The Rock Magnetic Properties, Grain Size and Mineral Composition of Winborne Dust and Sediment in the North Pacific Ocean

Dust, uplifted by wind from the continents, and transported through the atmosphere, leaves a geologic record across the earth; in loess deposits on the continents, red clays on the ocean floors and on the polar ice caps. If we can interpret the paleoclimatic and paleometeorology information preserved in these deposits, we can learn how continental climate and atmospheric circulation have varied over the course of time. Continental climate information is preserved in the composition of the dust. The mineral phases that comprise the surface of the continents are dictated by the geology of the parent rocks, but more importantly, the soils formed by the weathering of these continental rocks are extremely sensitive to the climate variables precipitation, temperature and seasonality. Records of atmospheric circulation processes are preserved in the spatial distribution patterns, flux and particle size of the deposited eolian material. In order to exploit the global paleoclimate and paleometeorology records, the relationships between the continental dust source areas, the transport process and the resulting deposits must be quantified. In this work, sediments and aerosols from the North Pacific Ocean are studied. The North Pacific contains the most spatially and temporally contiguous record of eolian material, transported from the deserts of northern Asia by the zonal westerly winds for millions of years. Aerosols, collected from research vessels in the North Pacific, surf ace sediments from across the entire ocean basin, and a sediment core from the central North Pacific were analyzed for rock-magnetic properties, grain size and mineralogy. This study provides a data set of geological measurements that are directly related to atmospheric processes, recent sedimentation, and eolian sedimentation over the last 8 million years. The aerosol samples record both the source region and transport history of the continental dust. Atmospheric dust concentrations are highest for those samples with the shortest transport time from Asia to the open ocean. Asian dust samples are characterized by high dust concentrations, fine grain size, and high concentrations of 2-20μm quartz and <2μm kaolinite. High latitude, Aleutian/Alaskan dust is characterized by low dust concentrations, coarse grain size and is relatively enriched in plagioclase and magnetic material. The aerosol is compositionally fractionated during the transport process, becoming relatively enriched in clay minerals at the expense of primary minerals. The surface sediments from the North Pacific preserve the relationships between transport process and physical characteristics observed for the aerosols. The rock magnetic properties, grain size and mineralogy of the aerosols are the same as the eolian surface sediments. The sediments display a steady decrease in the grain size across the entire basin, and the composition is fractionated towards a higher coercivity, and a plagioclase-depleted and kaoliniteand chloriteenriched composition with increasing distance from the source area. The eolian dust preserved in the down-core sediments records the onset of major eolian sedimentation to this region 3.8 million years ago. When the flux increased, the rock magnetic grain size increased, the composition of the minerals shifted from a kaolinite-enriched mineralogy to a chlorite enriched mineralogy, suggesting acidification of the source region and acceleration of atmospheric transport.

Precision, expressed as percent relative error of mineral peak areas.
Stratigraphic parameters and mass accumulation rates for Holes 885A and 886B.
Rock-magnetic measurements for Holes 885A and 886B.
Normalized mineral peak areas for the <2 µm fraction of Holes 885A and 886B.     Holes 885A and 886B in the North Pacific Ocean are largely eolian in origin Leinen, 1989;; thus, variations in the properties of these sediments are related to variations in atmospheric transport patterns. The present upwind source region for these eolian sediments is Asia, 1 / i \ and the sediments record both past climatic variation and tectonic activity from this continent. In addition, the terrigenous material may have been subjected to sediment reworking and diagenetic alteration. By examining a variety of parameters, we can attempt to deconvolve this climate-tectonic-transportpreservation signal into its individual components and learn something about the variation of each through time.
In this paper, we examine the downcore variation in mineralogy and rockmagnetic measurements to infer the paleoenvironmental conditions responsible for sedimentation proximal to Sites 885 and 886. The composition and concentration of terrigenous material preserved in deep-sea sediment are related to weathering, transport, and depositional processes as well as postdepositional alteration of continental material. Mineralogical composition is related to parent-rock composition as well as regional climate. Paleoclimatologists and atmospheric scientists have established that the mineralogy of present-day aerosols is related to the source area (M. Leinen et al., unpubl. data; J. Merrill et al., unpubl. data) and that the mineralogy of aerosols collected over the North Pacific is the same as that of surface sediments from eolian deposits . Furthermore, downcore variation in mineralogy has demonstrated a consistent relationship with other paleoclimate proxies and has allowed paleoclimatologists to interpret variation in past climates in the Asia-North Pacific region (Leinen, , 1989Schramm, 1989).
In an analogous manner, rock-magnetic studies have also been used to trace atmospheric samples to their source area  and to infer changes in atmospheric input to deep-sea sediments through time . In addition to compositional information, rock-magnetic properties are also useful proxies for the concentration and grain size of I ~ terrigenous deep-sea sediments. Finally, the rock-magnetic properties are very sensitive to reduction diagenesis, so that we can control for postdepositional alteration of the sediment.
The rock-magnetic analyses are nondestructive, so we were able to analyze the same samples for mineralogy. Because the mineralogy of atmospheric aerosols has been identified with broad regional source areas, we can use this information to interpret the mineralogy signal in deep-sea sediments if we can confirm that the original mineralogy has not been modified by transport, deposition, or postdepositional modification. Bulk mineralogy can be altered by transport and deposition because mineralogy varies with grain size. As material travels through the atmosphere and falls through the water column, large particles fall out more quickly than the fine fraction. In addition, winnowing of deep-sea sediments can result in size fractionation. The coarse fraction of terrigenous material is relatively concentrated in primary minerals such as quartz and plagioclase, whereas the clay minerals dominate the fine fraction. We control for this by examining the mineralogy in separate size classes in the deep-sea sediments. Diagenetic alteration may be discemeq by comparing the mineralogy record with the rock-magnetic properties; divergence of these two signals may indicate either a postdepositional modification of the sediment or a change in source.

