We present evidence for the absence of the George VI Ice Shelf during a brief period in the mid-Holocene and during one or more earlier interstadials or interglacials. Barnacle Bathylasma corolliforme shells sampled from ice shelf moraines at Two Step Cliffs on Alexander Island have been dated to c, 5750–6000 14C yr BP(c. 6550–6850 cal yr BP) and imply seasonally open water in the George VI Sound during this period. Other shells are beyond the range of radiocarbon dating and imply open water during one or more previous interglacial or interstadial period, prior to 40 000 14C yr BP. Our results show that the ongoing collapse of some Antarctic Peninsula ice shelves is not unprecedented.

A volcanic event, represented by both coarse ash and a prominent sulphate peak, has been detected at a depth of 85.82 m in a 90 m ice core drilled at Talos Dome, northern Victoria Land. Accurate dating of the core, based on counting annual sulphate and nitrate fluctuations and on comparison with records of major known volcanic eruptions, indicates that the event occurred in 1254 ± 2 AD. The source volcano is most likely to be located within the Ross Sea region. In particular, the glass shards have a trachytic composition similar to rocks from The Pleiades and Mount Rittmann (Melbourne volcanic province), about 200 km from Talos Dome. Sulphate concentration is comparable with that of violent extra-Antarctic explosive events recorded in the same core, but atmospheric perturbation was short-lived and localized, suggesting a negligible impact on regional climate. It is suggested that this eruption may represent the most important volcanic explosion in the Melbourne province during the last eight centuries; thus this event may also represent a valuable chrono-stratigraphical marker on the East Antarctic plateau and in adjoining areas.

A geomorphological sketch map of the Fossil Bluff area, showing the main morphological characteristics was prepared from aerial photographs taken by the British Antarctic Survey in 1995. Landforms and deposits mapped include those related to glaciers and the local hydrography, periglacial and nival activity, and wind and gravity action. Data mapped originally at a scale of 1:12 500 and simplified for presentation here at a scale of about 1:30 000, have been digitized as vectors using a GIS. In this way these data may easily be integrated with other environmental data for future research.

Antarctic petrels Thalassoica antarctica on Ardery Island, Antarctica (66°S, 110°E), experienced major reductions in breeding success and breeder survival over four seasons between 1984/85 and 1996/97. In 1996 the reason was revealed. A large snowdrift covered part of the study colony on the cliffs. Southern giant petrels Macronectes giganteus, normally lacking access to this area, exploited the snow for soft ‘crash landings”. After landing they waited for the disturbed birds to resettle on their nests and then used surprise to seize and kill a victim. Predation continued into the egg period, and only stopped after the snowdrift had melted. Giant petrels showed no interest in the eggs but, during the panic caused by their activities, South Polar skuas Catharacta maccormicki took the deserted eggs. Antarctic petrel mortality due to predation within the 1996/97 season amounted to 15.4% of experienced breeders, and breeding success was reduced to virtually zero. Weather data from the nearby Casey station over the 1980–96 period showed that a significant increase in precipitation has occurred, in combination with shifts in speed and direction of winds. We conclude that the decreases in breeding success and survival in earlier seasons were also related to increased snowfall and predation. Although similar predation behaviour by giant petrels has not been reported before, we think that it is long established and explains why nesting of the smaller fulmarine petrels is limited to steeper cliffs or sheltered sites. The complexity of the response seems unlikely to be predicted by our present understanding of how climate change affects ecosystems.

In the Amery Oasis of the northern Prince Charles Mountains, the glaciomarine Bardin Bluffs Formation of the Pagodroma Group was deposited between the Late Pliocene (<3.1 Ma) and Early Pleistocene (>1 Ma). The formation provides evidence of (i) a reduced East Antarctic ice sheet compared to that of the present day and (ii) a subsequent Plio–Pleistocene ice sheet expansion. The formation consists of two members. The older, basal Member 1 is c. 12.5 m thick and consists of relatively ice-distal silty, sandy and sparsely fossiliferous fjordal strata. Member 1 reflects largely ice-free marine sedimentation c. 250 km inland from the current Amery Ice Shelf edge. The member is restricted to the area about the north-eastern end of Pagodroma Gorge where it infills a chemically weathered erosion surface, cut in the form of a valley on the Permo-Triassic Amery Group. Weathering occurred during aerial exposure of the Amery Oasis in a warmer climate than that of today. The younger Member 2 exceeds 40 m in thickness and is made up of coarse ice proximal glaciomarine diamicts. It overlies disconformably Member 1 at Pagodroma Gorge. Elsewhere, Member 2 rests directly upon a smoothed and striated erosion surface, cut on the Amery Group, which was part of a fjord floor. This erosional surface and the facies contrast between the two members, indicates an East Antarctic Ice Sheet expansion and Lambert Glacier grounding-line advance.

