The marine record shows that over the last 350 ka Northern Hemisphere ice sheet volumes have fluctuated widely and only on rare short occasions have they been reduced to the present interglacial state. The fluctuations are well synchronized with hemispheric average summer insolation variations of 20 ka periodicity caused by changing orbital parameters. The development of a model which explains the varied amplitudes of the fluctuations and is consistent with the geological record embodies the following arguments: The transition from an interglacial state like today's to a glacial state is initiated when a summer insolation deficit causes a southerly extension of the North Atlantic-Arctic pack ice to 60°N latitude. The extension alters the subpolar low pressure patterns and thus causes a southward diversion of the European Gulf Stream flow. It also produces an enhanced warm West Greenland current. This current causes open seas as far north as Baffin Bay which provides moisture for rapid northern Laurentide ice sheet growth. After several glacial fluctuations driven by insolation variations, the southern Laurentide ice front may reach an extreme extension. This diverts the westerlies and the Gulf Stream thus weakening a dominant subpolar North Atlantic gyre and consequently producing a prolonged cutoff of the West Greenland current and a reduction of high latitude glacial precipitation. The subsequent high insolation can then melt back the eastern pack ice and restore the northern European Gulf Stream. This warms the high latitudes for a time sufficient to melt the continental ice, thus causing the transition back to the interglacial state.

An analysis of the record in the context of model suggests that the threshold deficit in average summer insolation that is required to initiate major glacial growth is influenced by the cooling effect of the Greenland ice cap on the seas to the east. The threshold level under conditions like today's is found to lie between −7 and −17 ly/day relative to the present. This threshold will not be crossed for at least 54 millenia due to an interval of smaller orbital eccentricity. Probable melting of the Greenland ice cap about 30 ka AP would ensure the extension of the present interglacial beyond 120 ka AP.

In southernmost South America, an incomplete radiometrically dated glacial chronology has been obtained by K–Ar dating for the interval 3.5-1 MY ago, and a more detailed chronology by C-14 dating for the last 25,000 years, with some older minimal ages. The first major glaciation was about 3.5 MY ago during the middle Pliocene. Little is yet known about glacial fluctuations during the interval 3.5-2.1 MY ago. Between 2.1 and 1 MY ago many glaciations occurred, probably including the greatest of late Cenozoic time which took place after 1.2 MY and, according to inconclusive evidence, before 1 MY ago. The Patagonian Gravel in its type area is mid-Pliocene to early Pleistocene glacial outwash that accumulated from the first to the greatest glaciations. During the late Pleistocene several glaciations occurred, but only the most recent has been radiometrically dated. During the last glaciation the glaciers were most extensive before 56,000 BP. Successively smaller advances that culminated about 19,500 BP and, probably, about 13,000 BP were separated by an interstade when glaciers shrank by more than half. The glaciers receded rapidly after 13,000 BP and were within their present borders by 11,000 BP; they remained so during the European Younger Dryas Stade 11,000-10,000 BP. Neoglacial regional readvances culminated 4600-4200 BP, probably 2700-2000 BP, and during the last three centuries; most glaciers reached their Neoglacial maxima during the first episode. Between readvances, the glaciers shrank within their present borders.

Ice wedges are wedge-shaped masses of ice, oriented vertically with their apices downward, a few millimeters to many meters wide at the top, and generally less than 10 m vertically. Ice wedges grow in and are confined to humid permafrost regions. Snow, hoar frost, or freezing water partly fill winter contraction cracks outlining polygons, commonly 5–20 m in diameter, on the surface of the ground. Moisture comes from the atmosphere. Increments of ice, generally 0.1–2.0 mm, are added annually to wedges which squeeze enclosing permafrost aside and to the surface to produce striking surface patterns. Soil wedges are not confined to permafrost. One type, sand wedges, now grows in arid permafrost regions. Sand wedges are similar in dimensions, patterns, and growth rates to ice wedges. Drifting sand enters winter contraction cracks instead of ice. Fossil ice and sand wedges are the most diagnostic and widespread indicators of former permafrost, but identification is difficult. Any single wedge is untrustworthy. Evidence of fossil ice wedges includes: wedge forms with collapse structures from replacement of ice; polygonal patterns with dimensions comparable to active forms having similar coefficients of thermal expansion; fabrics in the host showing pressure effects; secondary deposits and fabric indicative of a permafrost table; and other evidence of former permafrost. Sand wedges lack open-wedge, collapse structures, but have complex, nearly vertical, crisscrossing narrow dikelets and fabric. Similar soil wedges are produced by wetting and drying, freezing and thawing, solution, faulting, and other mechanisms. Many forms are multigenetic. Many socalled ice-wedge casts are misidentified, and hence, permafrost along the late-Wisconsinan border in the United States was less extensive than has been proposed.