Methods
Samples were collected at 150 cm intervals from Holes 885A and 886B with an aluminum sampling tool designed to collect undisturbed sediment; they were then extruded into 5-cm3 plastic cubes for rock-magnetic analyses. In addition, ODP paleomagnetic samples (7 cm3), collected at 150-cm intervals, 3 were also used. Sample spacing for the rock-magnetic study is approximately 75 cm, and 150 cm for the mineralogical analyses.
Susceptibility (Xlf, Xhf) was measured on a Bartington Instruments susceptibility meter at 0.47 and 4.7 kHz. The reported value (in 10-6m3/g) is the average of three replicates. The samples were demagnetized in a 100-mT alternating field. Anhysteretic remanent magnetization (ARM) was induced in a 0.1-mT steady field superimposed on a 100-mT alternating field. The ARM was measured on a cryogenic magnetometer, and the reported value (in 10-6 Am2/g) is the average of duplicate measurements. The Xarm was calculated by dividing ARM by the steady field (reported in 10-6m3/g). The final set of magnetic measurements were saturated isothermal remanent magnetization (SIRM) and isothermal remanent magnetization (IRM-0.3T)· Samples were saturated in a 1. and their applications may be found in the "Results" section (this chapter).
Readers unfamiliar with rock-magnetic techniques are referred to .
Samples were freeze-dried, weighed, and wet sieved at 63 µm; the >63 µm and <63 µm were then dried and weighed. The <63 µm fraction was treated to remove biogenic silica using a NaOH procedure (see Snoeckx, this volume). The NaOH procedure removes more amorphous material than other common extraction techniques, but does not alter the relative proportion of the various mineral 4 phases. Iron oxides were removed using the oxalic acid extraction technique of . Sediments were saturated with MgCl2, to reducedspacing variability caused by cation differences; they were then rinsed with warm deionized water, dried, and weighed. Sediments were wet sieved at 20 µm with the aid of a sonic dismembrator, and the <20 µm fraction was split into 2-20 µm and <2 µm size fractions by means of centrifugation. The 2-20 µm size fraction was spiked with a 10% by weight AI203 internal standard and the <2 µm fraction was spiked with a 10% talc internal standard. Samples were homogenized by grinding in a mortar and pestle under acetone, air dried, suspended in a deionized water slurry, and drawn onto duplicate (one air-dried, one glycolated) silver filters for X-ray analysis.
The X-ray analysis was run from 2° to 30°20 at 45 kV and 40 mA at 2° 20/min using Cu-ka radiation. Peak areas for smectite, illite, kaolinite, chlorite, quartz, and plagioclase were determined using Scintag DMS software. Precision of the peak areas is listed in Table 1. These precisions were calculated by comparing the area under the mineral curves between the glycolated and unglycolated samples. Smectite cannot be compared in this manner. Replicate scans and peak area resolution for this mineral were performed; the error for smectite determination is high (in excess of -30% for both size fractions). The relative proportion of kaolinite and chlorite was determined by the relative proportions of the kaolinite [002] and chlorite [004] peak areas. Mineral peak areas were normalized to the internal standard peak areas.
Sites 885 (44°4l'N, 168°16'E) and 886 (44°4l'N, 168°14'E) were drilled at about 5700 m water depth in the North Pacific red clay province. A total of 59 m of sediments were cored at Hole 885A and 69 m at Hole 886B. Three stratigraphic sedimentary units were recognized at each site: (I) Pleistocene to late Pliocene dd . h brown to brown red clay; (II) late Pliocene to late Miocene diatom ooze re1s / / and (Ill) a lower bro~layUi~h, originally dated as late Miocene (Rea, Basov, Janecek, Palmer-Julson, et al., 1993), but now recognized as older; late Miocene to late Cretaceous in Hole 886C (Ingram, this volume). Table 2 lists the stratigraphic parameters for Holes 885A and 886B.
Correlation between and composite depth of Holes 885A and 886B is detailed in Dickens et al. (this volume); all depths in this report are composite depths derived from that model. The depth intervals (in meters composite depth) for Unit I are 0-23.5 med; for Unit II, 23.5 to -52 med, and for Unit III, >52 med. The age models used in this report are derived from the magnetic reversal stratigraphy listed in Dickens et al. (this volume). These ages are in agreement with the Radiolaria stratigraphy reported by Morely (this volume). Ages derived from ichthyolith strontium isotopic ratios (Ingram, this volume) indicate slightly younger sediments in the lower Unit II samples, but older sediment in the Unit Ill sediments than in the magnetic age model. Hiatuses and low sedimentation rates explain the discrepancy between the Unit III ages. Because the radiolarian ages agree with the paleomagnetic estimates, the Sr data were not incorporated into the age model for this study.
The sedimentation rate is estimated by linear interpolation between each magnetic datum obtained from the composite depth model. It is important to recognize that the linear sedimentation rates (LSRs) derived from the composite depth section may be higher than those derived from the original cores. This discrepancy is cause for some concern when trying to estimate mass accumulation rates (MARs), as the seismic records indicate that the Site 885 sediment is indeed thinner than Site 886 (Rea, Basov, Janecek, Palmer-Julson, et al., 1993).
Most of the difference appears to occur in the lower brown clay of Unit Ill. The  Results Figure 1 illustrates the rock magnetic parameters plotted vs. composite depth, and Table 3 lists the magnetic measurements for Holes 885A and 886B.
The low-frequency susceptibility (Fig. lA) is primarily a measure of the concentration of magnetic iron oxides and is used as a proxy for the concentration 7 / \ of terrigenous material. Susceptibility is highest in stratigraphic Unit III, the lower brown clay unit, suggesting a large iron-oxide component in these sediments, which grades monotonically to low values in the lower part of Unit II, the diatom ooze (-46 med). Susceptibility remains low throughout the diatom ooze and increases at the base of Unit I, the upper red clay unit (-24 med). There are largeamplitude peaks superimposed on the general upcore increase in susceptibility, coincident with the ash layers described in the initial reports (Rea, Basov, Janecek, Palmer-Julson, et al., 1993).
The frequency dependence of the susceptibility (Xhf/X1f) is used to identify the presence of very fine (submicrometer), viscous, superparamagnetic grains.
Should the contribution of such grains be large (low Xhf/Xlf ratios), the susceptibility concentration proxies and grain-size parameters derived from these values will no longer be related to the concentration and grain size of terrigenous materials, as superparamagnetic materials make a disproportionately large contribution to the susceptibility. Figure lB illustrates the frequency dependence of the susceptibility; the frequency dependence in the lower and upper red clay units is relatively uniform and low, indicating that. the application of the magnetic proxies for terrigenous material is appropriate in these intervals. The diatom ooze unit displays a more variable frequency dependence signal, largely because of the low concentration of terrigenous material combined with a negative contribution of the biogenic silica to the susceptibility signal, resulting in a frequency ratio of two very small numbers. However, there does not appear to be a large viscous superparamagnetic contribution in this stratigraphic unit; again, the application of the magnetic parameters as terrigenous proxies is appropriate.
Anhysteretic remanent magnetization, expressed as Xarm (Fig. lC), is another measure of iron-oxide concentration, but it is also affected by the domain 8 state of the magnetic minerals. The domain state of magnetic material is related to the iron-oxide particle grain size, which, in tum, may be related to the terrigenous grain size, if fine magnetic particles are not randomly incorporated into larger terrigenous grains. The iron-oxide grain size is related to the terrigenous grain size because magnetic particles such as magnetite and hematite, derived from continental source areas, are subject to the same transport processes as the alumina-silicate grains. Smaller magnetic grains yield a larger Xarm signal. The Xarm mimics the upcore pattern observed for the susceptibility in the lower clay and diatom units, but reaches a broad maximum between 10 and 20 med in Unit I before decreasing toward the top of the core. This decrease at the top of the core indicates either a decrease in concentration of magnetic material, in conflict with the susceptibility signal, or an increase in the magnetic grain size. Examination of the ratio Xarm/Xlf (Fig. ID) indicates that the latter explanation is more likely.
The ratio Xarm/Xlf is inversely related to the ferrimagnetic iron-oxide grain size and is independent of magnetic concentration. The Xarm/Xlf ratio monotonically increases from the base of stratigraphic Unit III to the base of stratigraphic Unit II, indicating a general ferrimagnetic-iron-oxide grain-size decrease throughout the lower brown clay unit. The ratio abruptly increases and becomes more variable at the base of Unit IL The high variability in this unit is related to the low magnetic mineral concentration in the diatom ooze; however, a few comments may be made about the overall grain-size pattern. In general, the diatom ooze contains the finest grain size of all sediments from Holes 885A and 886B. There is an interval of low Xarm/Xlf between 34 and 30 med, indicating a relatively coarse magnetic grain size, and a sharp peak in the grain-size ratio at -28-30 med, indicating a very fine magnetic grain size. Above -28 med, the grainsize ratio decreases and becomes less variable throughout the remainder of the 9 . ooze and in all of stratigraphic Unit I. This pattern of variation suggests a diatom al increase in the ferrimagnetic grain size toward the top of the core. gener The S-ratio (IRM-0.3TISIRM1.2T) is a parameter sensitive to the composition of magnetic minerals. Magnetic minerals such as magnetite are easily magnetized, whereas minerals such as hematite require stronger magnetic fields to saturate them. Thus, the ratio of the proportion of magnetization stripped off of the magnetically saturated sample in a reverse field of 0.3 T is related to the proportion of "hard" to "soft" magnetic material present in a sample. Figure IE shows the S-ratio for Sites 885 and 886. The small-amplitude, high-frequency variability in Hole 885A sediments is a systematic laboratory error; one batch of samples was not completely saturated. The S-ratio is relatively uniform and high in the lower brown clay unit and in the lower part of the diatom unit, indicating a constant magnetic mineral composition throughout this interval. Low S-ratios are observed from 29 to 35 med; this is the same interval as the low Xarm/Xlf ratios in the diatom ooze. The concentration of hard magnetic minerals (such as goethite or hematite) does not explain the large magnetic concentrations indicated by the susceptibility or Xarm signal in Unit III. Thus, the magnetic carrier in these sediments must also include a softer component, such as magnetite. The concentration of hard-coercivity material begins to increase at the base of the uppermost red clay unit and displays a high-amplitude variability superimposed on ral HIRM increase throughout this unit. This increase in HIRM tracks the a gene increase in susceptibility; thus, the magnetic carrier in this unit is harder than the sediments in the lower brown clay unit.
Figures 2 and 3 illustrate the mineralogy for the <2 µm and 2-20 µm rerrigenous material at Holes Sites 885A and 886B. Mineralogy was run at twice the sampling interval as the rock-magnetic measurements. A study to derive weighting factors for the 2-20 µm size fraction has not yet been completed, so normalized peak areas are presented instead of absolute weight percent (Tables   4-5). The absolute values of the peak areas are only comparable for the individual mineral within each size fraction. Relative change is directly comparable among all minerals and size classes.
Smectite is produced by continental weathering processes, weathering of volcanogenic material, and authigenic formation in the sediment column; so the interpretation of this mineral group is complex. However,  states that most of the smectite in pelagic deep-sea sediments may be interpreted as continental weathering products. The concentration of smectite is relatively uniform in the <2 µm size fraction in both cores and through all units ( Fig. 2A).
For the 2-20 µm size fraction (Fig. 3A), smectite shows about a threefold increase in peak area between approximately 25 and 35 med and is relatively uniform and present in equal amounts in both stratigraphic clay Units I and III.
Illite is a ubiquitous terrigenous weathering product, generally associated with cool, dry environments, that is unlikely to form authigenically in the relatively thin sediments near Holes 885A and 886B. For both size fractions, there is a relative increase in illite weight percent centered on about 30 med (Figs. 2B and 3B). The magnitude of the increase is largest in the 2-20 µm size fraction and 11 smallest in the <2 µm size fraction. The <2 µm size fraction demonstrates the greatest variability in illite throughout all depths, and there is some suggestion of an illite increase in the upper 10 m of the Unit I red clay. The 2-20 µm size fraction has uniform and approximately equal illite concentrations in the upper and lower clay units.
Kaolinite is a terrigenous weathering product usually associated with strong hydrolysis. The error associated with kaolinite determination in this study is large due to the generally low kaolinite concentration in this study area, so only general comments may be made about variations in this mineral group. Kaolinite is present in highly variable amounts in all units for the <2 µm size fraction (Fig.   2C). The 2-20 µm size fraction contains a significant, but highly variable amount of kaolinite in only the diatom ooze unit (Fig. 3C).
Chlorite is a mineral phase that is usually used to indicate mechanical weathering of terrigenous sediments as it is highly susceptible to hydrolysis. The <2 µm fraction (Fig. 2D) shows low variability and low concentrations in the lower brown clay unit and high variability in the diatom unit and the upper red clay unit. Chlo rite concentrations increase in sedimentS shallower than 10 med. The chlorite concentr_ ation in the 2-20 µm size fraction (Fig. 3D)  Kaolinite to quartz (Figs. 4C and SC) is the ratio of a hydrolysis sensitive mineral to a primary mineral. In both size fractions, K/Q is generally low in Unit III, reaches a maximum at about 30 med, and decreases toward the top of the core.
This variation is more pronounced in the 2-20 µm size fraction, where K/Q goes to zero at the Unit I/II boundary.
The illite/quartz (Figs. 4D and SD) ratios display decreasing values in the <2 µm fraction from the bottom to the top of Unit III; l/Q is relatively uniform in the 2-20 µm fraction of this interval. In both size fractions, the lower part of Unit II displays uniform ratios; there is a peak in this ratio centered at about 30 med. This peak is much more pronounced in the 2-20 µm size fraction. The ratio decreases slightly in both size fractions from the bottom to the top of the upper red clay unit.
In both size fractions, chlorite/quartz ratios (Figs. 4E and SE), indicative of mechanical weathering, are generally very small in lower clay Unit III, show a slight increase at the Unit 11/111 boundary, and another small step up at -4S med.
There is a sharp peak in this ratio centered at approximately 30 med, followed by uniform values throughout the remainder of the diatom unit and Unit I.
Finally, we compare the rock-magnetic properties to the mineralogy, to examine how the two covary. Figure 6 illustrates the low-frequency susceptibility and the <2 µrn quartz peak area plotted vs. composite depth. The two variables covary in the upper red clay and diatom ooze units. However, the signals decouple at the Unit 11/111 boundary. This decoupling of the signal implies that the sedimentary material in Unit III is from a different source than the terrigenous material in Unit I. Figure 7 illustrates the illite/quartz ratio and the S-ratio plotted as a function of composite depth. These variables strongly correspond at about 30 med.
Illite/quartz increases occur at the same place that the rock magnetic parameter, S, indicates a shift from low to high coercivity iron-oxide mineralogy.  There is offset in the terrigenous MARs of Holes 885A and 886B in the upper clay unit. This could be the result of either systematic laboratory error, sediment winnowing, or focusing differences between the two sites. Because there are no systematic off sets in the primary rock magnetic parameters, or in the bulk MARs, we think the former is the case. Snoeckx and Rea (this volume) provide a detailed interpretation of the eolian accumulation at Sites 885 and 886. For the parameters presented, we display only the upper two units for the mass accumulation discussion; we have few samples from the lower clay unit, and the ages of these sediments are uncertain. Figure 9 displays the magnetic concentration parameters plotted as accumulation rates vs. age. The magnetic-iron-oxide concentration parameters (Xlf and HIRM) track the terrigenous MARs. These records show a small, LSRcontrolled peak at the base of Unit II (-7.5 Ma), low accumulation and low variability through the middle part of this unit, followed by an abrupt increase in accumulation at -3.8 Ma, which continues through all of Unit I. ases in the Unit II diatom ooze in the 2-20 µm size fraction, the <2 µm mere patterns are identical with the larger size class, and are not illustrated. For all minerals, the accumulation rate has small peaks at the base of Unit IT at -7 .5, 6.5 and 5.1 Ma and decreases to low values after 5 Ma. The MARs of the minerals stay low until an abrupt increase at about 3.8 Ma, which continues to increase throughout the late Pleistocene.