Airborne radar data acquired in 1995 by the Italian Antarctic Programme over Dome C in central East Antarctica were processed to develop maps of internal isochronous ice sheet layering around the EPICA ice core site. Three internal layers were traced continuously across the radar-survey area at ice depths of 1–2 km. The maps reveal that the ice core site is located where internal layers are near horizontal to depths of at least 2 km. The Italian radar data do not resolve internal layers below 2 km. However, radar data collected over this part of East Antarctica in the 1970s show the internal layers to depths of up to 4 km. These internal layers reveal the regional structure of ice to the west of Dome C. Layers from both surveys are dated through an existing chronostratigraphic link between the Vostok ice core site and Dome C. The pattern of internal layering at Dome C reflects relatively steady conditions of ice flow and accumulation for the last 100 000 years. However, for ice older than this we show that there is significant local variation in the thickness between internal layers and the ice-sheet base. Our maps provide an indication of the structure of the ice sheet from which the EPICA deep ice core will be taken.

Lake sediments in the Larsemann Hills contain a great diversity of biological and physical markers from which past environments can be inferred. In order to determine the timing of environmental changes it is essential to have accurate dating of sediments. We used radiometric (210Pb and 137Cs), radiocarbon (AMS 14C) and uranium series (238U) methods to date cores from eleven lakes. These were sampled on coastal to inland transects across the two main peninsulas, Broknes and Stornes, together with a single sample from the Bolingen Islands. Radiometric dating of recent sediments yielded 210Pb levels below acceptable detection limits. However, a relatively well-defined peak in 137Cs gave a date marker which corresponds to the fallout maximum from the atmospheric testing of atomic weapons in 1964/65. Radiocarbon (AMS 14C) measurements showed stratigraphical consistency in the age-depth sequences and undisturbed laminae in some cores provides evidence that the sediments have remained undisturbed by glacial action. In addition, freshwater surface sediments were found to be in near-equilibrium with modern 14CO2 and not influenced by radiocarbon contamination processes. This dating program, together with geomorphological records of ice flow directions and glacial sediments, indicates that parts of Broknes were ice-free throughout the Last Glacial Maximum and that some lakes have existed continuously since at least 44 ka bp. Attempts to date sediments older than 44 ka bp using 128U dating were inconclusive. However, supporting evidence for Broknes being ice-free is provided by an Optically Stimulated Luminescence date from a glaciofluvial deposit. In contrast, Stornes only became ice-free in the mid to late Holocene. This contrasting glacial history results from the Dålk Glacier which diverts ice around Broknes. Lakes on Broknes and some offshore islands therefore contain the oldest known lacustrine sediment records from eastern Antarctica, with the area providing an ice-free oasis and refuge for plants and animals throughout the Last Glacial Maximum. These sediments are therefore well placed to unravel a unique limnological sequence of environmental and climate changes in East Antarctica from the late Pleistocene to the present. This information may help better constrain models of current climate changes and ensure the adequate protection of these lakes and their catchments from the impacts of recent human occupation.