A physically plausible three-dimensional numerical ice flow model is used to examine the rate at which the Laurentide Ice Sheet could spread and thicken using as input likely values for the rate of fall of snowline and the amount of net mass balance over the growing ice sheet. This provides then both a test of the hypothesis of “instantaneous glacierization” and of the suggested rapid fall of world sea level to between −20 and −70 m below present at 115,000 BP. Two experiments are described: The first terminated after 10,050 years of model run with ice sheets centered over Labrador-Ungava and Baffin Island with a total volume of 3.0 × 106 km3 of ice, whereas the second was completed after 10,000 years and resulted in a significantly larger ice sheet (still with two main centers) with a volume of 7.78 × 106 km3 of ice. This latter figure is equivalent to the mass required to lower world sea level by 19.4 m. Our results indicate that large ice sheets can develop in about 10,000 years under optimum conditions.

Sorted circles, polygons, and stripes are reported from Alaska, Greenland, Baffin Island, Antarctica, and New Hampshire. From these studies and key references, all cases are found to have: (1) a mixed parent material, commonly till, composed of a wide range of clast sizes unsorted below frost table, (2) gutter depressions containing the largest stones and carrying summer drainage, and (3) tabular stones on edge in the gutters showing expansion-squeezing from the sides. The size of the unit cells, gutter to gutter, is a function of mean maximum clast size: smallest chips making forms 10 cm diameter across and largest forms 20 m across. The slope determines the shape: polygons, and nets form on slopes up to 2 or 4° depending upon amount of water and fines. Ellipses form on 3 to 6° slopes, and stripes form on 4 to 11° slopes. Clearly shape is an effect of solifluction. Lastly, time involves seasons of sporadic sorting until there is a stable end form with lichen-covered stone gutters and tundra-covered soil centers. The up-and-out mechanism, described by Corté, is the best known for the primary sorting. Larger sorted forms (2–20 m in diameter) are reported almost exclusively where nearly continuous permafrost exists. They form where the mean annual temperature is below − 4°C. Former permafrost is indicated where lichen and turf are dense and not overturned and where measured motion is nil. Small forms (under 1 m in diameter) are generated in a year or two where there is only deep annual freezing (0.1–2 m), but no permafrost.

The climate system consists of the atmosphere, the oceans, the cryosphere (land ice, snow, sea ice), the lithosphere, and the biomass. The behavior of the individual components of the system is governed by processes occurring over a broad range of time and space scales. The components are coupled by physical, biological, and chemical processes, and the coupled system seems capable of undergoing fluctuations on all time scales. In addition to these “internal” climatic processes, external processes (such as variability in the solar irradiance or human activities) must also be considered. Space and time scales of climatic variability are reviewed, with emphasis on the Holocene. Regional patterns of climatic variability may be associated with changes in the amplitude and longitudinal position of the long waves in the westerlies of midlatitudes, and with changes in the intensity and latitude of meridional circulation features such as the Hadley cell. Possible examples of this are mentioned. The variance spectrum of climatic time series is described and certain implications for climate modeling are suggested.

The surface temperature curve for the Northern Hemisphere was extended to include the years 1969 through 1973 following the same procedure used by H. C. Willett, J. M. Mitchell, Jr., and C. H. Reitan. The analysis showed a slight warming of 0.02°C between the periods 1965–1969 and 1970–1973, and a significant decrease in the number of negative temperature changes at individual stations (indicating a decrease in the total area experiencing temperature decrease).