Discussion
There are several changes in the mineralogy and magnetic properties that imply varying environmental conditions at Sites 885 and 886. Unit Ill, the lower brown clay unit, is characterized by high concentrations of fine-grained ferrimagnetic iron oxides, that contain a softer magnetic carrier than the Unit I sediments, low bulk and terrigenous MARs, and a mineralogy characterized by a large proportion of smectite relative to the other units. Unit II, the diatom ooze, is marked by low magnetic-iron-oxide concentrations and fine grain sizes, moderate sedimentation rates, low bulk and terrigenous MARs, and an interval of anomalous mineralogy and magnetic composition centered at 30 med. The Unit 1/11 boundary is marked by a step up in the terrigenous MAR and magnetic-iron-oxide concentrations. These increases continue to the top of the core and are accompanied by an increase in the iron-oxide grain size, as well as an increase in the chlorite concentrations.
The potentially important sediment sources proximal to Site 885 and 886 now and in the past include eolian terrigenous and volcanogenic sediment, siliceous biogenic material, and hydrothermal precipitates. The rock-magnetic and . al gy measurements mimic the major lithological changes, which account for m111er o f the variation observed for these parameters. There is close covariance of most o the mineralogy and magnetics in the upper two lithologic units (Fig. 6). The decoupling of the signal at the base of Unit II, specifically a drop in quartz concentration together with a sharp increase in the iron-oxide concentration, occurs at the same time as a shift toward a magnetic composition enriched (relative to the Unit I sediments) in a low-coercivity component. Large concentrations of iron oxides in deep-sea sediments are produced by hydrothermal, terrigenous sedimentary or volcanogenic processes. The sediment in Unit III is likely composed of a large hydrothermal component (see Owen et al. this volume). Hydrothermal sediment is characterized by a wide variety of magnetic carriers, such as magnetite, hematite or goethite, which were observed in shipboard smear-slide analyses (Rea, Basov, Janecek, Palmer-Julson, et al., 1993).
Although the magnetic composition of terrigenous material would depend on the source area composition, terrigenous material from a volcanogenic environment would be enriched in low-coercivity magnetic minerals. We hypothesize that the hydrothermal sediments in Unit III may also contain a small amount of eolian, andesitic, terrigenous material.  propose andesitic volcanism in western Mexico as a source for the Late Cretaceous terrigenous sediments identified in the bottom of Core LL44-GPC3. Plate tectonic calculations indicate that conditions may have been favorable for some andesitic eolian contribution in the past. The clay material in the top two units of the sediment column is likely from an Asian source, as indicated by the present location of the site and the covariance of the rock-magnetic and mineralogy me nts The increase in the importance of this eolian component is measure · ·d .,ed in the context of the climate and tectonic fo11Ces driving the signal. cons1 ei There are two types of tectonic variation that we need to consider to interpret the mineralogical and rock magnetic signals from Holes 885A and 886B.
The first is plate motion, which will influence the source contribution to the sediments. Based on the paleopole reconstruction of Sager and Pringle (1988), Holes 885A and 886B were located only 3° south of the present location at 45°N during the late Miocene deposition of sediments at the Unit II/III boundary (R. Larson, pers. comm., 1994), so plate motion should not complicate the interpretation of the eolian sediments above this interval. However, the basement at Sites 885 and 886 has been dated at -80 Ma (Keller, this volume).
Paleomagnetic plate rotations indicate that the plate was formed at about 16°N.
Thus, we need to control for plate motion when interpreting the Unit III sediments.
Low latitudes are influenced by easterly eolian transport from the North American continent. This supports our hypothesis for some supply of andesitic eolian material to the Unit III hydrothermal sediments.
Tectonic activity in the Asian source area, such as uplift of the Himalayan Mountains or Tibetan Plateau, could impact eolian sedimentation proximal to Sites 885 and 886 as a result of climatic forcing. Geological evidence indicates that uplift pulses are episodic (Copeland et al., 1990;Amano and Taira, 1992;Hovan and Rea, 1993), so the response in the sediment record would be abrupt if the upliftinduced climate forcing is linear. Uplift-forced climatic change (Kutzbach, et al., 1989;Ruddiman et al., 1989) is hypothesized to result in cooling of northern Asia and drying of the Eurasian interior. The impact on the eolian sediments in the North Pacific would be manifested as an increase in the mechanical weathering proxies such as in chlorite, quartz, illite, and plagioclase at the expense of · al weathering proxies such as kaolinite and smectite. This climatic change chemic would also produce an increase in terrigenous MARs, as particle production increased because of source region aridification. At Holes 885A and 886B, the shift from a kaolinite-rich mineralogy in the diatom ooze unit to chlorite-rich sediments in the upper red clay unit, concomitant with an increase in the rerrigenous MAR are supportive of dramatic environmental changes in Asia during the Pliocene.
The timing of this event in this region of the North Pacific is at about 3.8 Ma. Studies of present-day eolian transport identify the deserts of interior Asia as the source region for the loess deposits in China, as well as the North Pacific eolian sediments . A study by Ding et al. (1992) reports the onset of major loess deposition in China at 2.5 Ma. These authors state that the eolian source areas began drying out during the mid to late Pliocene because of plateau uplift. This assertion is supported by the mineralogy in the North Pacific sediments; specifically, the monotonic decline in the concentration of kaolinite in the 2-20 µm fraction of the late Miocene to early Pliocene Unit II sediments.
Although the eolian source regions developed in the late Pliocene, the meteorological conditions were not favorable for deposition on the loess plateau at that time. The mechanism invoked by Ding et al. (1992) for the lack of loess deposition during the late Pliocene involves atmospheric circulation changes, inferred from the model of Kutzbach et al. (1989). The authors state that the 500 mb quasi-stationary trough presently located to the east of the loess plateaus of China was not as well developed during the late Pliocene as it is presently because the Tibetan Plateau elevation was not as high as it is today. Thus, eolian transport from the desert regions was mainly west to east, as opposed to the present northwesterly flow, and the air-mass subsidence over the loess plateau, "ble for dust deposition there, was reduced compared with the present. respons1 These meteorological conditions would not preclude deposition of eolian material in the North Pacific; perhaps they even provided a more direct path between the deserts and the ocean than at the present time. The age offset between the onset of loess deposition in China at 2.5 Ma and the sudden increase in eolian deposition in the North Pacific at 3.8 Ma is consistent with the scenario proposed by these authors. The three small peaks in the MARs of the six mineral species in the lower part of Unit II at about 7.5, 6.5, and 5.1 Ma may also record smaller pulses of Asian aridification in the late Miocene.
Changes in the eolian transport and sediment deposition (current winnowing) in this region would be characterized by changes in the grain size and changes in the LSRs not accounted for by changing source strengths. Magneticiron-oxide grain sizes increase at the base of Unit I, and continue to increase throughout the unit. The increase can be explained by increasing transport speed of atmospheric circulation, caused by an increase in the pole-to-equator temperature gradient induced by glaciation. Snoeckx and Rea (this volume) provide a detailed analysis of this grain size variation at Sites 885 and 886.
Changes in LSR are generally accounted for by the varying strengths of the terrigenous and biogenic accumulation in these cores. There may be hiatuses in lower clay Unit III, but we do not have the sample resolution in this interval to discern this activity.
Another prominent feature of the sediments at Sites 885 and 886 is the sharp signal in the rock magnetic parameters Sand Xarm/Xlf coupled with the sharp change in mineralogy at 30 med . This signal appears in the middle of the diatom ooze unit, during the period with the lowest accumulation of terrigenous sediments in the entire sediment column. The S-ratio and Xarm/Xlf 20 C ould be the result of diagenetic alteration of the magnetic signal. Iron changes .d are soluble under reducing conditions, and the finest grains will be OXI es preferentially dissolved over coarser magnetic particles. The low Xarm/Xlf ratio between 30 and 34 med may be the result of such a grain-size modification. furthermore, the sharp Xarm/Xlf peak at 28-30 med could be caused by singledomain magnetite produced by magnetotactic bacteria active at the redox boundary in the sediment at the time of reduction diagenesis. This hypothesis is corroborated by the low S-ratio in this interval. Fine magnetite is more susceptible to dissolution than hematite, and this selective dissolution has altered the bulk magnetic composition in this interval, reducing the proportion of low-coercivity material. The total organic carbon shows a small peak at this depth (885A, maximum= 0.14%, 886B, maximum= 0.25%; Rea, Basov, Janecek, Palmer-Julson, et al., 1993), although the organic carbon concentrations are extremely low (mean TOC = 0.10%) everywhere in the core. The mineralogical variation could suggest a change in source supply, but it is difficult to envision a process that would supply such a short and intense burst of unusual clay mineralogy in the middle of the Pacific Ocean, especially with no increase in the terrigenous MAR. Because the terrigenous material is present in such low concentrations, and is accumulating in such small amounts, the minerals may be especially susceptible to alteration in this interval. The signal here is probably diagenetic in origin, marked by the dissolution of fine X-ray amorphous clay material in a relatively acidic environment and the creation of transitional minerals during clay halmyrosis.  proposes this explanation for mineralogical variations in sapropel layers in Eastern Mediterranean sediments. Although the organic carbon concentration here is obviously much lower than the sapropels in the Mediterranean, the very small concentrations of terrigenous material may render it more susceptible to 21 difi a tion The shape of the signal, with a sharp bottom and a gradational top, mo ic .                 "' ----~ Accumulation rates for the <2 µm minerals vs. age for Holes 885A and 886B. Aerodynamic particle size is calculated from the Al mass distribution on cascade impactor stages.
The concentration, grain size and composition of the mineral aerosol collected by the various samplers are reproducible, and all display the same temporal variation.
The atmospheric (INAA, aerodynamic particle size) analyses yield size and concentration information which is in good agreement with the analyses performed using standard deep-sea sedimentological techniques. Both data sets reveal that the aerosol physical characteristics are affected by both the source area and the transport pathway.

