-yr periods revealed by the 14C record and also by the major climatic deteriorations re. corded in Jurassian lakes (ca. 1500 A.D., ca. 800 B.C., and ca. 3500 B.C.) coincide with the ca. 2300-yr periods revealed by the earth's nondipole geomagnetic field. The present warming induced by anthropogenic factors should be intensified during the next few centuries by natural factors of climate evolution.

Lake-level data can be used to refine palaeoclimate reconstructions based on pollen data. This approach is illustrated for the European Holocene. Estimates of P-PET (precipitation minus potential evapotranspiration) were first inferred from modern pollen analogues. The pollen-based estimates were then compared with the status of lakes within a 5° radius. Analogues with P-PET anomalies inconsistent with the lake-level changes were rejected. The "constrained" sets of analogues were used to estimate continental-scale patterns of annual mean temperature and annual precipitation at 3000-yr intervals. Estimated temperature anomalies differed only slightly from the unconstrained reconstructions. Estimated precipitation anomalies, however, showed improved spatial coherence and increased regional contrast and were occasionally reversed in sign. The effect of the constraint was to impose a rational selection among almost equally similar modern pollen analogues with similar temperatures but widely varying moisture regimes. The resulting maps showed clear, spatially coherent patterns of change in precipitation as well as temperature, suitable for comparison with climate-model results. Further improvement of these maps will become possible as a more extensive coverage of lake-level data is obtained.

In the Río Mendoza valley, five Pleistocene drifts and one Holocene drift are distinguished by multiple relative-age criteria, including surface-rock weathering, development of rock varnish, moraine morphology, soil-profile development, and stratigraphic relationships. Several absolute ages suggest a preliminary chronology. During the oldest (Uspallata) glaciation, a system of valley glaciers flowed 110 km from the Andean drainage divide and 80 km from Cerro Aconcagua to terminate at 1850 m. Drift of this ice advance is older than a widespread tephra dated by fission-track at 360,000 ± 36,000 yr. During the Punta de Vacas advance, ice terminated at 2350 m, while during the subsequent Penitentes advance, the glacier system ended at 2500 m. A travertine layer overlying Penitentes Drift has U-series age of 24,200 ± 2000 yr B.P. The distribution of Horcones Drift, which is inferred to represent the last glacial maximum, delimits an independent ice stream that flowed 22 km down Horcones valley to 2750 m. A later readvance (Almacenes) reached 3250 m. Confluencia Drift is considered to be Neoglacial in age and extends downvalley to 3300 m. The moraine sequence is compared with those studied by Caviedes (1972) along Río Aconcagua on the Chilean flank of the Andes.

The Toba ash occurs extensively in the Indian subcontinent and marks a ca. 74,000-yr-old event. In the Bay of Bengal and Indian Ocean it is about 10 cm thick, whereas in several alluvial basins, it is usually 1-3 m thick. The latter occurs in a partly reworked state but as nearly chemically pure first-cycle sediments, The ash has a broad northwesterly dispersal pattern. Samples of ash from the Indian subcontinent compare closely with the youngest (74,000 yr B.P.) Toba Tuff and the deep-sea Toba ash in bulk chemical composition, REE signature, and bubble-wall shard morphology. However, a more proximally located and thicker (2-5 m) ash-bed, from the alluvial basins in the gneissic area and close to east coast, has a lower magnitude negative Eu anomaly, possibly because of minor contamination by feldspathic silt. Quaternary sediments in the central Narmada and middle Son basins contain rich late and middle Pleistocene mammalian and cultural records. Based on the presence of the ash layer marker and stratigraphic relations, late Pleistocene sediments within the subcontinent can be correlated with those from central India and the deep sea.

