4.1. Interpreting Os-Isotope Data
 Initial 187Os/188Os ratios reflect the isotopic composition of the whole sediment at time T0, thus representing the weighted average of hydrogenous-Os and detrital-Os from all sources. Mild leaches will, in the best of cases, dissolve only the hydrogenous fraction of the elements, leaving the detrital, mineral-bound osmium unattacked.
 Because of its anoxic nature, a large fraction of the Re from the deep sedimentary section must have been water-derived to account for its high content. Part of this Re has disintegrated in situ since deposition. As a consequence, part of the 187Os released through leaching will be post-depositional radiogenic Os, unrelated to synsedimentary supplies. It is impossible to isolate and label the initial 187Os fractions, using any extraction procedure. However, in old sedimentary rocks, the isochron method may yield a precise value for the initial isotopic composition of the sediment, provided the isochron regression-line is statistically robust (with a plausible age, when compared to independent stratigraphical or geochronological benchmarks). Unfortunately, this initial value will not label unequivocally the marine Os-signature, due to possible “contamination” by terrigenous material [Ravizza et al., 1991; Poirier, 2006], or by non-oceanic dissolved-Os [McArthur et al., 2008].
 Using a mild leach-procedure [Ravizza, 2007], the Os-isotope ratios obtained during this study are more radiogenic than those of the bulk sediment (between 0.8 and 12.6% higher 187Os/188Os; Table S1). One exception is the anoxic sample (at 199.8 mcd), which yielded a leach ratio 50% lower than the bulk analysis, but more than three times as high as the calculated initial value, using its Re/Os elemental ratio. These results suggest that the leach procedure leads to the retrieval of some, but not all, of the radiogenic 187Os produced in situ by Re disintegration. Leachates from the oxic section should thus provide a better hydrogenous-Os fraction sampling, considering its low Re/Os ratio. Given that values of the leached and un-leached measurements in the marine section all fall unto the range of the oceanic evolution, we believe the bulk digestions of the oxic section represent the hydrogenous osmium signature.
 Re (and Os) contents in the ACEX core are very enriched for their Corg (Figure S1), implying that many samples from this study must have been deposited in a low-restriction water-mass setting (i.e., short renewal time of the water mass, thus not limited in terms of metal inputs; see McArthur et al.  for Re-Os system/Corg linkages), and thus might yield some reliable Re-Os geochronological information. However, our sampling is not optimal for dating purposes, as we have only a single sample for a given stratigraphic level. Nevertheless, some adjacent levels (thus of close depositional age) below the hiatus and three separable discrete contiguous laminas from the same level (just above hiatus) do yield acceptable linear isochrons (Figure S2).
 Data from sample 199.8 mcd can be fitted through a regression line (age = 38.59 ± 0.82 Ma; Initial 187Os/188Os = 1.19 ± 0.05; MSWD = 0.23). This apparent age is ∼6 Ma younger than the stratigraphic age estimate of 44.4 Ma at 198.7 mcd from Backman et al. , for the lower limit of the hiatus. However, since “diatoms may suggest a late middle to early late Eocene age in the 202.5–203.5 mcd interval” [Stickley et al., 2008; Backman et al., 2008], our ∼39 Ma age at 199.8 mcd does not seem out of scope. Calculation of initial Os isotopic values for these samples using an age of 44.45 Ma yields 0.87 to 0.68, compared to 1.19 when using the age of 38.59 Ma (the y-intercept of the isochron). However, we agree that this age remains a simple 3-point isochron.
 A second isochron with a surprisingly similar age was obtained for sediments at 196.5 mcd (age = 38.1 ± 2.3 Ma; Initial 187Os/188Os = 0.59 ± 0.09; MSWD = 13, Figure S2), just overlying the hypothesized hiatus. This regression is much older than the presumed Burdigalian age assigned to the sediment, based on the presence of new genus of dinocysts Arcticacysta, (16–20 Ma, from weak or at least open to discussion biostratigraphic inferences [Sangiorgi et al., 2008; Jakobsson et al., 2007]). Accepting our Re/Os age estimates would essentially close the gap in the ACEX core and, thus modify significantly the acknowledged chronology. On the contrary, if one accepts the Burdigalian age of this interval [Backman et al., 2008; Sangiorgi et al., 2008], our result would then indicate some mixing between multiple sources of Os at the time of deposition brought by surface waters that would dominate the very low-Os euxinic deeper water [McArthur et al., 2008]. In view of the huge and diverse watershed of the area during the late Eocene (i.e., Precambrian North-American crust, Phanerozoic Asian and European crust), this possibility cannot be totally discarded. In this unlikely scenario, the data set yielding an age about 20 Ma in excess vs. the stratigraphic time scale, would not reflect a true isochron, but rather a ∼18 Ma-old mixing line of two components (or more, considering the scatter).