Introduction
Eolian deposition is the primary source of sediments in some regions of the Earth's oceans. Since the transport and deposition of eolian sediments are dependent upon source area climate and atmospheric circulation, these deposits are potentially a powerful tool for inferring climate changes through time. In order to fully exploit this paleoclimate record, it is necessary to determine how climate and atmospheric circulation processes control eolian transport and sedimentation. While there is a wealth of information for both aerosol and sediment composition and grain size, few studies in the North Pacific have analyzed both aerosol and sediment data in a strictly analogous manner. This research characterizes the mineralogy, rock-magnetism and grain size in a suite of North Pacific atmospheric samples using techniques normally employed for deep sea sediment analysis and relates the measurements to air-mass trajectories. The goal is to generate data from atmospheric samples that can be directly compared with the ancient eolian record.

Aerosol Transport
Many investigators have studied the contribution of continental material in the total particulate aerosol load presently found over the North Pacific Ocean (e.g. . One of the most spatially and temporally complete suite of samples was collected as part of the SEAREX (sea-air exchange) program . This program monitored the aerosol concentration over the Pacific Ocean using a network of island and shipboard collection sites for periods of weeks to years. The goal of this program was to identify the sources, transport mechanisms and fluxes of material in the marine atmosphere.
Many results of the SEAREX and other long-term sampling programs are relevant for paleoclimate reconstructions. Continuous annual sampling demonstrated that atmospheric Al concentration, a measure of continental dust concentration, varies seasonally Uematsu, et al., !983;. Peak aerosol Al concentrations in the North Pacific occur from February to June; the mineral concentration decreases by a factor of 2-12 during the July to January clean period. Periods of high atmospheric Al concentration are associated with Kosa events in Japan and dust storms in Asia .
In addition to temporal variability, the atmospheric Al concentration varies spatially. In the North Pacific during 1981 through 1982, the largest Al concentrations occurred at high latitudes; mean values ranged from 0.89 µg•m-3 at Shemya (54°44'N) to 0.05 µg•m-3 at Fanning (3°55'N). This spatial trend in concentration is observed for the high-dust, clean periods and the average annual concentration .
The seasonal and geographic trends noted for the mineral aerosol concentration are also observed in the measured or calculated depositional flux to the sea surface derived from Al concentration. The total aerosol flux is the sum of both the wet (75-85% of the total deposition) and dry deposition, and so varies with precipitation as well as dust concentration . The calculated mineral aerosol flux is similar to the accumulation rate of eolian sediments in the North Pacific  with most of the annual dust deposition occurring during short term dust events.

Composition of aerosols
Since the goal of the SEAREX program was the estimation of the relative importance, transport mechanisms and flux to the ocean of all material in the marine atmosphere, and because the aerosols are collected in small quantities, elemental concentrations of the marine aerosol were the primary analytical tool. These The interpretation of the Al concentration in deep sea sediments is more complex. Deep sea sediments are composed of aluminosilicates from different sources; eolian transport, hemipelagic and riverine input, authigenic formation and hydrothermal activity. Furthermore, aluminum is also incorporated into biogenic material . Multivariate statistics can be employed to differentiate between various aluminum-bearing end-members and estimate their relative importance in deep sea sediment .
However, this type of statistical analysis precludes studying the compositional evolution of a single end-member source over time, a point that is usually of interest for paleoclimatic work.
A separate consideration is that the Al concentration of aluminosilicates is relatively invariant for some common clay minerals; for example, smectite, illite and chlorite; each contain about 10% Al by weight. However, the Al concentration of kaolinite is about 20% and there is no Al in quartz, which is a very common constituent of soil-derived aluminosilicates. Thus, unless the aerosol source is homogenous, and there is no down wind fractionation of the mineralogy caused by particle size fractionation, the Al concentration may yield inaccurate estimates of mineral aerosol concentration. Finally, because variuos mineral phases are thermodynamically stable in different continental weathering environments, mineralogy is a compositional parameter which is particularly well suited for paleoclimatic work.
While there are many studies of the relationship between mineralogy and continental environments (Dixon andWeed, 1989, Chamley, 1989) as well as of mineralogy distributions in deep sea sediments (Goldberg, 1961;Biscaye, 1976;, there are relatively few studies of the mineral composition of aerosols collected over the North Pacific Ocean. This information is especially critical since aerosol samples can be related to specific meteorological conditions and to their continental source area with trajectory analysis, which would constrain the relationship between aerosol properties with transport and potential source areas.  studied the composition of dust in the eastern equatorial Pacific, using both elemental and X-ray diffraction techniques. The authors noted mineralogical differences across the intertropical convergence zone (ITCZ), but there was a lack of elemental variation across this boundary. Ferguson and others (1970) studied the mineralogy in both the clay and silt size aerosol collected in the eastern North Pacific, and concluded that the aerosols were mineralogically similar to eolian sediments from the North Pacific. Blank and others (1985) quantitatively determined the <2 µm size fraction mineralogy of a set of aerosols collected over the northwest Pacific and compared it with the mineralogy in the same size class of eolian surface sediments. The concentrations of illite, kaolinite, chlorite, quartz, and plagioclase were statistically identical in the surface sediments and aerosols. Smectite concentrations were unequal, but the authors allow that this may be due to analytical and/or sampling variability as opposed to natural processes. Aerosols from both the western and eastern North Pacific margins were studied by . The mineralogy of both the clay ( <2 µm) and silt (2-20 µm) size fraction was determined; the same mineral phases were identified as in  work. A companion study included an analysis of the synoptic meteorological conditions associated with the individual aerosol samples ) which related the compositional variability of the aerosols to differences in air mass trajectory pathways. Different mineral assemblages were associated with different potential continental source areas. All the mineralogy work mentioned above used nylon meshes as the collection medium. Quantitative aerosol concentration cannot be determined using this technique (see methods).
Buat-Menard et al., (1983) used single particle analysis to identify the mineralogy of wet and dry bulk deposition samples from Enewetak. The authors detected differences between the rain and dry deposition mineralogy, but could not determine whether this variation was a function of the different sampling interval or fractionation by scavenging processes occurring during precipitation. While individual particle analysis provides aerosol mineralogy and grain size information, this technique is not practical for deep sea sediments. Again, the atmospheric concentration of the mineral aerosol can only be estimated qualitatively using this sampling technique.

Grain Size
Grain size is a commonly measured parameter of deep sea sediments and aerosols. The information is useful because grain size is an important control on the composition, chemical cycling, entrainment, transport and sedimentation of particles in both atmospheric and deep sea environments. Grain size measurements are generally made on the chemically isolated terrigenous fraction of deep sea sediment and represent individual aluminosilicate particle sizes  based on the hypothesis that more turbulent and stronger atmospheric circulation can transport larger particles. Direct comparison of the sediment grain size with aerosol grain size is not straightforward; atmospheric scientists collect data representative of in situ particle size, which may consist of particle aggregates as well as individual aluminosilicate grains. A common sampling technique for estimating aerosol aluminosilicate grain size is fractionation with cascade impactors; these separate aerosol particles based on their aerodynamic particle size. Estimates of the aluminosilicate particle size are made by determining the mass of Al on each impaction stage and taking a geometric mean of the mass distributions over the range of size classes collected.
The mass median size distribution of aerosol particles containing Al was measured from cascade impactor samples collected during the 1979 SEAREX experiment at Enewetak . Samples from the second stage, with a mass median cutoff diameter of 3.0 µm, contained the highest concentration of Al during both the wet and dry seasons. The geometric mass median diameter ±1 Standard deviation for these samples was 2.0 ± 4.8 µm . A similar study at American Samoa indicated a mean diameter of 1.9 ± 2.3 µm (AriJiloto, et al., 1987). Both studies considered particles between median cutoff diameters of 7.4 and 0.49 µm. Total deposition (not size-fractionated) samples collected on several island locations in the North Pacific were studied by Uematsu and others (1985). The samplers (conical funnels) employed for total deposition have a larger cutoff diameter than the cascade impactors, but this cutoff has not been quantified. While the authors do not report the grain size distribution for the samples, they do include flux estimates for the <20 µm and <20 + ~20 µm size fractions, which were separated using a 20 µm mesh. The authors state that the ~ 20µm size fraction may account for up to 20 -50% of the total dust flux, so it is clear that large size fractions should generally be included in analyses of atmospheric aerosols. In addition, these published estimates of dust flux suggest that the proportion of coarse material increases for samples collected at higher latitudes. Aerosol particles larger than 15µm were collected during the ADI OS (Asian Dust Input to the Oceanic System) experiment in 1986 . Analysis of 4 total deposition atmospheric samples using SEM/EDXA (scanning electron microscope/electron dispersive x-ray analysis) provided data on the 20-200 µm size range. The authors observed an increase in the number and mass flux of particles > 75 µm coincident with the arrival of a dust plume from Asia. No information was given on the complete grain size range of the aluminosilicate aerosols.

