Maximum and minimum Northern Hemisphere ice-sheet reconstructions were necessary because glacial geological studies in the Arctic are inconclusive (Reference Hughes, Hughes, Denton, Andersen, Schilling, Fastook, Lingle, Denton and HughesHughes and others, in press). One body of work favors ice sheets which extended to the margins of the Arctic continental shelf, but another body has these ice sheets terminating near present Arctic shorelines. Ice sheets were reconstructed assuming isostatic equilibrium and rock–ice density ratios of 3 and 4. For the maximum reconstruction, a density ratio of 3 requires removing 172 m of ocean water, Laurentide ice up to 3230 m high with 2930 mmean thickness, Scandinavian ice up to 2710 m high with 2700 m mean thickness, and Antarctic ice up to 4050 m high with 2330 m mean thickness; whereas a density ratio of 4 requires removing 164 m of ocean water, Laurentide ice up to 3800 m high with 2920 m mean thickness, Scandinavian ice up to 2520 m high with 2190 m mean thickness, and Antarctic ice up to 4050 m high with 2310 m mean thickness. By contrast, for the minimum reconstruction with a density ratio of 4 only 126 m of ocean water was removed, Laurentide ice was up to 3480 m high with 2690 m mean thickness, and Scandinavian ice was up to 2520 m high with 2190 m mean thickness.

During the last interglaciation 12000 years ago mean global sea-level was about 6 m higher than at present. This sea-level rise could have resulted from complete collapse of the Greenland ice sheet or the Antarctic ice sheet west of the Transantarctic Mountains, or partial collapse of an Antarctic ice-sheet drainage basin east of the Transantarctic Mountains. Since the West Antarctic ice sheet is mostly grounded on a continental shelf that is permanently below sea-level, it is a marine ice sheet and is believed to be inherently unstable. For this reason, we chose to collapse it for the CLIMAP 125000 years b.p. experiment using our marine ice-sheet disintegration model.

Two West Antarctic ice-sheet disintegration histories were modeled, both starting with a maximum West Antarctic ice sheet which extended to the continental shelf margin during the previous ice age (Reference Stuiver, Stuiver, Denton, Hughes, Denton and HughesStuiver and others, in press). Disintegration was triggered and maintained by rising sea-level during the glacial–interglacial transition. Isostatic rebound was assumed to nearly keep pace with disintegration so that approximate isostatic equilibrium was maintained. Both disintegration histories began with ice-stream grounding lines retreating from sills at the continental shelf margin as sea-level rises. Initially the ice-sheet calving fronts and grounding lines nearly coincided because a confined ice shelf cannot form beyond the continental-shelf margin.

In the ice-shelf disintegration history, the West Antarctic grounding lines retreated faster than the calving fronts so that floating ice shelves were created in the Ross, Amundsen, Bellingshausen, and Weddell Seas. The grounding line in the Amundsen and Bellingshausen Seas temporarily halted along the barrier of volcanos and mountains extending from the Edsel Ford Range to the Antarctic Peninsula, but the calving ice-shelf front continued to retreat. When this ice shelf no longer effectively buttressed Thwaites and Pine Island Glaciers, their grounding lines retreated over the barrier and an ice shelf was created in the Byrd Subglacial Basin. This ice shelf, together with the Ross and Filchner-Ronne Ice Shelves, completed the ungrounding of the West Antarctic ice sheet. These ice shelves remained intact during the last interglaciation, so surface albedos did not change dramatically even though surface elevations did.

In the calving-bay disintegration history, the West Antarctic grounding lines and calving fronts retreated at the same rate so that extensive floating ice shelves never formed. Instead, calving bays migrated up surging ice streams and carved out the heart of the West Antarctic marine ice sheet. Seasonally open seas replaced the West Antarctic ice sheet so both surface albedos and surface elevations changed dramatically. Our modelling results pinpoint Pine Island Bay, where Thwaites and Pine Island Glaciers now terminate, as an active calving bay that may control the Holocene and present-day collapse of the West Antarctic ice sheet, provided that high bedrock sills are not present.