The Antarctic Ice Sheet is today and has been for several million years a major component of the global climate system, responsible for deep- and bottom-water formation and eustatic sea-level change. It is also a source of "noise" in the oxygen isotopic record, which limits the value of that record to other studies through most of the Cenozoic.

At present, the history of the Antarctic Ice Sheet is unknown: it has been inferred from low-latitude proxy data-principally from oxygen isotopic measurements on deep-ocean benthic foraminifers-and from the record of eustatic sea-level change adduced from sediments on low-latitude margins (e.g., Miller et al., 1987; Haq et al., 1987). However, these inferences are both ambiguous and in disagreement (e.g., Sahagian and Watts, 1991; Barker, 1992), which not only leaves the history unresolved but also limits the credibility and usefulness of both sets of proxy data. For example, there is a dispute over whether the principal increases in Antarctic ice volume, affecting the benthic isotopic record, occurred at ~35 Ma, at 16-13 Ma, or only after ~3 Ma. Within these various hypotheses, assumptions have to be made about the constancy of equatorial surface temperatures, or the high-latitude surface origins and temperatures of intermediate to deep waters at low latitudes, that may be incorrect. Similarly, changes in grounded ice volume provide the only generally accepted repeatable, rapid-acting cause for global eustatic sea-level change, yet the timing and amplitudes of sea-level change adduced from low-latitude margin sediments are disputed. Additionally, changes occur at times when there is no independent evidence for the existence of substantial volumes of grounded ice on Antarctica or elsewhere. Further, the isotopic and sea-level estimates of grounded ice volume disagree substantially with each other, at both long and short periods, through most of the Cenozoic. Other proxies (bottom-water formation and ice-rafted detritus) are less useful because, although they are certainly produced directly by the present ice sheet, their relationship with ice volume changes is much less direct and is easy to misinterpret. Onshore Antarctic evidence of glacial history is sparse and at present controversial; the argument continues as to how stable the Antarctic Ice Sheet has been (e.g., Webb and Harwood, 1991; Denton et al., 1993; Barrett, 1996).

The opportunities for learning more about ice-sheet history are limited. Deep and intermediate waters of the Southern Ocean have generally been corrosive to the carbonate tests almost exclusively used in isotopic analysis, so the problems of indirect estimates of ice volume from distal proxy data will persist. Some progress may be made by detailed analysis at very high resolution of carbonate sections from a large number of lower latitude sites, but the solutions will remain ambiguous. Other proxies, and the onshore geological record, seem unlikely to be able to help. However, the Antarctic margin sediments hold a direct record of Antarctic ice-sheet fluctuation that can help resolve the ambiguities of ice-volume change and clear the way for more useful interpretation of isotopic and sea-level data in the future.

The long-term aim of the suite of linked ANTOSTRAT drilling proposals, reviewed and adopted by the Detailed Planning Group, is to provide an estimate of variations in the size of the Antarctic Ice Sheet through the Cenozoic. This will necessarily include periods when the ice sheet was much smaller and warmer than it is today, reaching the margin only occasionally and in a few places, with significant fluvial sediment transport and deposition elsewhere. It is therefore necessary for drilling to sample both East and West Antarctic glacial history and to distinguish a small interior ice sheet, barely reaching the margin, from a much larger ice sheet with a large coastal ice budget. This means making use of numerical models to suggest what might have been the patterns of past glaciation and drilling in different regions, as the models or other relevant information might suggest. For example, Figure F1 (from Huybrechts, 1993) shows a glaciological model of ice sheets that cover only parts of the continent during warmer conditions. It is clear that some regions will be more sensitive to particular stages of ice-sheet volume change than others and that no single region will provide a complete history. The models provide the means of combining data from different regions of the Antarctic margin into a complete history of ice-sheet development. For a particular region (areas where proposals are or will be focused are starred in Fig. F1), it will be important to establish when the ice sheet first arrived at the margin, when there were changes in the rate of ice transport to the margin (that might have influenced both the size of the particular ice catchment and the geometry of sedimentation), and what was the sensitivity of ice in this region to Milankovich-scale change. The record of these events and sensitivities lies within glacially transported sediments deposited at the margin and is in most cases accessible to drilling.