Large-scale bedrock morphology and relief of two key areas, the Jutulsessen Nunatak and the Jutulstraumen ice stream are used to discuss glascial history and landscape development in western and central Dronning Maud Land, Antarctica. Two main landform components were identified: well-defined summit plateau surfaces and a typical alpine glacial landscape. The flat, high-elevation plateau surfaces previously were part of one or several continuous regional planation surfaces. In western Dronning Maud Land, overlying cover rocks of late Palaeozoic age show that the planation surface(s) existed in the early Permian, prior to the break-up of Gondwana. A well-develoment escarpment, a mega landform typical for passive continental margins, bounds the palaeosurface remnants to the north for a distance of at least 700 km. The Cenozoic glacial landscape, incised in the palaeosurface and escarpment, is exemplified by Jutulsessen Nunatak, where a c. 1.2 km deep glacial valley system is developed. However, the prominent Penck-Jutul Trough represents some of the deepest dissection of the palaeosurface. This originally tectonic feature is today occupied by the Jutulstraumen ice stream. New topographic data show that the bed of the Penck-Jutul Trough is situated 1.9±1.1 km below sea level, and that the total landscape relief is at least 4.2 km. Today's relief is a result of several processes, including tectonic faulting, subaerial weathering, fluvial erosion, and glacial erosion. It is probable that erosion by ice streams has deepened the tectonic troughs of Dronning Maud Land since the onset of ice sheet glaciation in the Oligocene, and continues today. An attempt is made to identify major events in the long-term landscape development of Dronning Maud Land, since the break-up of the Gondwana continent.

The northern Prince Charles Mountains overlook the western side of the 700 km long Lambert Glacier–Amery Ice Shelf drainage system. Within these mountains, at Amery Oasis (70°50′S, 68°00′E) and Fisher Massif (71°31′S, 67°40′E), the Cenozoic glaciomarine Pagodroma Group consists of four uplifted Miocene and Pliocene–early Pleistocene formations here named the Mount Johnston, Fisher Bench, Battye Glacier and Bardin Bluffs formations. These are composed of massive and stratified diamicts, boulder gravels and minor laminated sandstones, siltstones and mudstones. Each formation rests on either Precambrian metamorphic rocks, or on Permo-Triassic fluvial strata. The unconformity surfaces are parts of the walls and floors of palaeofjords. The Miocene Fisher Bench Formation exceeds 350 m in thickness at Fisher Massif, where the yet older Miocene (or Oligocene) Mount Johnston Formation overlies basement rocks at up to 1400 m above sea level. Individual formations contain either Miocene diatoms, or else Pliocene–early Pleistocene diatom-foram assemblages. The diamicts are interpreted as fjordal ice-proximal or ice-contact sediments, deposited seawards of tidewater glacier fronts located some 250 to 300 km inland of the present ocean margin. Each formation records an ice recession following a glacial expansion.

We present preliminary results of the first detailied surveys of the former Larsen-A Ice Shelf, Larsen Inlet and southern Prince Gustav Channel, where disintegration of small ice shelves in the past ten years has exposed the seafloor. Glacial troughs in the Larsen-A area, Larsen Inlet and Prince Gustav Channel reach 900–1100 m depth and have hummocky floors. Farther south-east, the continental shelf is shallower (400–500 m) and its surface is fluted to smooth, with the density of iceberg furrowing increasing towards the shelf edge. Acoustic profiles show a drape of transparent sediment 4–8 m thick in Prince Gustav Channel, thinning southwards. In cores, this drape corresponds to diatom-bearing marine and glacial-marine mud. In the Larsen-A area and Larsen Inlet, acoustically opaque sediment includes proximal ice shelf glaciomarine gravelly and sandy muds, and firm to stiff diamicts probably deposited subglacilly. These are overlain by thin (up to 1.3 m) glaciomarine muds, locally with distinctive diatom ooze laminae.

Although Sims Island (73°17′S/78°33′W), a bedrock peak in Carroll Inlet on the English Coast of eastern Ellsworth Land (Fig. 1), had previously been observed and photographed from the air (e.g. Rowley & Thomson 1990, fig. C.6.2), no landing had been made prior to the present study. Based on proximity and apparent lithological similarities, the island was thought likely to consist of alkaline volcanic rocks similar to those exposed on nearby Snow Nunataks (Fig. 1; e.g. Rowley & Thomson 1990). Smellie (1999) tentatively included it the Bellingshausen Sea Volcanic Group (BSVG), which consists of Miocene-Recent alkaline volcanic outcrops scattered across Alexander Island and eastern Ellsworth Land. More specifically, he provisionally assigned it to the Mount Benkert Formation (MBF), a probable subglacial volcanic succession exposed on three of the four Snow Nunataks. This note is based on observations made and samples collected during a short visit to Sims Island on 6 March 1996.