A simple linear response model describing the functional relationship between ocean carbon cycle parameters and paleoclimate tracers (atmospheric pCO2, δ13C, CaCO3 saturation) was derived from a set of sensitivity experiments performed previously using a three-dimensional carbon cycle model. The linear model is optimally fitted to ice and marine sediment core records for the last 120,000 yr to estimate the carbon cycle parameter changes that could have caused the observed reduction of atmospheric CO2 partial pressure during the last glaciation. The analysis indicates that the glacial pCO2 reduction was primarily caused by a strengthening of the biological POC pump and a retardation of the oceanic circulation. An increase in deep-sea alkalinity and a change in the advective pattern of the ocean circulation have a smaller impact on atmospheric CO2 but are necessary to explain the full set of paleoclimate tracers.

Multiple tills separated by interbedded lake sediments were temporarily exposed during open-pit mining at Sanford Hill, near Newcomb, in the central Adirondack Mountains of northeastern New York. Radiocarbon ages of wood fragments from brown clay between two tills at that site indicate an age older than 55,000 yr B.P. A pollen profile in the brown pond clay (Tahawus clay) records a transition from initial domination by pine, spruce, and birch to an oak pollen zone. The upward disappearance of spruce and decrease of pine and birch are accompanied by diversification and increase in hardwood pollen. This assemblage includes several warm-climate hardwood species that are not represented in the postglacial flora of the region, suggesting that the biota existed during an interglacial or interstadial interval that was warmer and/or longer than postglacial time. Accordingly, the Tahawus clay is tentatively assigned to the Sangamon interglaciation, and probably to marine oxygen isotope substage 5e. Truncation of the Tahawus clay by glacial erosion may account for the absence of a Sangamon paleosol.

Low, nearly continuous terraces of similar age are present along streams in drainage basins that range in size from Drift Creek (190 km2) to the Umpqua River (11,800 km2) in the Oregon Coast Range. Radiocarbon ages from near the bose of fluvial sediments underlying these terraces are clustered at about 9000-11,000 14C yr B.P. Beveled bedrock surfaces (straths) that underlie the fluvial sediments are 1-8 m above summer stream levels and are present along most of the nontidal reaches of the rivers that we studied. Where exposed, the bedrock straths are overlain by 2-11 m of fluvial sediment that consists of a bottom-stratum (channel) facies of sandy pebble-cobble gravel and a top-stratum (overbank) facies of sandy silt or silt. Eight radiocarbon ages from the fluvial sediments allow correlation of the lowest continuous terrace over a wide area and thus indicate that a regional aggradation episode occurred in Coast Range drainage basins during the Pleistocene-Holocene transition. The cause of such widespread aggradation is unknown but may be related to climate-induced changes in the frequency of evacuation of colluvium from hollows, which are common in all drainage basins in the region.

A thick deposit of Old Crow tephra was discovered in a bluff exposure along the middle Holitna River near the Kulukbuk Hills (61°20′N latitude, 157°10′W longitude) in interior southwest Alaska. This locality is the southwesternmost-known deposit of Old Crow tephra in Alaska. Thickness and grain-size data from this site support a source volcano in the eastern Aleutian arc. Pleistocene stratigraphic sequences in the lowland are dominated by upward-fining eolian sand-sheet deposits and loess separated by organic silt. These deposits record at least two episodes of regional glaciation and an intervening nonglacial period (marine oxygen isotope stage 3, stage 5, or both). Old Crow tephra crops out near the top of the lower upward-fining eolian unit, indicating that the ash erupted near the end of an interval of periglacial eolian sedimentation. The sequence of eolian deposits that contain Old Crow tephra probably accumulated during the latter part of marine oxygen isotope stage 6, whereas the overlying eolian sequence formed during the last glaciation (stage 2). This stratigraphic position is consistent with other stratigraphic contexts for the tephra and with fission-track and thermoluminescence ages of ca. 140,000 ± 10,000 yr B.P.