Great strides have been made in recent years in the collaborative interpretation of seismic data from the Antarctic margin through the ANTOSTRAT initiative (see Cooper et al., 1994, 1995; Barker and Cooper, 1997). Together with the simplicity of the modern Antarctic glacial regime (compared with that of the Arctic), these advances have led to the rapid emergence and application of a unifying model of glacial sediment transport and deposition (Alley et al., 1989; Larter and Barker, 1989; Bartek et al., 1991; Cooper et al., 1991; Kuvaas and Kristoffersen, 1991). Briefly, almost all ice transport to the ice-sheet margins takes place within broad, rapidly moving ice streams. Rapid flow is enabled by low-friction basal conditions, whose main source is the existence of an overpressured and undercompacted, unsorted, shearing basal till. The necessary shear ensures that ice transport is accompanied by till transport, and virtually all of the transported till is melted out, dropped, and deposited very close to the grounding line, where the ice sheet becomes ice shelf before calving into icebergs and drifting north. The ice stream therefore essentially erodes and transports inshore of the grounding line and deposits directly offshore in a high-latitude analogue of the low-latitude subaerial erosion/shoreline/marine sedimentation system (the level of erosion, however, is not sea level but the ice-stream base). Further, the grounding line advances and retreats under the influence of upstream ice provision and basal sediment supply--and sea-level change--that are all related to climate. The very large prograded sediment wedges beneath the Antarctic margin were developed during a series of glacial maxima, when the ice sheet was grounded all the way to the continental shelf edge (Fig. F2).

The glacial sedimentation regime has other characteristics. Shelf-edge progradation is usually focused into broad "trough-mouth fans" opposite the main ice streams, and the shelf is overdeepened (generally to between 300 and 600 m depth, but in places much deeper) and inward sloping. Continental slopes are often steep, and in places turbidity current transport of the unstable component of slope deposition (with downcurrent deposition of suspended fines) has produced large, mainly pelagic/hemipelagic sediment drifts on the continental rise (Kuvaas and Leitchenkov, 1992; Rebesco et al., 1996, 1997; Escutia et al., 1997; Fig. F3). Sediment supply to the slope and rise is highly cyclic, with large quantities of unsorted diamicton deposited during glacial maxima and very little during interglacial periods.

Three depositional environments are recognized: shelf topsets and slope foresets of the prograded wedge and proximal hemipelagic drifts on the continental rise. Of these, the shelf record is potentially the least continuous. There, sediment is preserved mainly as a result of slow subsidence from cooling and from flexural response to the topset and foreset load. The sediment is prone to re-erosion during the next glacial advance. The topsets tend to mark only the major changes in glacial history so that the more continuous foreset record is an essential complement. The proximal rise drifts may not always be present and are as yet sparsely sampled, but potentially they contain an excellent record, closely related to that of the upper slope foresets from which they are largely derived. Existing seismic data and drill sites from around Antarctica have demonstrated the coarse (but not as yet the fine-scale) climate record in continental rise sediments and the likely climatic sensitivity of margin wedge geometry (Barker, 1995a). These sites have also revealed the partial nature of the shelf topset record (Hayes, Frakes, et al., 1975; Barron, Larsen, et al., 1989).

The continental shelf is an area of high biogenic productivity during interglacial periods. Although long-term sediment preservation on the shelf is limited by the erosional effects of grounded ice sheets during subsequent glacials, biogenic interbeds are preserved within sequence groups composed mainly of thick glacial diamicton topsets and foresets. In addition, glacially eroded deeps can preserve expanded Holocene sections that may be continuous and essentially biogenic, if the ice-sheet grounding line is sufficiently remote that ice-rafted debris is minor or absent and the sections are well protected from bottom--current action. Such sections can provide a record of decadal and millennial variability that can be compared with high-resolution records from low latitudes and from the ice sheet itself. This environment is available on the inner shelf of the Antarctic Peninsula at Palmer Deep (Domack and McClennen, 1996; Leventer et al., 1996) and was sampled during Leg 178.

Different parts of Antarctica have had different glacial histories, as the numerical models of ice-sheet development (e.g., Fig. F1) suggest. The present Antarctic Ice Sheet comprises an East Antarctic component grounded largely above present sea level and a West Antarctic component grounded mostly below sea level. Marine-based (West Antarctic) ice sheets are considered less stable. There is evidence from around Antarctica that, although East and West Antarctic climates were coupled in the past, changing approximately in phase, the climate of West Antarctica (including the Antarctic Peninsula) has varied about a consistently warmer baseline (e.g., Kennett and Barker, 1990; Wise et al., 1992). Although East Antarctic glaciation extends to 35 Ma or earlier, West Antarctic glaciation probably began more recently, during generally colder times. Further, there is strong evidence that Northern Hemisphere glaciation has been the main contributor to global sea-level change over the past 0.8 m.y. (and probably 2.5 m.y.) and has therefore partly driven the more subdued changes in Antarctic glaciation. Another significant local control may have been the Transantarctic Mountains, which probably attained much of their present elevation and influence on the East Antarctic Ice Sheet during late Cenozoic time. The Antarctic Peninsula has many specific glacial and paleoglacial features but has also a unique regional tectonic environment that both constrains and enhances the opportunities for drilling.