Rock Magnetics
Rock magnetic parameters (susceptibility, anhysteretic remanent magnetization, isothermal remanent magnetization) are indicative of the concentration, grain size and composition of magnetic material (for example, magnetite and hematite) which occurs naturally in soils and soil-forming parent rocks. Rock magnetic measurements are commonly employed for the analysis of sediments because they are a rapid, non-destructive means of estimating the concentration and grain size of the total terrigenous component of deep sea sediments Doh, et. al., 1988, Bloemendal and). In addition, susceptibility can be measured on whole cores, providing a continuous proxy for rerrigenous activity over time. Studies have used rock magnetization as a diagnostic tool for mineral aerosol origin , Chester, et al., 1984  This study quantifies the characteristics of the mineral aerosol collected over the open North Pacific Ocean using standard sedimentological techniques for mineralogy, particle size and rock magnetization determination. These measurements are readily accomplished on aerosol samples collected with meshes, due to the large sample mass of collected material. The mineralogy was determined by powder X-ray diffraction using an internal standard in order to quantify the amount of each mineral phase present. Both the <2 µm and 2-20 µm size fraction mineralogy were determined for most samples. Elzone particle size is measured from 1 µm to 63 µm.
Rock magnetization parameters, susceptibility (X), anhysteretic remanent magnetization (ARM) and isothermal remanent magnetization (IRM) are also determined for each of the samples. The measurements obtained for each sample are then compared with air mass trajectories, in order to relate the properties of the sample to the source areas and transport pathways.
The mineralogy measurements are also made on cascade impactor and high volume samples collected during one cruise. These mineralogy analyses are not quantified in terms of absolute weight percent, since no internal standard is used in the analysis. The cascade impactor sample mineralogy is determined for the different size classes collected on separate stages. Thus, both the high volume and cascade impactor sample mineralogy may be compared with the aluminum concentration in order to determine whether size fractionation of the mineralogy has any effect on the aluminum concentration in a sample. Finally, mass median diameter size estimates from the distribution of aluminum mass on the cascade impactors may be qualitatively compared with the particle size determination from the mesh samples.
This set of comparisons will allow us to relate standard atmospheric analyses with standard sedimentology determinations.

Sampling
Aerosol sampling took place on 2 cruises in the North Pacific in the spring of 1986 and 1987 (Fig. 1) were hung outboard of the research vessel where they were dampened by high relative humidity but not washed by waves. The meshes were deployed far forward of the ships' stacks in fair weather (no rain or fog) and only when the wind was off the bow of the ship. Any potential contamination of the meshes from stack gases or shipboard activity was recorded.
The mesh collection technique is useful for a sedimentological analysis; since large volumes of air are sampled, large numbers of particles may be collected rapidly.
From an atmospheric scientist's point of view, there are several limitations to the meshes. First, it is not possible to calculate atmospheric particle concentrations quantitatively, because there is no way to determine the volume of air sampled accurately. Additionally, the overall collection efficiency of the meshes is low (-30%) . The efficiency likely varies with absolute wind speed and relative humidity. The collection efficiency of the meshes may vary with particle size so that the samples are not representative of the total aerosol, because they bias against fine particles .  found that the mineralogy of aerosols collected concurrently on pumped filter samplers and on meshes were compositionally identical. The one exception was microcline, but the authors assert that this difference was an artifact. There are no comparative studies of grain size or rock magnetism between meshes and other techniques. In summary, the meshes provide samples suitable for sedimentological analyses, but are a crude technique for quantitative atmospheric studies.
Fourteen mesh samples were collected on the research cruises. The sampling intervals ranged from 15 to 117 hours (Table 1). Eleven samples produced enough material for quantitative mineralogical analysis of both the <2 µm and 2-20 µm size fractions; the remaining 4 samples were quantitatively analyzed for mineralogy of the <20 µm size fraction. Twelve samples provided enough material for Elzone grain size analysis. All samples were subjected to rock-magnetism analysis.

Pumped Samples
High volume cascade impactors or bulk samplers were operated on both cruises ( diameter corresponding to the AED of the particle. Any particles too fine to be collected on the impactor stages are collected on a final cellulose filter. The instrumental configuration used on the SEAREX cruise may bias against particles greater than 10 µm, resulting in unreliable collection efficiency from stages 0 and 1 . The 50% aerodynamic cutoff diameters for each stage are: Sample Processing

Meshes
The aerosol particles are washed off the nylon mesh for analysis with repeated rinses of deionized water. Each duplicate sample mesh is processed separately. The rinse water is filtered onto pre-weighed 0.45 µm Nuclepore membranes, and the airdried filter is re-weighed to determine the sample mass. The dried aerosol sample is scraped from the filter and stored in a vial for analysis. The large volume(> l liter) of water necessary to extract the aerosols from the mesh washes any sea salt from the sample. Bulk sample quantities ranged from 0.0047 g to 0.8852 g per mesh.
Pumped Samples The cellulose substrate must be removed from the aerosol sample for X-ray diffraction analysis as the cellulose produces a high background on X-ray diffraction scans which interferes with the low intensity signal from the aerosol aluminosilicates.
In addition, removal of the filter substrate concentrates the aerosol sample, which serves to intensify the aluminosilicate signal. Finally, by processing a larger portion of the filter than can be exposed to the x-ray beam at a single time, any effects from uneven distribution of particles on and within the filter substrate will be minimized.

78
The cellulose filter was oxidized by a low-temperature ashing technique. A 1 to 95 cm2 section of cellulose filter or a single impaction substrate strip was cut with plastic scissors and placed in an acid-washed Teflon vial. The samples or blanks were combusted in a LFE Corporation model LTA-505 low temperature asher at 50 Watts, corresponding to a temperature below 100°C for 17 to 71 hours. After the cellulose was fully combusted, the residual material was mixed into a slurry with a small amount of deionized water. This slurry was pipetted onto a pre-weighed zero background quartz plate for X-ray diffraction analysis.

Rock Magnetism
Samples were packed into 1.5 cm3 Nalgene cylinders for analysis. Cotton was techniques are referred to .

Grain Size
The grain size distribution of the aerosol mesh samples was determined using a model 180 Elzone Particle Analyzer. Elzone analysis measures the volume of electrolyte displaced by a particle when it passes through a small orifice between two electrodes. The volume is converted to equivalent spherical diameter and the instrument is calibrated using latex spheres of a known diameter and volume.
Samples were analyzed to provide a continuous grain size distribution (in equivalent spherical diameter) between 1 and 63 µm.
After the aerosol had been analyzed for rock-magnetism properties, it was wet-sieved at 63µm. The> 63µm samples were weighed and stored. The suspended <63 µm sediment was shaken and a subsample of this slurry was pipetted into a filtered solution of 4% sodium pyrophosphate and stirred continuously for analysis using a 120 µm diameter orifice tube. This procedure was repeated, sieving at 20 µm and using a 48 µm orifice tube to provide a more accurate measurement of the <20 µm size fraction. The 20-63 µm sieved size fraction was weighed and stored. The <20 µm size fraction was subsequently used for quantitative mineralogical analysis. 128 channels were measured for each orifice size and the results for both analyses were combined using a smoothing program to provide a continuous grain size profile. 80

Mineralogy
The <20 µm mesh aerosols were saturated with MgCl2, to reduce ct-spacing variability in clays with exchangeable cations; they were then rinsed with warm deionized water. The <20 µm fraction was split into 2-20 µm and <2 µm size fractions by centrifugation. The 2-20 µm and the <2 µm fractions were spiked with a 10% talc internal standard. Samples were homogenized by grinding in a mortar and pestle under acetone, air dried, suspended in a deionized water slurry, and drawn onto silver filters for X-ray analysis.
Both the mesh and pumped aerosol samples were X-rayed from 2° to 45°20 at 45 kV and 40 mA at 1° 20 per minute, using Cu-ka radiation. Peak areas for smectite, illite, kaolinite, chlorite, quartz, and plagioclase were determined using Scintag DMS software. The relative proportion of kaolinite and chlorite was determined by the relative proportions of the kaolinite [002] and chlorite [004] peak areas. Mineral peak areas were normalized to the internal standard peak areas for aerosol mesh samples.

Air Mass Trajectories
We used isentropic trajectory analysis to estimate the movement of the dustladen air masses from which our samples were collected. This method provides a sample-specific assessment of potential source areas and transport pathways. The calculations are based on the global meteorological analysis of the National Meteorological Center. This is a gridded set of horizontal wind and geopotential height fields, at 2.5° latitude and longitude spacing, available twice daily. The winds are interpolated from the isobaric levels to isentropic surfaces at 5K intervals. A detailed description of this technique may be found in . Examples of 81 the application of this technique to mineral aerosol transport are presented in .