Deception Island is the most active volcano in the Antarctic Peninsula region. It is a large basalt–andesite shield volcano with a 10 km-wide restless caldera (Port Foster) and a complicated history of pre- and post-caldera eruptions. There has been no modern volcanological investigation of the entire island and it remains a largely unknown volcanic hazard. The pre-caldera period on the island began with the low-energy eruption of tephras from multiple centres (Fumarole Bay Formation), possibly by subaqueous fire fountaining during shoaling and likely initial emergence of the volcano. It was followed by subaerial effusive to weakly pyroclastic (Strombolian/Hawaiian) activity that constructed a small basaltic shield (Basaltic Shield Formation), and a large eruption that vented about 30 km3 of magma (Outer Coast Tuff Formation). The latter eruption may have been triggered by an influx of compositionally different magma into the main chamber feeding the volcano, and the evidence suggests that it was associated with a significant involvement with water (seawater?). The eruption was followed by caldera collapse, and there have been several small incremental caldera “collapses” subsequently. Post-caldera eruptions were all small-volume and predominantly phreatomagmatic (Baily Head and Pendulum Cove formations), but magmatic eruptions constructed several small lava deltas around the coast and also produced a local carapace of scoria and thin lavas, particularly around the caldera rim (Stonethrow Ridge Formation). Although the caldera is presently resurging, interpretation of the eruptive history of the island suggests that future eruptions are likely to be small in volume and will have only a limited regional impact.

New outcrops of Hope Bay Formation and Larsen Basin rocks have been exposed by recent ice-retreat on Tabarin Peninsula, northern Antarctic Peninsula. Quartzose, very low-grade metasedimentary rocks, cropping out at Balegno Nunatak (63°28′53″S, 57°00′31″W) are assigned to the Permian–Triassic Hope Bay Formation, and dioritic rocks are assigned to the Antarctic Peninsula batholith. Sedimentary rocks exposed at Rubulis Nunatak (63°30′52″S, 57°07′29′W) closely resemble Lower Cretaceous sedimentary rocks cropping out at nearby Troilo Nunatak, western Tabarin Peninsula. These rocks are correlated with the lower Gustav Group (?Barremian–Coniacian) of the Larsen Basin. This new record confirs that the Larsen Basin extends over southern Tabarin Peninsula, where basin sediments are thought to be faulted against sediments of the Antarctic Peninsula magmatic arc.

Moisture sources for snow falling in Dronning Maud Land (DML), Antarctica, are calculated for 1998 using three dimensional 5-days backward air parcel trajectories. The drilling site of the European Project on Ice Coring in Antarctica (EPICA) in DML is chosen as the main arrival point (75.0°S, 0.01°E). A distinction is made between trajectories with and without snowfall at arrival. Of the snowfall trajectories, 40–80% are located in the Atlantic Ocean within four days before arrival. Evaporation along these trajectories is largest three to four days before arrival. The air parcels are then located between 40° and 60°S in the Atlantic Ocean where surface temperatures range between 0° and 20°C. A case study for May 1998 shows that when snow falls exceptionally high temperatures and wind speeds prevail in the atmospheric boundary layer. The position of the trajectories in the boundary layer suggests a source region for this event between 40° and 50°S and 20° and 60°W in the Atlantic Ocean, where sea surface temperatures vary between 5° and 15°C.

High resolution firn core records of the oxygen isotope ratio (δ18O) and trace chemical species were extracted from a high accumulation site on Law Dome, East Antarctica. Inter-core comparisons were conducted and regional events identified in cores 5 km apart. High resolution dating of one of the firn cores was established using a co-located Automatic Weather Station (AWS) equipped with a snow accumulation sensor, allowing dating of individual precipitation events in the firn core record. Variations in the δ18O and trace chemical records were compared with meteorological conditions at the mesoscale and the synoptic-scale. Particular focus was given to an abrupt change in sea salt concentrations and δ18O within a depth range that appears from AWS accumulation data to have been deposited over a 24 hour period. The abrupt change in the firn core record was found to be consistent with an abrupt change in meteorological conditions. Direct comparisons between high resolution firn core records and meteorological conditions will greatly facilitate the interpretation of signals preserved in deep ice cores.