Fe Oxide Mineral Source Determinations Samples collected from the shelves and coastal lands around the Arctic Ocean were analyzed by the procedures below, clustered into composition groups for each mineral and each source area as defined by unique lithic grain compositions, and used to match grains from sediment cores collected from the Arctic Ocean for provenance determination (Darby and Bischof, 1996; Darby, 1998). This large data set consists of over 13,450 analyses of individual Fe oxide grains. All samples were wet-sieved into 45-63 µm and 63-250 µm fractions, dried, and the magnetic Fe oxide minerals separated by hand magnet and then Frantz magnetic separator set at 0.3 amp. The magnetic minerals were combined and mounted in epoxy plugs, ground to expose the grains, polished, photographed, and identified with reflected-light ore-microscopy (1000x). Each identified grain was numbered on the photograph of the mount and analyzed for 12 elements by an auto-microprobe. This analysis counted each element for 20 seconds or 20,000 counts and standards were Fe oxides similar to the minerals analyzed (Darby and Bischof, 1996). The 12 elements are all known to occur in one or more of the nine mineral types (Fe, Ti, Mn, Mg, Ca, V, Cr, Si, Al, Zn, Nb, and Ta). The mineral types were: fresh ilmenite (TiO2 < 51%), altered ilmenite, magnetite, magnetite with other phases, hematite, ferric-ilmenite (ilmenite with < 50% exsolved hematite), titano-hematite (hematite with <50% exsolved ilmenite), titanomagnetite, and chromite. Special care was made to avoid altered parts of grains such as weathered rims. The analyzed grains in the source area data set were grouped by mineral type and clustered into composition groups for each mineral. The range for each element was about that of what was normally determined by replicate analysis (i.e., approximately ±5% of the analyzed value). More than 2000 different mineral composition types were recognized for all nine Fe oxide minerals among the 41 Circum-Arctic source areas, but fewer than 150 of these contained more than 10 grains and up to 70 grains. These larger composition groups account for more than 80% of the matched grains in shelf samples that were matched to these source areas. Groups with less than 3 grains were excluded unless they were the only grain groups for a particular mineral in a source area. The source groups were tested using discriminant function analysis (DFA) where the probability of group membership had to exceed 0.95 and the probability of each grain being closer than any other grain to the group centroid had to exceed 0.1. About 30% of the analyzed source area grains failed these tests and were not used. To match grains to a source area, each grain in a sea ice sample was compared to all source area composition groups for that mineral using DFA. The same group membership probability was used (0.95) but a much more conservative probability of 0.5 was used for closeness to the group centroid. This assures that each matched grain is nearly identical to the group centroid for each element and even closer in composition than some of the grains used to form that group. This reduces considerably the error associated with confusing similar source groups, but it also causes nearly 40% of the analyzed grains to fail in matching to a source. All of the computer programs used for grouping, testing the uniqueness of each group, and matching grains from core samples to these source groups have been developed in our lab. We have over 70,000 Fe oxide analyses in our data bank. Of these about 20,000 are from known source areas (including the East Coast of the US) and known geologic provinces. The rest are from deep-sea sediment cores that have received sediment from various sources. All of the known source samples have been grouped into unique composition types for that source area or region (river or limited coastal zone of use in provenance determination). Below is the range of values for each mineral analyzed in the large circum-Arctic data set to show the wide range of compositions that are possible for these Fe oxide minerals. Fresh Altered Titano- Hematite Titano- Ferro- Magne- tite Magne- tite Chromite Ilmenite Ilmenite Magne- tite Hematite Ilmenite + Others number 5147 1076 2188 256 415 1160 1654 991 75 TiO2 30.0-51.0 51.0-98.1 1.1-46.6 0-25.8 5.0-73.9 30.2-86.8 0-50.2 0-56.0 0-8.4 FeO 1.4-67.5 0.7-47.8 2.6-70.0 37.8-95.5 1.6-90.6 4.0-66.1 31.1-99.3 30.1-99.0 21.3-64.0 MnO 0-20.2 0-14.9 0-6.3 0-2.0 0-18.2 0-16.1 0-2.2 0-8.4 0.3-8.8 MgO 0-6.9 0-3.7 0-19.7 0-9.0 0-8.0 0-8.7 0-20.4 0-9.4 0.2-12.8 Si)2 0-51.0 0-15.8 0-48.5 0-60.2 0-20.7 0-20.4 0-30.5 0-24.1 0-7.6 Al2O3 0-25.6 0-8.9 0-19.7 0-8.0 