Results
Meshes Table 3 shows the mass of dust collected on each mesh as well as the Elzonedetermined geometric mean diameter for the aerosols. The aerosol mass recovered from the meshes ranged from 0.005 to 0.885 grams per mesh panel. Paired compaf!sons indicate that the replicate meshes collect dust in a reproducible manner at a 95% confidence limit for each cruise. Figure

Meshes
The mineralogy of the aerosols was determined for both the <2 µm and the 2-20 µm size fractions for most samples. Samples smaller than -20 mg of material per mesh were not size-fractionated and the bulk sediment mineralogy was determined.
The mineral phases quantified are smectite, illite, kaolinite, chlorite, quartz and plagioclase. For all analyses, paired comparisons indicate no statistically significant differences between replicate meshes at a= 0.05 (Table 5).
We wanted to use the mineralogy to: examine the relationship between source area and transport mechanism with compositional variation; compare our results to other studies; and, compare mineralogy of the pumped samples collected on the Moana Wave cruise. In light of these plans, we reduced the raw diffraction patterns in the following manner. The peak areas for each phase were normalized to the internal Standard (10% talc) to correct for inter-sample differences in peak intensity. This normalization effects the variance between samples, but does not change the relative proportion of minerals within a sample. The normalized peak areas were multiplied by weighting factors derived for North Pacific deep sea sediments (Heath and Pisias,84 197 9), in order to estimate the absolute weight percent of each mineral phase present.
While these weighting factors may not be accurate for aerosol samples, they allow us to compare the results of these analyses with previous studies of deep sea eolian sediment mineralogy and aerosol mineralogy. To facilitate comparison with other studies and with the pumped aerosol samples, the data are normalized to 100%. This expresses the data on an "amorphous-material-free" basis. In the <2 µm size fraction, smectite comprises 3% of the diffracting material, chlorite 5% and kaolinite 9% (Table 5, Figure 9). The primary minerals quartz and plagioclase are present in approximately equal concentrations, 7 .5% and 7% respectively. The 2-20 µm size fraction contains 2% smectite, 7% kaolinite and 5% chlorite (Table 5, Figure 8). The quartz and plagioclase concentrations are 19% and 18% respectively.
There are no significant mineralogical differences between the ADIOS and Moana Wave cruise samples. Another way to examine the mineralogy is to look at differences in the concentrations of the various mineral phases between the two size classes. The clay minerals smectite and illite are present in significantly higher concentrations (0.6% and 20% higher, respectively) in the< 2µm size fraction relative to the 2-20 µm size fraction for paired comparisons of all samples at a=0.05.
Quartz and plagioclase are concentrated 12% and 11 % higher in the 2-20 µm size fraction than in the <2µm size fraction.

Pumped Samples
Samples from two cascade impactor and three high volume samplers which were analyzed for mineralogy from leg 1 of the Moana Wave cruise, and samples from two high volume samplers were analyzed from the second leg of the cruise. We  (table 7). This analysis, together with the observation that the large aliquot samples HV 43 and HV 197 are not significantly different for all parameters suggests sample inhomogeneity is a problem for small high volume sample aliquots.
These samples (<4 cm2) were excluded from the data set.
Field blanks and procedural blanks were analyzed with the pumped samples.
The ashed residue from all the blank samples was negligible. Peaks were present in some diffraction patterns from the field blank samples. However, the peaks were intermittent and the phases could not be conclusively identified. Common ct-spacings included 6.33A, 3.63A and 3.18A. Sodalite or a sodium -calcium sulfate phase have reflections at these ct-spacings, but a multiple line match could not be made for any single sample. The ct-spacing at 3.18A overlaps with the ct-spacing used to identify plagioclase, thus this mineral phase will be affected by these contaminating phases.
Low temperature oxidation of the cellulose substrate should not affect the aluminosilicate fraction of the aerosol. This assumption was assessed by running clay standards through the ashing procedure. A few milligrams of the clay standards montmorillonite (smectite), illite and kaolinite were run through the low temperature asher and there was no modification of the diffraction pattern in any of the standards.
In addition to the clay standards, we ran two samples collected on the mesh samplers through the low temperature asher; the diffraction pattern of the aluminosilicate fraction was not affected by the ashing process.
The data may be compared with Arimoto's (written communication, 1995,  interval as the samples analyzed for mineralogy. When this is the case, both sample identification numbers are listed on the figure. The variation of the elemental and ashed data is illustrated for the total samples (Fig. 12), and by stage for the two cascade impactors (Fig. 13).  There are also important temporal changes in the transport time and potential source areas for each cruise. For the Moana Wave 1986 cruise, the most significant change is the drop in transport time from eastern source areas for the later part of the cruise. These short transport times from the Aleutian Islands and Alaska indicate a higher potential for these source areas to influence the physical characteristics of the aerosols. The shortest transport time from Asia, the most dust productive region which impacts the North Pacific, occurs during the sampling interval for HV7 and Mesh 1. Temporal changes in the ADI OS 1987 transport regime are much simpler than those for the Moana Wave cruise. The potential source areas do not appreciably change for the first four sampling intervals, but the transport time from Asia and Japan increases from 3 days at the beginning of the cruise to 8 days at the end. Mesh 5 is influenced by the most southerly Asian sources for both the Moana Wave and 92 ADIOS samples. Mesh 6 is influenced by the most northerly Asian sources for this cruise, as well as some transport from eastern source areas.

Discussion
Comparison with other aerosol studies The majority of the chemistry work which has been performed on North Pacific mineral aerosols does not include a sample specific trajectory analysis, as perfonned in this work. Rather, the mineral concentration and grain size information is expressed as seasonal or latitudinal averages. Thus, direct comparison between this work and previous analyses is limited to comparing broad ranges of information.
For the pumped samples, the mineral aerosol concentration, based on an 8% Al concentration for mineral dust, and using the high volume and cascade impactor samples for the calculation (which was the value used in the previous studies), ranged from 4.1 µglm3 to 0.09 µglm3. These estimates are within the range of mineral aerosol concentrations previously published for North Pacific aerosols , and references therein). The grain size estimates, based on the cascade impactor aluminum and iron mmd range from 1.14 µm to 5.33 µm; Arimoto and others (1985) report a mmd, based on aluminum concentration, of 2.0 µm ± 4.8 µm for mineral aerosol samples collected at Enewetak. The Elzone geometric mean diameter of the mesh samples ranges from 3.06 µm to 7.62 µm. There are no previously published estimates of electronic particle size measurements for North Pacific aerosol to compare with the mesh samples. Comparison of the mineralogy results with previous aerosol mineralogy work is incorporated into the discussion of aerosol vs. air mass trajectories.

Meshes vs. Pumped Samples
The pumped and the mesh analyses can be compared in three ways; concentration, grain size and composition. These comparisons are necessarily qualitative because the mesh samples and pumped samples were not collected over exactly identical time intervals. Figure 18 shows

Aerosols vs. Air Mass Trajectories
We can make some general statements about the expected effect of source area and transport pathway on the concentration, grain size and composition of the aerosols. Transport time will predictably effect all of the physical parameters. Dust concentration and grain size will decrease with increasing transport time as particles settle out of the dust plume, with coarse particles sedimenting out more rapidly than fine particles. The composition of the aerosols will become increasingly dominated by clay minerals, as the coarser grained primary minerals settle out of the atmosphere with transport time. In addition to transport effects, all of the physical parameters will vary with changes in source area.
We test our predictions by examining the relationship between source area and transport time with the measured variables. The air masses pass over multiple source areas, at different speeds, on their way to the sampling stations. The small number of observations preclude statistically decoupling transport effects from source area effects with multivariate statistical analysis. In addition, it is not possible to completely isolate the effects of an individual source area with this data set, as the open ocean samples integrate many source areas, due to the long transit times.
The simplest transport effects to understand are the relationships between decreasing dust concentration and grain size with increasing transport time. Figure 23 illustrates the relationship between dust concentration and transport time for each sample, broken down by source area. For example, the three points at the highest dust concentration, 0.0125g/hr, represent ADIOS mesh 2. The air mass for this sample  The comparison of the mineral phase concentration with transport time is illustrated in figures 27 and 28. These figures include mineralogy data from the western North Pacific (which have been normalized to 100%) from Leinen,et. al.,199 4. The collection techniques and analytical procedures used by Leinen, et al., 19 94 were identical to those in this study. The samples from this previous study were collected from the margins of the North Pacific Ocean, thus transport time from the source areas is much shorter, and there is less mixing between multiple source areas.
This affords us the opportunity to examine a broader range of transport times for the mineralogy data. The correlations between transport time and the mineralogy of the aerosols are presented in table 11; significant correlations (a=0.05) are highlighted in bold font. The source areas Siberia and Alaska had too few air mass trajectories to perform correlation analysis. The relationship between transport time and particle fractionation predicts that the concentration of the clay (primary) minerals should be positively (negatively) correlated with transport time, as the coarse grained primary minerals preferentially settle out with increasing time. This mechanism is consistent with the significant positive correlations for smectite, illite, chlorite, and negative correlations with quartz and plagioclase. Kaolinite, however, is negatively correlated with source areas from Japan and Asia in the <2µm size fraction, but is positively correlated with transport time for the Aleutian source area in the 2-20µm size fraction.
The alternative mechanism for this mineralogy variation is mixing between Asia/Japan source areas and high latitude Aleutian/ Alaska/Canada source areas. The observed trends in mineralogy regressed against Asian transport predict that the high latitude source area would be relatively enriched in the clay minerals smectite, illite and chlorite and depleted in the primary minerals quartz and plagioclase as well as kaolinite clay. While most of these mineralogy differences between the two source areas are plausible, the requirement that plagioclase be depleted in the high source area is in conflict with both the grain size data and the geological differences between these two source areas. Thus, we conclude that the mineralogy variation is the result of both mixing and transport driven mineral fractionation.
The measurements for the pumped samples were also evaluated against transport time from the different source areas. The mineralogy regression against transport time was performed separately because the pumped samples were not differentiated into separate size fractions for XRD analysis. Transport times from Asia (>60°N) and Japan are correlated with illite and inversely correlated with plagioclase for the pumped samples. Transport time from the Aleutians is correlated with kaolinite. The aerosol size, aluminum concentration and ash residual are not significantly correlated with transport time. There are fewer significant transport time -mineral phase correlations for the pumped samples, but the analysis demonstrates the same trends as that generated for the mesh samples.