The simultaneous operation of regressive and progressive pedogenetic pathways raises the possibility that soil evolution may exhibit patterns far richer and more complex than increasing development over time. This possibility is explored via a numerical model incorporating the relative rates of progressive and regressive pedogenesis and feedbacks between these rates and the degree of soil development. This model may exhibit deterministic chaos and sensitive dependence on initial conditions with realistic parameter values. Variations in profile development in a region of the North Carolina Coastal Plain where soil-forming factors are relatively constant is consistent with deterministic chaos. Chaotic soil evolution suggests that soil development may reflect the interplay between progressive and regressive soil-forming processes in addition to—or instead of—the age of a surface or deposit. The former may produce a state of development which is unique to a particular time, sensitively dependent on the (unknown) initial conditions, and not simply related to age. Soils and their genetic signatures as indicators of relative ages and Quaternary environments may need to be reinterpreted in this light.

Rock varnish coatings tend to become thicker, darker, and more continuous over time, leading to the use of changes in overall varnish color and the percentage of clast surfaces covered by varnish as relative-age indicators. Manganese is the most characteristic element of subaerial rock varnishes, and the progressive development of varnish coats can be quantified by measuring the amount of Mn accumulated on a given area of rock surface. Manganese oxides were dissolved off varnished clasts collected from alluvial surfaces on the Soda Mountains piedmont in the Mojave Desert, California, and the amount of Mn was measured using inductively coupled plasma emission spectroscopy. On the distal piedmont, maximum varnish development increases from a mid- to late-Holocene surface, typically containing up to 0.15 mg/cm2 of accumulated Mn, to an early- to mid-Holocene surface with up to 0.21 mg/cm2. However, varnish is less developed on a nearby late Pleistocene surface, suggesting extensive abrasion of clasts on the Pleistocene desert pavements or disturbance of the clasts. Varnish is better developed on the proximal piedmont, typically containing up to 0.30 mg/cm2 of Mn, although varnish from a Pleistocene surface is again no better developed than from a nearby early- to mid-Holocene surface. These data demonstrate that rock varnish can show significant spatial variation in degree of development on geomorphic surfaces of similar age, and imply that collecting varnish as old as a geomorphic surface may be difficult on surfaces as young as late Pleistocene.

The most viable sources for ubiquitous loess deposits on the eastern Snake River Plain, Idaho are aggraded river valleys, active alluvial fans, and fluctuating pluvial lake margins associated with regional late Pleistocene glaciation of the northern Rocky Mountains. Stratigraphic studies on the Idaho National Engineering Laboratory document two distinct loess deposits, separated by a well-developed paleosol, resting on basaltic lava. Baked sediments beneath this lava yielded thermoluminescence (TL) age estimates of 108,000 ± 13,000 and 101,000 ± 7000 yr, and baked organic matter gave radiocarbon ages of >32,000 yr B.P., consistent with an earlier K/Ar age for the flow of 95,000 ± 25,000 yr. The overlying two loess deposits yielded TL age estimates of 74,000 ± 6000 and 28,000 ± 3000 yr. The available geochronology indicates that the latest period of loess deposition commenced ca. 40,000 to 35,000 yr ago and ceased approximately 10,000 yr ago, which is generally coincident with the inferred timing of regional Pinedale glaciation and pluvial lake expansion. We estimate that the penultimate loess depositional episode dates between 80,000 and 60,000 yr ago, which is significantly younger than previous age estimates of 140,000 to 150,000 yr based on stratigraphic position. We speculate that this period of loess deposition may correlate with documented early to middle Wisconsinan glaciation and a high stand of pluvial lakes in the Basin and Range province.

