In the upper Quaternary, there are diatoms in the interglacial samples, implying seasonally open water, whereas the barren glacials probably indicate near-permanent sea-ice cover, perhaps with intermittent polynyas (Pudsey, 2000).
In the Pliocene, diatom abundance reached a maximum at ~4.3 Ma, which is consistent with the warm period in the Southern Ocean inferred by Hodell and Warnke (1991) and Bohaty and Harwood (1998). Diatoms occur throughout the cycles, suggesting that there was at least seasonally open water through glacials and interglacials. Glacial sea ice was likely of much shorter seasonal extent than in the Quaternary (Hillenbrand and Fütterer, Chap. 23, this volume). There is no evidence for anoxia, so in order to suppress bioturbation, the laminated units must have been much more rapidly deposited. This could mean deposition of each turbidite during only hours to days, with relatively long intervals between turbidity currents.
In the Miocene, generally lower diatom content in the laminated sediments, coupled with high sedimentation rates, suggests that diatom productivity was maintained throughout the cycles but that biogenic silica was diluted at the site by high terrigenous input of laminated sediment. Lower fine-grained terrigenous input during deposition of the bioturbated units increased the concentration of diatoms and ice-rafted sand as well as allowed thorough mixing of the sediment by burrowing organisms.
The fine grain size measured throughout the section down to the upper Miocene (Fig. F4) suggests that bottom currents have remained predominantly weak throughout deposition. Even the relatively sandy samples contain a considerable proportion of clay. The silts interpreted as contourites are very thin (usually 1-2 mm) (Shipboard Scientific Party, 1999) and do not have erosive bases. This implies that currents strong enough to winnow clay-size particles and deposit only silt did not persist for very long. Median grain sizes in the range of 8-11 
(4-0.5 µm) and modes clustering around 8 are consistent with bottom currents comparable to the average 6 cm/s measured at mooring sites on Drift 7 (Camerlenghi et al., 1997). Silt modes around 6 (Fig. F4C) were measured in samples that included a turbiditic component.
Poorly sorted or unsorted sand interpreted as ice rafted is present throughout the section, mainly in the bioturbated facies (Fig. F4). Hassler and Cowan (Chap. 11, this volume) examined over 300 gravel clasts from Sites 1096 and 1101 as well as Site 1095 and were able to match all lithologies to Antarctic Peninsula source rocks. This is relevant to the possibility suggested by Webb and Harwood (1991) that during the early Pliocene the West Antarctic Ice Sheet (including the ice sheet on the Antarctic Peninsula) may have disappeared completely. If this were the case and glacier ice no longer reached the coast, the supply of IRD to Site 1095 would have ceased, except perhaps for very rare clasts from distant sources in East Antarctica. It is not impossible that the ice sheet may have been very much smaller during "interglacial" parts of the depositional cycles (cf. Hall et al., 1997), but there must have been a source of sediment-laden ice not far from Site 1095 to account for the continued presence of IRD.
The model for late Quaternary deposition on Drift 7 shows a grounded ice sheet extending to the shelf edge during glacial periods (Larter and Barker, 1991; Pudsey and Camerlenghi, 1998). The ice sheet transports sediment to the upper slope, which is very steep in this region (>10°) (Rebesco et al., 1998). Here it accumulates for only short periods before mass-flow processes transport it into deeper water. During interglacials the ice margin is at or near the present-day coast and most glacial sediment remains on the shelf. The geometry of seismic reflectors beneath the shelf and slope has been interpreted to result from similar alternating positions of the ice margin in the past (Larter et al., 1997).
According to this model, the alternation of bioturbated, sandy facies and laminated, silty facies downcore at Site 1095 could simply represent alternating interglacial and glacial conditions. If this were the case one might expect a reasonably regular cyclicity with prominent orbital periods of 19/23, 41, and 100 k.y. However, real continental margin sediments probably do not respond in a linear way to climate forcings (see below). The persistence of cyclically alternating facies throughout lithologic Units I and II supports the seismic interpretation of alternating positions of the ice margin. Only in Unit III does the more uniform style of sedimentation (laminated claystone without sandy or bioturbated intervals) suggest the nature of glaciation may have been different.
The rather curious results from the MATLAB spectral analysis may be attributed to a less-than-ideal input function. Even if growth and decay of the Antarctic Peninsula ice sheet had been smooth and cyclic (a questionable assumption), terrigenous sediment supply to the continental rise would have increased sharply each time the ice sheet advanced to the shelf edge and, conversely, decreased each time ice retreated back over the shelf. The high sedimentation rate glacials and low sedimentation rate interglacials therefore cannot match the age model used in the frequency analysis, which is linear between paleomagnetic datum points. The simple, visual technique of counting peaks and troughs in physical properties (corresponding approximately to glacial-interglacial cycles) is actually more useful for these sediments than is a rigorous spectral analysis.
A likely complicating factor is differing amplitudes of cycles at different times in the past because of changing phase relationships in the precession, obliquity, and eccentricity forcings. During some glacials the ice sheet might not have reached the shelf edge at all or have remained there for only a short time; conversely, some interglacials might scarcely have allowed time for meltback to the coast before the next ice advance. This could account for cycles of regular spacing but differing thickness of the bioturbated and laminated parts of the cycles, for example, the color cycles in Figure F6. Changes in sediment supply in the late Pliocene apparently included suppression of the turbidite source, leading to the accumulation of bioturbated, sandy sediment once every ~160 k.y. (Fig. F6), plus a change in the source or concentration of magnetic mineral grains approximately every 100 k.y. (magnetic susceptibility cycles in Fig. F6). In the upper Pliocene this cyclicity is not present.
At Site 1095, which is near the distal edge of Drift 7 and has been influenced by turbidity currents flowing through the adjacent channel, processes intrinsic to a submarine channel-lobe system may also have resulted in alternation of dominantly turbiditic (laminated and rapidly deposited) and hemipelagic (bioturbated, containing IRD, and slowly deposited) facies, independent of climate forcing. On continental margins in low and mid-latitudes, seismic and core studies have documented channel avulsion with associated lobe progradation or abandonment, and these processes have commonly been inferred in ancient sedimentary successions (review by Bouma et al., 1989). Although a degree of climatic control on continental rise sedimentation is to be expected (e.g., sea level change, particularly in the Neogene), local sediment supply related to drainage basin geology and tectonic uplift may mask any orbital cyclicity.