Implications for interpretation of eolian sediments
The interpretation of paleoclimate and paleometeorology from eolian deep sea sediments generally includes the assumptions that the grain size is related to atmospheric transport vigor and the mineralogy is related to source area. While this data set is generally consistent with these assumptions, the relationship between atmospheric transport and aerosol physical properties is not that simple. The grain size is affected by variation in source area, and the mineralogy is fractionated by differential settling during transport. Thus, paleoclimate/meteorology studies of eolian sediments should include multiparameter analyses, as well as sample distributions on a large enough spatial scale to capture transport-related compositional and grain size gradients.
100 Conclusions Atmospheric dust concentration and mineralogy data generated from mesh samplers, cascade impactors or high volume pumped samples are comparable, and all sampler types display the same temporal variation in the physical properties.
However, aerosol size estimates generated by Elzone analysis for mesh samples and elemental mass distribution from cascade impactors, while comparable for fine grained samples, diverge for samples with large Elzone size estimates.
The concentration, grain size and composition of aluminosilicate aerosols collected over the North Pacific Ocean reflect variation in both the source area of the dust and atmospheric transport driven fractionation of the aerosol. While there is strong evidence for both transport and mixing influences on the composition and size of aluminosilicate mineral aerosol, a larger data set is required in order to estimate the relative importance of both processes.
This data set supports mixing between a dominant Asia dust source and a weak high latitude dust source, and transport driven fractionation of the aerosol. Asia is a strong, fine grained dust source which is relatively enriched in 2-20µm quartz and <2µm kaolinite. The high latitude source area is a weak, coarse-grained dust source which is relatively enriched in plagioclase and magnetic minerals.
Transport related mineral fractionation drives the mineralogy towards a clay mineral enriched and primary mineral depleted composition with increasing transport time. These trends are observed for all source areas studied, in both the mesh sample mineralogy and the data generated from cascade impactors and high volume samples.                 Subsequent research has demonstrated that the grain size (Dauphin, 1983;, composition  and flux (Leinen, 1989) distributions of surface sediment properties in the North Pacific Ocean display a pattern which is consistent with dust transport through the atmosphere from Asia. There is a maximum in the flux of quartz beneath the axis of the zonal westerlies, and the concentration decreases from west to east. Eolian material accumulates very slowly in the North Pacific Ocean. Present-day eolian mass accumulation rates are highest between 38-40°N in the Northwest Pacific (-2,000 mg/cm-2fky-1) and rates decrease to the east, north and south (Leinen, 1989). The decreases are related to distance from the major eolian source (arid regions of Asia) and the predominant transport pathway, high-altitude tropospheric westerlies associated with large storms (Leinen, 1989). Grain size also decreases from west to east, and the mineral distribution patterns mimic the trends displayed by the flux and grain size properties.

~··
Atmospheric scientists have also studied dust transport from Asia into the North Pacific. Continuous sampling through an annual cycle, measured during the SEAREX program in the North Pacific, demonstrated that atmospheric Al concentration, a measure of continental dust concentration, varies seasonally . Peak aerosol Al concentrations in the North Pacific occur from February to June; the mineral concentration decreases by a factor of 2-12 during the July to January clean period. Periods of high atmospheric Al concentration are associated with Kosa events in Japan and dust storms in Asia .
In addition to temporal variability, the atmospheric Al concentration varies spatially. In the North Pacific during 1981 through 1982, the largest Al concentrations occurred at high latitudes; mean values ranged from 0.89 µg•m-3 at Shem ya (54 ° 44'N) to 0.05 µg•m-3 at Fanning (3°55'N). This spatial trend in concentration is observed for the high-dust, for clean periods and for the average annual concentration .
The seasonal and geographic trends noted for the mineral aerosol concentration are also observed in the measured or calculated depositional flux derived from Al concentration. The total aerosol flux is the sum of both the wet (75-85% of the total deposition) and dry deposition, and so varies with precipitation as well as dust concentration .
The calculated mineral aerosol flux is similar to the accumulation rate of eolian sediments in the North Pacific  with most of the annual dust deposition occurring during short term dust events. The water of the central gyre is unproductive and the mineral tests of biogenic materials which are produced are not preserved, since the sediment surf ace is well below the CCD in this region, and the bottom water is undersaturated with respect to silica. The contribution of hemipelagic material drops off rapidly with increasing distance from the continents, but this component is difficult to distinguish from eolian sediment because the lithogenic component of hemipelagic sedimentation contains the same mineral phases as the eolian component.  have developed a technique for distinguishing between hemipelagic and eolian sedimentation based on the shape of the terrigenous grain size distribution; the hemipelagic sediment has a flatter distribution and is coarser grained compared with the eolian material. This technique can be used to determine the dominant sedimentary component, but cannot quantitatively partition the sediment into the two 189 sources. The other component which may contribute to the sediment in the central basin is ash derived from the circum-Pacific island arc chains; however, this component can be quantitatively distinguished from the eolian sediments by statistical partitioning based on chemical analyses (Weber, et al., 1995;Olivarez, et al., 1991). The samples studied in this research cover the full range of sedimentation regimes in the North Pacific Ocean, but primarily consist of samples from the central basin.

Composition
The average composition of the eolian <2 µm size class in the North Pacific red clays is 30-40% illite, 10-15% chlorite, 10-15% quartz, 10-15% plagioclase, 10-15% kaolinite and 0-5% smectite (Leinen, 1989). Quartz and illite concentrations are highest in a latitudinal band centered at -30°N . Chlorite concentrations are highest in the north. Kaolinite concentrations are high in latitudinal bands extending from the western and eastern borders into the center of the Pacific basin, but the distribution pattern does not display a continuous band across the ocean . Smectite concentrations are highest along the margins of the North Pacific basin .
Grain Size J. Paul Dauphin studied the particle size distribution of quartz in the silt-size fraction of North Pacific surf ace sediments (Dauphin, 1983). Dauphin determined that the mass median diameter of the 2.5-68.3 µm size fraction decreased from 10 µm at 150°E to 7 µmat 140°W. The pattern of quartz grain size distribution shows the same latitudinal distribution as the quartz concentration in surface sediments; the latitudinal band of coarsest grain size is centered at -30°N.  have measured the 1-30 µm grain siz.e of 40 surface sediment samples in the North Pacific Ocean. In addition to the previously observed trend of decreasing grain size with increasing transport distance from Asia, the authors were able to discriminate between hemipelagic and eolian sediments based on the shape of the grain siz.e distribution curve. Janecek (1985) determined the siz.e of eolian surf ace sediment particles in the 1-16 µm siz.e range. The median grain siz.e decreased from 4 µm in the western Pacific (DSDP site 578) to 2.4 µm in the eastern Pacific (site LIA4-GPC-3).

Rock magnetics
The are no maps of the surf ace sediment rock magnetic properties in the North Pacific. Rock magnetic measurements provide information on the phase, concentration and grain siz.e of magnetic minerals, primarily iron oxides, in sediments. Since the iron oxide concentration, grain siz.e and composition are highly correlated with the same parameters of the total aluminosilicate (terrigenous) component of sediments , these measurements have been extensively used in down core studies to provide continuous proxies of terrigenous sedimentation through time. The measurements are non-destructive, rapid, and can be conducted on whole cores. Doh and others (1988) used the rock-magnetism properties of sediment recovered from core LL-GPC3, located at 30°N and 157°W in the central North Pacific, to study the paleoceanographic changes of the eolian sediment deposited through time at this location. The authors were able to document changes in the iron oxide flux, composition and grain size based on the rock magnetic analyses which recorded the evolution of sedimentary processes through time. These analyses agreed well with previous studies of the terrigenous composition and grain siz.e which tracked the movement of the site from the easterly wind regime into the westerly wind regime of the North Pacific. In addition, the onset of northern hemisphere glaciation was expressed as increasing iron oxide grain size and flux, and changing iron oxide composition. This temporal variation in rock magnetic properties should be expressed as spatial variability in the surface sediments, and will be investigated as part of this research.
Here, we again measure the mineralogy and total terrigenous grain size as well as determine the spatial distribution patterns of rock magnetic properties (susceptibility, anhysteretic remanent magnetism and isothermal remanent magnetism) in North Pacific surface sediments. For the grain size and mineralogy, we include coarser size fractions than previous work; mineralogy is measured in both the <2 µm and 2-20 µm size fractions and grain size is determined over a 1to63 µm size range. We make all measurements on the same samples, so that we can characterize the covariation of the different sediment properties. We have analyzed a set of aerosol samples collected over the North Pacific Ocean using identical analytical techniques, for comparison with the surface sediments. The aerosol samples have been related to meteorological analyses which allows us to directly compare the physical properties with the source area and transport pathway of eolian sediments.
The goal of this research is to provide an integrated data set which relates various eolian sediment proxies with present day eolian sedimentation processes,

Methods
The data set consists of a suite of 168 surface sediment samples collected from the North Pacific (Table 1) Further explanation of these variables and their applications may be found in the results section. Readers unfamiliar with rock-magnetic techniques are referred to .
After the rock-magnetism was determined for the bulk samples, the samples were treated to remove biogenic calcite with a buffered acetic acid extraction.
Samples were then wet sieved at 63 µm; the >63 µm and <63 µm fractions were then dried and weighed. The <63 µm fraction was treated to remove biogenic silica using a sodium carbonate procedure. Iron oxides were removed using the oxalic acid extraction technique of . Sediments were saturated with MgCl2, to reduce d-spacing variability caused by cation differences; they were then rinsed with warm deionized water.

