1.1. Problems With the Traditional Approach
 Traditionally, fluxes of material to the seafloor have been estimated from mass accumulation rates (MAR) between dated sediment horizons [e.g., Lyle and Dymond, 1976; DeMaster, 1981; Curry and Lohmann, 1986; Rea and Leinen, 1988; Sarnthein et al., 1988; Lyle et al., 1988; Mortlock et al., 1991]:
where ρdry is the dry bulk density of the sediment (g cm−3), LSR is the linear sedimentation rate (cm kyr−1), z2 and z1 are the depths of sediment horizons 2 and 1 (cm), and t2 and t1 are the corresponding sediment ages (kyr), generally obtained from 14C dating or tie points in the δ18O stratigraphy.
 Mass accumulation rate of a specific sedimentary constituent (MARi) is then obtained simply by multiplying the total mass accumulation rate by the concentration of the constituent of interest. The latter should be averaged over the entire core section between z1 and z2 ([i]z1z2).
It is important to realize that variations in [i] between z1 and z2 should not be multiplied by the sedimentary rate interpolated between these boundaries to calculate variations in flux, since it cannot be ascertained that the sediment accumulation rate remained constant over the depth interval considered.
 Although still widely used, this approach has several major shortcomings: First, the temporal resolution that can be obtained by this procedure is inherently limited by the need to take the difference between two measured ages. Increasing the resolution of the flux record requires dating sediment horizons that are in closer proximity, which automatically increases the relative error on (t2 − t1) and on MAR estimates. Second, MAR calculations rely on knowledge of the dry bulk density of the sediment (ρdry), which is estimated by sampling and drying known volumes of sediment or by establishing linear or polynomial fits with carbonate concentration [Lyle and Dymond, 1976; Curry and Lohmann, 1986; Froelich et al., 1991]. Either of these approaches is associated with significant uncertainties. Third, and maybe most importantly, MARs thus calculated do not distinguish between the contribution from vertical fluxes originating from the overlying waters, and lateral fluxes resulting from sediment redistribution by bottom currents. Failing to take into account horizontal supply or removal can be very misleading if, as is often the case, MARs are interpreted in terms of fluxes originating from the overlying surface water [e.g., Francois et al., 1993; Pondaven et al., 2000; Marcantonio et al., 2001a; DeMaster, 2002].
1.2. The Incidence of Sediment Redistribution on the Seafloor
 Sediment redistribution can be either postdepositional or syndepositional. Postdepositional redistribution is due to erosion, transport and redeposition of sediments that have been initially deposited at an earlier time. This is relatively rare and can generally be recognized from disturbed stratigraphy and chronology, clearly ruling out an affected core for reconstruction of fluxes originating from surface water. On the other hand, syndepositional redistribution describes a situation in which resuspension, lateral transport and final deposition occur simultaneous to or soon after initial deposition. Syndepositional redistribution is not always readily identifiable, because the integrity of the sediment stratigraphy and chronology is maintained. As a result, it is often overlooked.
 Sediment redistribution on the seafloor occurs on a variety of scales. It can result in large and conspicuous drift deposits (e.g., Feni and Gardar drift in the NE Atlantic; Bermuda Rise in the NW Atlantic; The Meji drift in the NW Pacific), which are often targeted by paleoceanographers because of their high sediment accumulation rates and the high-resolution records they provide. Evidently, the accumulation rates of sediment constituents measured at these sites are not used to reconstruct paleofluxes from the overlying water. However, there is increasing evidence that syndepositional redistribution can occur much more commonly, but more subtly so that it is not so easily recognized.
 Recently deposited surficial sediments and particularly phytodetritus that accumulate on the seafloor can be readily redistributed by relatively weak bottom currents [Beaulieu, 2002]. Particle fluxes measured with sediment traps deployed within the bottom boundary layer show evidence for particle “rebound” (i.e., resuspension of particles that have reached the seafloor but have not become incorporated into the sediment; Walsh et al. ). Photographic evidence shows that phytodetritus accumulates preferentially in small depressions. It can thus be readily surmised that the interaction between bottom currents and small to mesoscale and large topographic features on the seafloor could often and significantly affect accumulation rates without impairing the stratigraphy or chronology of the sedimentary sequences. This is corroborated by high-resolution seismic reflection profiles of the seafloor that often show wide variations in the thickness of layered sediment deposited in the same area. Among many examples, a spectacular illustration of this process was recently reported by Mollenhauer et al.  (Figure 1). The tendency for pelagic and hemipelagic sediments to be winnowed preferentially from topographic highs and deposited preferentially in lows, thus tending to smooth out, over time, underlying roughness in basement topography, has been known since the early days of marine seismic reflection profiling. The process of “syndepositional redistribution” and its effects are well described in the following quotation from Ewing et al. [1964, p. 17] summarizing their observations from crossings of the Mid-Atlantic Ridge: “On the northern and middle crossings, the sediments are mainly in pockets, and intervening areas are almost or entirely bare. A large percentage of the pockets have almost level surfaces. These facts suggest that the sediments deposited on the ridge flow easily after reaching the bottom here. Where impounded, the ridge sediments apparently develop cohesiveness and will not flow easily if subsequently tilted.” An excellent illustration of this sediment “ponding” effect as it occurs on the flanks of the Mid-Atlantic Ridge can be found in the review by Ewing and Ewing [1970, Figure 6]. The distribution of sediments on the Mid-Atlantic Ridge may be an especially dramatic example, but we would submit that it illustrates a process that is widespread, perhaps ubiquitous, in the ocean.