Between 60,000 and 40,000 B.P., northeastern Queensland, south New South Wales, and southeastern South Australia were drier than at present. From 40,000–30,000 B.P. a colder climate than at present is indicated from one New Guinea area. Dryness became even more accentuated in northeastern Queensland, whereas many lakes filled up in the southern mainland, probably because of increasing precipitation effectiveness there. Before the end of this period colder conditions than now were already giving rise to slope instability in the Snowy Mountains of New South Wales.

The period of 25,000–15,000 B.P. saw the greatest lowering of the New Guinea treeline, reaching an extreme at 17,000 B.P. when glaciers also achieved their maximum extent. This was the time of extensive glaciation in Tasmania and small glaciers formed in the Snowy Mountains. Estimates of the lowering of mean annual temperature range from 6°–10°C. Northeastern Queensland experienced its driest Late Quaternary climate; lakes were contracting throughout the southern mainland and the final phase of substantial desert dune building took place before the period ended.

In the Snowy Mountains ice retreat began before 20,000 B.P., as did the construction of clay dunes in the southern semi-arid belt, a process demanding higher temperatures. However, in New Guinea and Tasmania ice retreat and treeline rise did not begin till after 15,000 B.P. Temperatures rose rapidly and everywhere most of the ice had gone by 10,000 B.P., when some lakes filled up in southern Australia, implying an increase in absolute precipitation.

In the last 10,000 years climate has been relatively stable although there are some indications that temperature and rainfall were marginally higher than now between 8000 and 5000 B.P. Since then, lake levels have oscillated; a brief, limited resumption of periglacial activity took place in the Snowy Mountains and there were small glacier advances in New Guinea.

A variance spectrum of climatic variability is presented that spans all time scales of variability from about one hour (10−4 years) to the age of the Earth (4 × 109 years). An interpretive overview of the spectrum is offered in which a distinction is made between sources of variability that arise through stochastic mechanisms internal to the climatic system (atmosphere-ocean-cryosphere) and those that arise through forcing of the system from the outside. All identifiable mechanisms, both internal and external, are briefly defined and clarified as to their essential nature. It is concluded that most features of the spectrum of climatic variability can be given tentatively reasonable interpretations, whereas some features (in particular the quasi-biennial oscillation and the neoglacial cycle of the Holocene) remain fundamentally unexplained. The overall spectrum suggests the existence of a modest degree of deterministic forms of climatic change, but sufficient nonsystematic variability to place significant constraints both on the extent to which climate can be predicted, and on the extent to which significant events in the paleoclimatic record can ever manage to be assigned specific causes.

Pingos are large frost mounds which develop in permafrost as the result of the segregation of massive ground-ice lenses. At least two genetic varieties of pingos, open- and closed-systems, form under differing conditions of climate, topography, and groundwater occurrence. Active pingos are known to occur in many high latitude regions. Pingo scars are the degeneration products of pingos. Ideally they are ramparted, circular depressions, although they may be expressed in a variety of divergent forms due to differing conditions of topography, substrate materials, degree of thermokarst overprint, and erosional/depositional histories. Pingo scars occur in many modern permafrost regions. Presumed pingo scars have also been identified in many regions beyond the present permafrost limit and therefore may have utility in reconstructing former permafrost environments.

It has been suggested that during the last glaciation the Innuitian Ice Sheet existed over the eastern Queen Elizabeth Islands. This is based on the pattern of postglacial emergence over this area and the timing of driftwood penetration into the interisland channels. Alternative interpretations of both sets of data raise questions about the presence of the Innuitian Ice Sheet at this time. Field observations on northeastern Ellesmere Island, plus additional data pertaining to the presence of multiple tills and “old” radiometric dates on lacustrine deposits, shelly tills, and raised marine features suggest that the maximum glaciation over this region, equivalent to the Innuitian Ice Sheet, predates the last glaciation, Palaeoclimatic conditions are also discussed in relation to these data. It is suggested that during the last glaciation of the Queen Elizabeth Islands there was a convergent but not coalescent advance of the existing upland ice-fields. This noncontiguous ice cover over the Queen Elizabeth Islands is termed the Franklin Ice Complex. It is suggested that the term Innuitian Ice Sheet be reserved for contiguous older glaciations over this same area.