Grain Size
The grain size distribution of the sediment samples was determined using a model 180 Elzone Particle Analyzer. Elzone analysis measures the volume of electrolyte displaced by a particle when it passes through a small orifice between two electrodes. The volume is converted to equivalent spherical diameter and the instrument is calibrated using latex spheres of a known diameter and volume.
Samples were analyzed to provide a continuous grain size distribution (in equivalent spherical diameter) between 1 and 63 µm .
After the sediment had been processed to remove biogenic silica, calcium carbonate and iron oxides, the suspended <63 µm sediment is shaken and a subsample of this slurry is pipetted into a filtered solution of 4% sodium pyrophosphate and stirred continuously for analysis using a 120 µm diameter orifice tube. This procedure is repeated, sieving at 20 µm and using a 48 µm orifice tube to provide a more accurate measurement of the <20 µm size fraction. The 20-63 µm sieved size fraction is weighed and stored. The <20 µm size fraction is subsequently used for quantitative mineralogical analysis. 128 channels are measured for each orifice size and the results for both analyses are combined using a smoothing program to provide a continuous grain size profile between 1 and 63 µm. 194

Mineralogy
The <20 µm fraction was split into 2-20 µm and <2 µm size fractions by means of centrifugation. The 2-20 µm size and the <2 µm fraction was spiked with a 10% talc internal standard. Samples were homogenized by grinding in a mortar and pestle under acetone, air dried, suspended in a deionized water slurry, and drawn onto silver filters for X-ray analysis.
The samples were X-rayed from 2° to 32° 28 at 45 kV and 40 mA at 1° 28 per minute, using Cu-ka radiation. Peak areas for smectite, illite, kaolinite, chlorite, quartz, and plagioclase were determined using Scintag DMS software. The relative proportion of kaolinite and chlorite was determined by the relative proportions of the kaolinite [002] and chlorite [004] peak areas. Mineral peak areas were normalized to the internal standard peak areas and converted to mineral weight percent using the weighting factors of Heath and Pisias ( 1969).

Rock magnetics
For all discussion, samples further south than 20°S are excluded from consideration, due to paucity of data below this latitude. The magnetic measurements are made on unprocessed samples, thus recording the influence of all the sedimentary processes effecting the sample site. The concentration of magnetic material in a sample is determined from the susceptibility (X) data, which is a measurement of the magnitude of the magnetic moment measured while the sample is in an magnetic field. This measurement is made at two frequencies, with differences in the high and low frequency susceptibility due to the contribution of ultra fine, or superparamagnetic, particles. The high frequency susceptibility (Xhf) and low frequency susceptibility (Xlf) have means of 0.63± 0.65 and 0.66±o.67 µm3fkg ( Table 2). The magnetic concentration is lowest along the equatorial region where terrigenous material is diluted with the high flux of biogenic (primarily calcium carbonate) sedimentation (Fig. 1). µAm2fkg. The spatial distribution of SIRM closely mimics the distribution of the susceptibility (Fig. 2).
The HIRM is the amount of remanent magnetism remaining after the magnetically saturated sample has been exposed to a weaker reversed field. This measurement depends both on the concentration and the coercivity (a function of the composition) of magnetic material. As an example, magnetite is a "soft" or low coercivity mineral; all of the induced magnetism from the SIRM is stripped off in the weaker reversed field. Conversely, hematite is a "hard" or high coercivity material.
Since "hard" minerals are not easily magnetized, some of the magnetism from the saturation step is retained after exposure to the weaker reversed field. Thus, this measurement indicates the concentration of high coercivity material in the sediment.
The mean HIRM is 425±579 µAm2fkg. Again, the spatial distribution of the HIRM mimics the Xlf and the SIRM (Fig. 3).
The final magnetic concentration measurement is the anhysteretic remenant magnetism (ARM), here expressed in the same units as the susceptibility, as Xarm.
This measurement is sensitive to both the concentration and the grain size of the magnetic material, with finer particles yielding a higher measurement than larger particles. The mean Xarm is 11.5±7.8 µm3fk:g. Unlike the previously discussed measurements of magnetic concentration, the highest values for Xarm occur in the central basin from the equator north to Hawaii (Fig. 4). Since the Xarm values are sensitive to both magnetic concentration and magnetic particle size, normalizing these measurements to concentration will yield a parameter which is inversely proportional to magnetic particle size. A map of the spatial distribution of the ratio Xarm/Xlf is presented in figure 5. The bull's eye in the southeast quadrant of the map is located over the east pacific rise, and likely represents the magnetic signature of hydrothermal sediments, which would contribute very fine grained iron oxide partjcles to the sediment. In general, the XarmlXlf values of samples south of 10°N are high, indicating fine magnetic particle sizes. The ratio is low around the margins of the northern basin, consistent with a large contribution of coarse terrigenous and volcanogenic magnetic material.
The final magnetic parameter is the S-ratio, -IRM/SIRM, which is a compositional parameter expressing the proportion of high coercivity material in the sediment. Sediment containing a large proportion of high coercivity material, such as hematite, will have a lower S-ratio than sediments with a large proportion of low coercivity material such as magnetite. Samples with the lowest S-ratio are located in the central North Pacific (Fig. 6). The relatively high proportion of high coercivity material in this region is related to the dominance of terrigenous sedimentation.
Volcanogenic material has a high proportion of magnetite, but material from the deserts of northern Asia should have a higher proportion of hematite produced by the weathering conditions in this region.

Grain Size
Tue median grain size for the <63 µm terrigenous fraction of the samples is 6.1±4.6 µm ( Table 3). The largest grain sizes are found on the southeast quadrant of the map area, from the spreading ridge/hydrothermal areas (Fig. 7).

Mineralogy
There are six mineralogy phases which were quantified in this study; smectite, illite, kaolinite, chlorite, quartz and plagioclase (Table 4) Kaolinite is present in concentrations of 17±11 % in the <2 µm size fraction and 8±7% in the 2-20 µm size fraction. The <2 µm size fraction displays moderate concentrations along the northern margin of the basin, and high concentration off the coast of Central and South America (Fig. 14). The 2-20 µm size fraction kaolinite is most common off the coast of Central and South America (Fig. 20).
Chlorite ranges from 5±3% and 3±2% in the <2 µm and 2-20 µm size fractions, respectively. The highest chlorite values tend to occur near the northeast and eastern margin of the basin in the <2 µm size fraction (Fig. 15). The 2-20 µm chlorite concentration is high near the margins and in the region around the Hawaiian Islands (Fig. 21).

Discussion
The fundamental hypothesis investigated in this research is that the rock magnetics, grain size and mineralogy of the sediments are indicative of the sediment source area and the transport dynamics. One important question is whether each of the parameters convolve both the source signal and the dynamics, or whether some parameters are more sensitive to source region and other to transport process. When interpreting the eolian sediment record, particle size has been used as a measure of transport vigor, and composition as an proxy for either changing source area or changing weathering regime in the same source area.
In order to investigate this problem, a series of plots were constructed illustrating the variation of the measured parameters with transport distance  Figure 25 shows the magnetic composition parameter S, plotted as a function of longitude. There is a general decrease in the ratio, indicating an increase in the relative proportion of high coercivity material with distance from the continent. This is consistent with both diminishing hemipelagic/volcanogenic input across the ocean, but likely also includes fractionation of the eolian aerosol during transport. As dense, magnetic particles preferentially settle out, the aerosol is enriched in fine grained aluminosilicates coated with high coercivity iron oxides. A large change in the proportion of high coercivity to low coercivity material is required to change the Sratio, thus this observation is not necessarily in conflict with the mechanism proposed for explaining the discrepancy between the aerosol and sediment Xann values.
The rock magnetic grain size proxy Xann/X and the aluminosilicate geometric mean and median grain diameter are illustrated in figure 26. There is a steep decrease in grain size from 130°E to l 60°E, followed by a small, but steady decline in grain size across the entire North Pacific basin. The aerosol measurement of Xann/X fall well below the sediment values, and this again is explained by some dissolution of iron oxides in the ocean.
The aerosol Elzone particle size measurements fit the sediment observations. Figure 27 shows the variation of the weight percent of each of the sediment size fractions separated during sample processing. There is very little >63µm material in North Pacific surface sediments. The 20-63µm size fraction shows the same continentally influenced drop off between 130°E and 160°E observed in the previously discussed observations. The 2-20µm size class remains relatively constant, but the <2µm size class steadily increases across the basin. It has been suggested that 202 wind transported aluminosilicates achieve equilibrium with respect to the particle . and the zonal wind velocity after a few thousand kilometers (Janecek and Rea, s1z.e 198 5;Janecek and Rea, 1983;; this hypothesis is not supported by the sediment observations in this work.
Fractionation of aerosol mineralogy has been described by , continents. The sediment is characterized by low magnetic concentration, extremely small magnetic particle size, fine aluminosilicate particle size, and is rich in smectite.

203
The East Pacific Rise spreading/hydrothermal area is characterized by very large aluminosilicate particle size, very fine grained magnetic particle size, low magnetic concentration and unique mineralogy.
Large There is good agreement between the magnetic concentration, the overall composition and particle size of eolian sediments and aerosols collected over the North Pacific Ocean. The eolian material is fractionated as it travels through the atmosphere. The aluminosilicate and magnetic particle size decreases across the North Pacific Basin. The aerosols are also compositionally fractionated during sediment transport; the S-ratio decreases, indicating a relative enrichment in high coercivity material with increasing distance from the source area. There is no coherent pattern in the <2µm mineralogy across the basin, but the 2-20µm size fraction demonstrates an increase in chlorite and kaolinite and a decrease in the plagioclase concentration across the basin.