As many as six levels of emerged Holocene coral terraces occur along 40 km of coastline on the Huon Peninsula, Papua New Guinea, recording uplift history since culmination of the postglacial transgression. The Holocene reef crest, ca. 6000 yr B.P., is tilted down to the northwest, parallel to the coast and concordant with the deformation of the last interglacial coral reef terrace, and descends from 23 to 12 m in the study area. The pattern and rate of deformation have been uniform in the late Quaternary because average uplift rates have remained the same since the last interglaciation. The Holocene terraces described here are erosional features with regressive encrusting corals, developed upon the Holocene transgressive reef. The multiple levels represent episodic, probably coseismic uplift, which has occurred repeatedly in the last ca. 6000 yr. Significant longshore variation in the age of the lowest terrace, from 1700 to 2500 yr B.P., suggests independent coseismic uplift on different sectors of the coast. This is supported by age-height relationships of the higher Holocene terraces. Nonlinear uplift during the Holocene, with recurrence intervals increasing toward the present, is clearly recorded by the regressive terraces in each subregion. Some of the Holocene regressive terraces grade laterally into fluvial terraces capped with debris-flow deposits, probably reflecting seismically triggered mass movement.

Cave entrance and rock shelter infillings are positioned within the Pleistocene chronology for three areas of France (northern Alps, Franche-Comté, and Périgord). Despite minor local variations, it is possible to identify regional types with a consistent depositional record over long intervals of time. The interregional variability relates to the frequency and position of the gaps within the infillings. Sites in the northern Alps have not yielded any artifacts older than the Upper Paleolithic (Magdalenı́an), and dated sedimentary sequences do not go back beyond the Older Dryas. More complete sequences in Franche-Comté contain Mousterian industries. Two major gaps occur here, one lasting about 50,000 yr during the Lower Glacial and Lower Pleniglacial periods, and the other lasting some 10,000 yr during the Upper Pleniglacial. The sequences are more complete in the Périgord, where all major climatic phases have been identified with no significant sedimentary gaps. Differences in sedimentary records can be explained by variations in regional conditions during the last interglacial-glacial cycle and especially by the distance from the maximum advance of ice daring glaciation.

New radiocarbon dates provide ages for two submerged strandline features on the Massachusetts inner shelf. These ages provide limited control on a relative sea-level (RSL) curve for the late Wisconsinan and Holocene. The curve indicates a late Wisconsinan high stand of RSL of +33 m about 14,000 yr ago and a very short-lived relative low stand of about -43 m at about 12,000 yr ago followed by a rise to present sea level. Rapid changes of RSL around 12,000 yr ago may be related to changes in global glacial meltwater discharge and eustatic sea-level change shown by dated corals off Barbados. Variations in the magnitude and timing of RSL change from south to north along the coast of the western Gulf of Maine are due to greater crustal depression and later deglaciation to the north.

Holocene emergent reefs and notches are well distributed on Rota and Guam. Relative sea-level changes at these islands are reconstructed based on geomorphological observations and borings on present and emergent reefs, together with 54 radiocarbon dates. Sea level rose gradually to a maximum of 1.8 m between 6000 and 4200 yr B.P. and reached its highest level by 4200 yr B.P. on both islands. After 3200 yr B.P. abrupt uplift caused emergence of the reef. By subtracting the tectonic effect, we obtained the sea-level change in the Marianas: sea level reached its present level by 4200 yr B.P. and has remained almost stable since then. Reconstructed late Holocene sea-level change in the Mariana Islands provides constraints on geophysical models of sea-level variations.

In order to establish paleoenvironmental conditions during the late Quaternary, four cores from the Basin of Mexico (central Mexico) were drilled in Chalco Lake, located in the southeastern part of the basin. The upper 8 m of two parallel cores were studied, using paleomagnetic, loss-on-ignition, pollen, and diatom analyses. Based on 11 14C ages, the analyzed record spans the last 19,000 14C yr B.P. Volcanic activity has affected microfossil abundances, both directly and indirectly, resulting in absence or reduction of pollen and diatom assemblages. Important volcanic activity took place between 19,000 and 15,000 yr B.P. when the lake was a shallow alkaline marsh and an increase of grassland pollen suggests a dry, cold climate. During this interval, abrupt environmental changes with increasing moisture occurred. From 15,000 until 12,500 yr B.P. the lake level increased and the pollen indicates wetter conditions. The highest lake level is registered from 12,500 to ca. 9000 yr B.P. The end of the Pleistocene is characterized by an increase in humidity. From 9000 until ca. 3000 yr B.P. Chalco Lake was a saline marsh and the pollen record indicates warmer conditions. After 3000 yr B.P. the lake level increased and human disturbance dominates the lacustrine record.