Frost creep and gelifluction are the cold-climate representatives of mass-wasting processes that occur in a broad range of environments. Neither process requires permafrost, and frost creep can be inhibited by its presence at shallow depth. Acting in various combinations, frost creep and gelifluction produce distinctive lobate and terrace-like landforms, which are easy to recognize while fresh and active, but difficult to distinguish from mudflow lobes, earthslides, and similar deposits after they have been modified by other processes. Large frost creep and gelifluction features are currently active in many tundra environments that experience only deep seasonal freezing; thus they are not generally considered to be indicators of permafrost. Most radio-carbon-dated lobes and terraces, however, seem to have originated at times when permafrost was more widespread than it is today. This is true in the Colorado Front Range, where the formation of lobes and terraces appears to have been initiated by rapid melting of ice-enriched permafrost during the warming phases of frost-heave cycles that were centuries or millennia in duration. There is growing evidence that lobes and terraces developed in many parts of the world between about 3000 and 2500 BP; the climatic significance of their formation during this interval is open to several interpretations. Long-term average rates of frontal advance, calculated for deposits in Colorado, Australia, Greenland, Yukon Territory, Alaska, Scotland, and Norway, range from 0.6 to 3.5 mm per calendar year, significantly slower than maximum rates of movement measured on the surfaces of active lobes and terraces in comparable environments; the features are clearly not as effective at transporting debris as was previously supposed. Variations in past rates of downslope soil movement, estimated from close-interval dating of buried humus horizons or plant remains overrun by the advancing fronts of lobes and terraces, provide a sensitive record of climatic change. The dated humus layers are also suitable for detailed pollen analyses and soil chronosequence studies.

At present the age of ice from ice cores taken at depths where the seasonal isotope fluctuations no longer are measurable is estimated from the Dansgaard-Johnsen-Nye analysis of the longitudinal thinning of ice sheets. A significant error occurs in such an age estimate if ice cores are taken from a hole drilled through ice that has flowed from a region of the ice sheet where ice slides over its bed to a region where it cannot slide over its bed. The Camp Century drill hole is in ice which may have had such a flow history. In the zone where sliding ceases to take place, the horizontal ice velocity decreases in magnitude in the lower part of the ice sheet while it increases in the upper part. The average ice velocity remains unchanged in value. As a consequence the upper part of the ice sheet is stretched and the lower part is compressed as ice moves through the sliding-no sliding transition zone. The upper part is stretched to a strain of the order of 1/2 and the lower part is compressed to a strain that is of the order of 1/2. The age of ice from the Dansgaard-Johnsen-Nye analysis is underestimated; the error in the age is of the order of h/a, where h is the ice thickness and a is the accumulation rate. (Larger errors occur if the theory of Johnsen et al. is used to determine the age of the ice.) An error in age of a similar magnitude exists if ice flows from a region in which sliding does not take place into a region in which it does. The Byrd Station drill hole is in ice which probably has such a flow history. In this situation the age of the ice is overestimated. If the annual isotope fluctuations are detectable the sliding-no sliding zone effect will make it appear that a sudden change in the accumulation rate occurred at the time the ice at the upper surface of the ice sheet passed over this zone.

A basic assumption in some climatic theories is that, given the physical properties of the atmosphere and the underlying ocean and land, specified environmental parameters (amount of solar heating, etc.) would determine a unique climate and that climatic changes therefore result from changes in the environment. The possibility that no such unique climate exists and that nondeterministic factors are wholly or partly responsible for long-period fluctuations of the atmosphere-ocean-earth system, is considered. A simple difference equation is used to illustrate the phenomena of transitivity, intransitivity, and almost-intransitivity. Numerical models of moderate size suggest that almost-intransitivity might lead to persistence of atmospheric anomalies for a whole season. The effect of this persistence could be to allow substantial anomalies to build up in the underlying ocean or land, perhaps as abnormal temperatures or excessive snow or ice. These anomalies could subsequently influence the atmosphere, leading to long-period fluctuations. The implications of this possibility for the numerical modeling of climate, and for the interpretation of the output of numerical models, are discussed.