 Syndepositional redistribution appears to be particularly common and important in regions with a dynamic bottom water circulation (e.g., Southern Ocean, northern and western Atlantic, equatorial Pacific), where the accumulation rates of laterally redistributed sediments are often many times larger than the fluxes of material sinking from the overlying surface waters [e.g., Suman and Bacon, 1989; Francois et al., 1993; Frank et al., 1999, 2000; Dezileau et al., 2000; Marcantonio et al., 2001a; P. Loubere et al., Lower biogenic fluxes in the eastern equatorial Pacific at the last glacial maximum: Reconstruction of calcite paleo-fluxes, submitted to Paleoceanography, 2003, hereinafter referred to as Loubere et al., submitted manuscript, 2003]. MARs measured by the traditional method in these regions thus often can seriously overestimate the actual settling rates of material from the overlying water column. This problem is compounded by coring biases. The need for high temporal resolution in paleoceanographic reconstructions encourages collection of sediment cores in areas of high accumulation rates, which is achieved with increasing success as surveying and positioning techniques improve. These areas are mostly zones of preferential deposition of redistributed sediment (or sediment focusing). This causes a bias in the available core collections that has resulted in significant overestimates of the sedimentary sink in budget calculations based on MARs [e.g., Pondaven et al., 2000; DeMaster, 2002], a bias that (again) was well recognized as early as 1964: “New measurements of carbonate deposition are needed in areas of draped sediments, rather than in pockety areas, to establish reliable rates of sediment accumulation” [Ewing et al., 1964, p. 34]. It is thus imperative that a means of identifying and correcting for lateral sediment redistribution be routinely applied, if one is to use correctly the sedimentary record to quantify past variations in particle rain from surface waters.
 Bacon and Rosholt , Bacon , Suman and Bacon , and Francois et al.  have proposed and developed a method based on normalization to 230Th to quantify syndepositional redistribution and estimate vertical fluxes from the sedimentary record. A further advantage of this approach is that each 230Th measurement provides a flux estimate. Thus, unlike the conventional approach of estimating MARs, there is no trade-off between the resolution and precision of the flux estimate, and the resolution at which flux variations can be recognized is limited only by bioturbation. In addition, dry bulk density does not intervene in the flux calculation by 230Th normalization, thereby eliminating yet another source of error.
 Even though the approach was first proposed 20 years ago and its use has been increasing in recent years, its application is still far from universal, and many recent studies still use MARs to estimate vertical fluxes from late Quaternary sedimentary records [e.g., Ikehara et al., 2000; Thomas et al., 2000; Latimer and Filipelli, 2001; Lyle et al., 2002; Hyun et al., 2002]. Slow acceptance of the 230Th normalization approach by the paleoceanographic community has been in part the result of contradictory views in the earlier literature regarding the validity of the method's key underlying assumption. Bacon  argued that the flux of 230Th scavenged to the seafloor must always be nearly equal to its known production rate in the water column, and thus could be used as a reference to estimate the settling flux of sedimentary particles. However, this view was not universally accepted, particularly during the initial stages of development of the method, and it has been also argued that the flux of 230Th to the seafloor could considerably exceed production in the water column in regions of high particle flux [Mangini and Diester-Haass, 1983; Scholten et al., 1990; Thomas et al., 2000]. Wider use of the approach may have also been discouraged by the slow sample throughput of the alpha-spectrometric method traditionally used for 230Th analyses and by the need for somewhat specialized equipment.
 These obstacles have now been overcome. Our present understanding of the behavior of 230Th in the water column [Rutgers van der Loeff and Berger, 1993; Scholten et al., 1995, 2001; Moran et al., 1997, 2002; Vogler et al., 1998; Yu et al., 2001; Chase et al., 2003] and recent modeling studies [Henderson et al., 1999; Marchal et al., 2000] provide a stronger basis for asserting the validity while also recognizing the limits of the 230Th normalization approach. In addition, the development of Inductively Coupled Plasma Mass Spectrometry (ICP-MS) has dramatically improved sample throughput and has replaced alpha-spectrometry for 230Th measurements in sediments [e.g., Shaw and Francois, 1991; Hinrichs and Schnetger, 1999; Choi et al., 2001; Shen et al., 2002]. The widespread use of this innovative and versatile analytical technique is such that essentially every modern geochemical laboratory now has the wherewithal to measure 230Th in sediments. The convergence of improved understanding of the geochemical behavior of 230Th in the ocean and improved analytical capabilities sets the stage for the widespread use of 230Th normalization in late Quaternary paleoceanographic studies.