James Croll's Physical Theory of Secular Changes of Climate emerged during an age of revolution in geology that included the rise of the glacial theory and the search for its underlying causes. According to Croll, periods of high eccentricity are associated with the persistence of long glacial epochs, within which glaciations occur in alternate hemispheres when winter is at aphelion every ~11,000 years; however, astronomical forcing is only able to produce glaciation by means of physical agencies (climate feedbacks) that amplify the small effects of varying seasonal irradiation. Croll understood the importance of interglacial deposits because they provided evidence for the occurrence of multiple glaciations within his long glacial epochs. He was aware of the limitations of the terrestrial record and suggested that deep-sea sediments would contain a continuous succession of glacial-interglacial cycles. Contrary to a widespread view, however, Croll was not envisaging the advent of palaeoceanographic exploration avant la lettre, but instead was drawing attention to the inadequacy of the land record as a testbed of his astronomical theory. Yet, the marine record did eventually deliver a test of astronomical theories almost exactly 100 years after the publication of his 1875 book Climate and Time in their Geological Relations. Here, we provide an historical account of the technological and scientific developments that led to this and a summary of insights on astronomically paced climate changes from marine, terrestrial and ice core records. We finally assess Croll's ideas in the context of our current understanding of the theory of ice ages.

At Lago Buenos Aires, Argentina, 10Be, 26Al, and 40Ar/39Ar ages range from 190,000 to 109,000 yr for two moraines deposited prior to the last glaciation, 23,000�16,000 yr ago. Two approaches, maximum boulder ages assuming no erosion, and the average age of all boulders and an erosion rate of 1.4 mm/103 yr, both yield a common estimate age of 150,000�140,000 yr for the two moraines. The erosion rate estimate derives from 10Be and 26Al concentrations in old erratics, deposited on moraines that are >760,000 yr old on the basis of interbedded 40Ar/39Ar dated lavas. The new cosmogenic ages indicate that a major glaciation during marine oxygen isotope stage 6 occurred in the mid-latitude Andes. The next five youngest moraines correspond to stage 2. There is no preserved record of a glacial advance during stage 4. The distribution of dated boulders and their ages suggest that at least one major glaciation occurred between 760,000 and >200,000 yr ago. The mid-latitude Patagonian glacial record, which is well preserved because of low erosion rates, indicates that during the last two glacial cycles major glaciations in the southern Andes have been in phase with growth and decay of Northern Hemisphere ice sheets, especially at the 100,000 yr periodicity. Thus, glacial maxima are global in nature and are ultimately paced by small changes in Northern Hemisphere insolation.

The Cenozoic glacial history of East Antarctica is recorded in part by the stratigraphy of the Prydz Bay—Lambert Graben region. The glacigene strata and associated erosion surfaces record at least 10 intervals of glacial advance (with accompanying erosion and sediment compaction), and more than 17 intervals of glacial retreat (enabling open marine deposition in Prydz Bay and the Lambert Graben). The number of glacial advances and retreats is considerably less than would be expected from Milankovitch frequencies due to the incomplete stratigraphic record. Large advances of the Lambert Glacier caused progradation of the continental shelf edge. At times of extreme glacial retreat, marine conditions reached > 450 km inland from the modern ice shelf edge. This review presents a partial reconstruction of Cenozoic glacial extent within Prydz Bay and the Lambert Graben that can be compared to eustatic sea-level records from the southern Australian continental margin.

The modern benthic fauna of the Antarctic continental shelf is characterized by the lack of active, skeleton-breaking (durophagous) predators such as crabs, lobsters and many fish, and the dominance in many areas of epifaunal suspension feeders. It has often been remarked that these ecological characteristics give the fauna a distinctly Palaeozoic feel, with the assumption that it may be an evolutionary relic. We now know that this is not so, and fossil evidence shows clearly that many of the taxa and life-styles that are absent now were previously present. The modern fauna has been shaped by a number of factors, important among which have been oceanographic changes and the onset of Cenozoic glaciation. Sea-water cooling, and periodic fragmentation of ranges and bathymetric shifts in distribution driven by variability in the size and extent of the continental ice cap on Milankovitch frequencies will all have caused both extinction and allopatric speciation. The modern glacial setting with relatively low terrestrial impact away from immediate coastal regions, and scouring by icebergs are the key factors influencing the ecology and population dynamics for the modern Antarctic benthos.

Ocean Drilling Program (ODP) Site 677 provided excellent material for high resolution stable isotope analysis of both benthonic and planktonic foraminifera through the entire Pleistocene and upper Pliocene. The oxygen isotope record is readily correlated with the SPECMAP stack (Imbrie et al. 1984) and with the record from DSDP 607 (Ruddiman et al. 1986) but a significantly better match with orbital models is obtained by departing from the timescale proposed by these authors below Stage 16 (620 000 years). It is the stronger contribution from the precession signal in the record from ODP Site 677 that provides the basis for the revised timescale. Our proposed modification to the timescale would imply that the currently adopted radiometric dates for the Matuyama–Brunhes boundary, the Jaramillo and Olduvai Subchrons and the Gauss–Matuyama boundary underestimate their true astronomical ages by between 5 and 7%.