The postglacial uplift pattern indicated by elevations of ice-marginal glaciomarine deltas in coastal New England, deposited between approximately 15,000 and 14,000 yr B.P. during ice retreat from northeastern Massachusetts into southwestern Maine, is very similar to that previously recorded for glaciolacustrine deltas of similar age from inland areas of New England. Multiple regression analyses of elevations from both sets of deltas show an extremely close fit to tilted flat surfaces that rise 0.852 m/km to the N 28.5°W along the coast and 0.889 m/km to the N 20.5°W in western New England. The close similarity of uplift pattern in areas where elevation data are from different base-level media, along with additional shore-line evidence, indicates (1) that both areas are part of the same crustal postglacial uplift block, (2) that postglacial uplift was delayed until after 14,000 yr B.P., and (3) that little or no eustatic sea-level change occurred between 15,000 and 14,000 yr B.P., during which time the margin of the late Wisconsinan Laurentide ice sheet retreated about 100 km from Boston, Massachusetts, into southwestern Maine. Elevation data from even younger glaciomarine deltas in the coastal area indicate that soon after the ice margin reached southwestern Maine and adjacent New Hampshire (ca, 14,000 yr B.P.), eustatic sea level rose rapidly 7-10 m during the time that the ice margin retreated 5-10 km, which may have occurred during an interval of only 50-100 yr, Our new data not only confirm the delayed postglacial uplift model previously described for western New England, but also indicate that little or no eustatic sea-level change occurred during a substantial period of early deglaciation. However, at about 14,000 yr B.P., sea level rose rapidly. Postglacial uplift in the region apparently began between 14,000 and 13,300 yr B.P., before the retreating ice margin reached eastern Maine.

A progradational regime of falling sea level and/or high sediment input has produced extensive beach ridge plains in northwest Alaska during the last 4000 yr. Eleven Chukchi Sea beach ridge complexes, oriented at various angles to wind fetch, provide a cumulative history of longshore transport and erosion. Archaeological and geological upper limiting radiocarbon ages (n = 59) allow correlations between depositional units on seven beach ridge complexes. Progradation started 4000 yr B.P. at nearly all complexes, as eustatic sea level stabilized. Two disconformities or truncations are found on most of the complexes, providing time-parallel storm horizons, dated at 3300-1700 and 1200-900 14C yr B.P. Between 1700 and 1200 14C yr B.P. most of the complexes prograded, indicating the predominance of less-stormy conditions. Modern synoptic patterns that produce Chukchi beach ridge erosion are linked to northerly shifts in North Pacific storm tracks. The regionwide beach ridge erosional truncations correlate with records of glacier expansion, heightened precipitation evident in tree-rings, stream flooding, and shelf deposits reworked by storm surges.

Palaeoclimatic estimates of mean annual temperature and rainfall in the southern Tanganyika basin between 25,000 and 9000 yr B.P. have been established from two pollen sequences based on the best-analogue method. The results give evidence of a mean temperature decrease of about 4.2°C during the last glaciation, a value consistent with that previously obtained in the catchment area on the Burundi Highlands. This cooling was synchronous with a decrease of mean annual precipitation of about 180 mm/yr. Postglacial climatic conditions were established by 12,700 yr B.P., with warming and wetness continuing to increase from this date onward. These new palaeoclimatic data will be useful for hydrological reconstructions of Lake Tanganyika, particularly during the last glacial age for which the magnitude of water-level fall has been a controversial issue; our rainfall estimates are more consistent with low values (-250 to -300 m fall) than with high ones (-600 m) previously proposed.