Quaternary stratigraphy in the northeastern region has developed slowly because researchers have been few and because quartzose bedrock and extensive forest have inhibited field study. The earliest stratigraphic studies antedated 1850, but in most parts of the region the stratigraphic units still recognized today were identified in the present century. Ancient and modern sequences are generalized and are shown in charts.

A review of work on buried paleosols in the disciplines of pedology, Quaternary geology, and archaeology is presented under the headings of (1) the problems of identification, (2) techniques of study, (3) buried paleosols and Quaternary stratigraphy, (4) archaeological stratigraphy and dating, (5) layered soils, and (6) past environment from buried paleosols. It is suggested that future pedological research of interest to Quaternary studies should concentrate on clarifying what is a soil as opposed to a weathered sediment, what processes and features are peculiar to pedogenesis as opposed to diagenesis, and what are the relationships between soil-site conditions and soil characteristics.

Tongue-shaped and lobate rock glaciers are recognized in most alpine regions today. For the tongue-shaped, two situations emerge: those with buried glacier ice (debris-covered glaciers) called ice-cored rock glaciers, and those with interstitial ice known as ice-cemented rock glaciers. Those with ice cores are revealed by depressions between rock glacier and headwall cliff (where a former glacier melted), longitudinal marginal and central meandering furrows, and collapse pits. Ice-cemented rock glaciers ordinarily do not possess these features. As applied to 18 rock glaciers in the Colorado Front Range, 11 of 12 east of the Continental Divide are ice-cored, 6 west of the Divide are ice-cemented. The majority of lobate rock glaciers in the Colorado Front Range are on the south sides of valleys, and, except for talus, are the most voluminous form of mass wasting. All those active and above treeline have characteristics common to all rock glaciers. In addition, they originate from talus, contain interstitial ice, move outward from valley walls at 1–6 cm/yr, and transport more debris as a process of erosion than heretofore realized. Block fields and block slopes, in polar and alpine regions, are thin accumulations of angular to subrounded blocks, on bedrock, weathered rock, or transported debris. They extend along slopes parallel to the contour. Block streams are similar but extend downslope normal to the contour and into valleys. They are made of interlocked blocks without interstitial detritus, but many have finer material deeper inside. The fabric of surface blocks indicates that motion most likely occurred during a periglacial time when interstitial debris, now washed or piped out, permitted movement of the whole deposit.

New evidence from recent field and seismic investigations in the Lake Michigan basin and in the type areas of the Valders, Two Creeks and Two Rivers deposits necessitates revision of late-glacial ice-front positions, rock- and time-stratigraphic nomenclature and climatic interpretations and deglaciation patterns for the period ca. 14,000–7,000 radiocarbon years B.P. The previously reported and long accepted pattern of deglaciation for the Lake Michigan basin started with a regular retreat from the Lake Border Morainic System, with a minor oscillation marked by the Port Huron moraine(s) and then an extensive Twocreekan deglaciation followed by a major (320 km) post-Twocreekan advance (Valders). However, we now record a major retreat between the times of the Lake Border and Port Huron moraines, followed by a gradual retreat from the Port Huron limit and interrupted by a minor standstill (deposition of Manitowoc Till), a retreat (Twocreekan) and a readvance (Two Rivers Till). No Woodfordian or younger readvance was as extensive as had been the preceding one. This sequence argues for a normal, climatically controlled, progressive deglaciation rather than one interrupted by a major post-Twocreekan (formerly Valderan) surge. This revision appears finally to harmonize the geologic evidence and the palynological record for the Great Lakes region. Our investigations show that Valders Till from which the Valderan Substage was named is late-Woodfordian in age. We propose the term “Greatlakean” as a replacement for the now misleading time-stratigraphic term “Valderan”. The type section and the definition of the upper and lower boundaries of the Greatlakean Substage remain the same as those originally proposed for the Valderan Substage but the name is changed.

A series of lectures as comprehensive as those reported in this issue leaves very little for a general reviewer to say. The lecturers have included many of the leading specialists in each of the subfields of palaeoclimatology, upon whose work a reviewer must lean. Moreover there has recently been published a fine synthesis of what is known about the overall character of Quaternary and Holocene climates (U.S. GARP Committee, 1974) (hereafter referred to as GARP-74). All I can do in this situation is to point out some disturbing and persistent problems that tend to defy solution.