4.2. Paleoenvironmental Implications of Os-Isotope Data
 The fact that the deep, reduced core section has very radiogenic initial Os values suggests continental sources as major inputs of Os. This interval corresponds to a lacustrine stage in the Arctic basin [Jakobsson et al., 2007], with little or no connection to open ocean basins. During this interval, oceanic floor expansion was limited in the Arctic (thus with reduced hydrothermal alteration of unradiogenic seafloor, if any). Overprinting of inherited radiogenic Os signatures by fresh-water mass [McArthur et al., 2008] may easily account for the ACEX lower section data. Moreover, the temperature during the Paleocene/Eocene was much higher than today (more than 23°C in the Arctic, during the Paleocene/Eocene Thermal Maximum [Sluijs et al., 2006]). An intensified hydrological cycle with precipitation exceeding evaporation has been suggested, especially at high latitudes [Zachos et al., 2001; Brinkhuis et al., 2006]. Such factors should have enhanced weathering rates over surrounding continental landmasses, which include many old cratons, and thus high terrestrial radiogenic-Os (and Re) inputs to “Arctic lake” waters, during this period. Limited exchange with open ocean (for example through Turgay Strait [Brinkhuis et al., 2006]) could also be involved during the Eocene, and account for the lowest isotopic ratios recorded at 45–46 Ma, whereas global seawater had a 187Os/188Os ratio of 0.5–0.6 at that time (Figure 2).
 When using the ACEX age model from Backman et al. , two environmental transitions are apparent from our dataset, above and below the 26 Ma hiatus. The strong enrichments in Re and Os and the elevated 187Os/188Os ratios observed during these transitions would thus have to be interpreted in relation with the paleogeography of the Lomonosov Ridge [Sangiorgi et al., 2008]. This hiatus, with non-deposition and likely some erosion, would be relating to a tectonic shoaling of the ridge. According to Sangiorgi et al. , the sediment just below the hiatus indicates enhanced current velocity and/or shallow environments and depositional conditions, at or near sea level, with minimal transport offshore by currents. The presence of a “drop-stone” reported by reported by Moran et al.  in these layers, would thus indicate inception of winter sea-ice with ice-floes spreading local material, as illustrated by Dionne  in modern environments.
 However, a more plausible explanation, if we admit that the two Re/Os age estimates are more reliable than the existing biostratigraphic data surrounding the 198.7 mcd “event”, would simply involve a major drainage event at the opening of Fram Strait, some 38 Ma ago, or, a major flood, depending on the topography of the paleolake infilling the Arctic basin prior to this event. In this scenario, one can recalculate all initial Os ratios (Table S1) using the corresponding revised age model, which puts off a linear sedimentation rate of 1.75 m/Ma between 151.3 mcd, the deepest 10Be age, and 196.5 mcd, the Re/Os age. The output, shown on Figure 2 (right), is an Arctic sedimentary sequence that if fully marine since ∼38 Ma, suggesting an inception of North Atlantic waters ca. 20.5 Ma earlier than previous estimate. Near-future measurements of samples in the 150–190 mcd interval will allow us to strengthen this hypothesis.
 The full incorporation of the Arctic Ocean into global oceanic circulation at either 17.5 Ma or more likely 38 Ma, should have conversely resulted in some input of radiogenic-Os into the North Atlantic (as radiogenic as the deep ACEX sediments initial values). However, this event seems unrecorded based on current literature. Two reasons might be evoked: 1) the radiogenic-Os spike from the Arctic was too small to be recorded or 2) current sampling resolution of available record is too low to reveal a transient excursion potentially associated with a short drainage/flood event and the subsequent Arctic ventilation. However, considering the anoxic conditions in the Arctic basin during its early stages, the residence time of Os must have been short, and dissolved Re-Os must have been scavenged and buried very rapidly, thus removing rapidly these elements from Arctic waters.
 When sedimentation resumed, following the opening of Fram Strait, oxygenated waters invaded the Arctic Basin, thus leading to depositional conditions drastically distinct from the earlier ones. The transition to fully oxic conditions occurs at the base of sub-unit 1/4, at 192.94 mcd (dated at ca. 17.5 Ma by Jakobsson et al.  and Backman et al. , and about 20 Ma earlier based on Re-Os data), where the sediment color changes from grey to brown. Examination of Os-isotope data suggests that the “marine” signature (related to linkage with the North Atlantic) occurred slightly before, as isotopic ratios in the underlying sub-unit 1/5 (down to 194.44 mcd) already resembles that of the global ocean trend (Figure 2, left), thus supporting the assignment of the 198.7–196.5 mcd interval to a drainage/flood layer (where the presence of coarse material would not then necessarily indicate ice-rafting deposition!).