0-14.4 0-12.7 0-16.0 0-9.3 0-18.3 Cr2O3 0-4.5 0-0.8 0-12.7 0-7.4 0-5.1 0-0.6 0-13.0 0-15.3 30.3-58.5 ZnO 0-10.0 0-6.4 0-6.0 0-8.7 0-8.9 0-9.0 0-25.0 0-6.6 0-6.4 V2O3 0-2.8 0-1.4 0-4.1 0-2.0 0-2.2 0-1.7 0-2.9 0-3.4 0-0.9 CaO 0-27.0 0-28.3 0-46.9 0-10.3 0-22.1 0-20.1 0-11.7 0-19.2 0-1.9 Nb2O5 0-5.4 0-4.7 0-6.7 0-2.3 0-5.5 0-2.2 0-3.2 0-3.3 0-1.9 TaO 0-0.9 0-1.1 0-2.1 0-0.7 0-1.3 0-0.7 0-1.1 0-0.8 0-0.8 References Bischof, J.A. and Darby, D.A., 1997, Mid to Late Pleistocene ice drift in the western Arctic Ocean: evidence for a different circulation in the past. Science, v.277: 74-78. Darby, D.A., 1998. Mysterious iron-nickel-zinc spherules. Canadian Journal of Earth Science, 35:23-39. Bischof, J. and Darby, D., 1999, Quaternary ice transport in the Candian Arctic and extent of Late Wisconsinan glaciation in the Queen Elizabeth Islands. Candian J. Earth Sciences, 36:2007-2022 Darby, D.A., Bischof, J., Spielhagen, R., Marshall, S., and Herman, S., 2002, Arctic ice export events and their potential impact on global climate during the late Pleistocene. Paleoceanography.(in press) Darby, D.A., 1990. Evidence for the Hudson River as the dominant source of sand on the U.S. Atlantic Shelf. Nature, v. 346, p.828-831. Darby, D.A. and Bischof, J.F., 1996. A statistical approach to source determination of lithic and Fe-oxide grains: an example from the Alpha Ridge, Arctic Ocean. Journal of Sedimentary Research, v. 66, n. 3, 599-607. Darby, D.A. and Y.W. Tsang, 1987. The variation in ilmenite element composition within and among drainage basins: implications for provenance. Journal of Sedimentary Petrology, 57(5):831-838. Fe Oxide Mineral Alteration (primarily ilmenite) 0 = Unaltered, TiO2 < 48%, R (reflectance) = 20%, anisotropic 1 = Stage 1 alteration, weak pleochroism, TiO2 48-52%, weak anisotropism, no internal reflections, R=18%. 2 = Stage 2 alteration, non-pleochroic, R=19%, no internal reflections, gray efforescence, TiO2= >52 <55%. 3 = Stage 3 alteration, non-pleochroic, R=20%, brown or reddish-brown internal reflections, TiO2= 55-<75%, light gray spots of cryptocrystalline TiO2 (cloudy appearance). 4 = Stage 4 alteration, yellow or white internal reflections, cloudy appearance, TiO2 = >75%, R=21-22%. Exsolution Type (only mineral types 4, 5, or 7) 00 = none 03 = flasers (<1( and 1-5(); (note: flasers are sigmoidal laminae, all dimensions are widths) 05 = lenses (>5() + flasers (<2() 06 = lenses (>5() + elongated blebs (<1() 07 = lenses (>5() + flasers (<1() 10 = straight laminae (<5() 11 = straight laminae (<5() in 2 directions 12 = straight laminae (<5() in 3 directions 13 = straight laminae (<5() in 4 or more directions 14 = lenses in 2 directions 20 = straight laminae (1-5() 21 = straight laminae (1-5() in 2 directions 22 = straight laminae (1-5() in 3 directions 23 = straight laminae (1-5() in 4 or more directions 24 = straight laminae (<1() 25 = straight laminae (<1() in 2 or more directions 26 = straight laminae (<1() and (1-5() 27 = straight laminae (<1() and (>5() 28 = straight laminae (1-5() and (>5() 29 = straight laminae and flasers (<2() 30 = lenses >5( 31 = lenses 1-5( 32 = lenses <1( 33 = lenses <1( and 1-5( 34 = lenses <1( and >5( 35 = lenses 1-5( and >5( 36 = straight laminae and lenses (any widths) 37 = lenses (<1() + rods (<1(); rods are elongated cylinders 38 = lenses ((1() + rods ((1() 39 = wedges ((10( at widest part) 40 = wedges ((10( at widest part) 41 = wedges ((10( at widest part) + straight laminae (<1() 42 = wedges ((10( at widest part) + straight laminae (1-5() 43 = wedges ((10( at widest part) + straight laminae (>5() 44 = wedges ((10( at widest part) + lenses (<1() 45 = wedges ((10( at widest part) + lenses (1-5() 46 = wedges ((10( at widest part) + lenses (>5() 47 = wedges ((10( at widest part) + flasers (<1() 48 = wedges ((10( at widest part) + flasers (1-2() 49 = wedges ((10( at widest part) + flasers (>2() 50 = flasers (<1() 51 = flasers (1-2() 52 = flasers (>2() 53 = flasers (<1() + straight laminae (<1() 54 = flasers (1-5() + straight laminae (1-5() 55 = flasers (<1() in 2 directions 56 = flasers (1-2() in 2 directions 57 = flasers (>2() in 2 directions 58 = rods (<1() 59 = rods (>1() 60 = irregular elongated blebs (>2(), oriented 61 = irregular elongated blebs (<2(), oriented 62 = irregular elongated blebs (>2(), random orientation 63 = irregular elongated blebs (<2(), random orientation 64 = irregular elongated blebs (>2(), in 2 directions 65 = irregular elongated blebs (>2(), in 3 or more directions 66 = straight laminae (>2() with irregular elongated blebs within 67 = straight laminae (2-5() with irregular elongated blebs within 68 = wedges (>10() with irregular elongated blebs within 69 = flasers (>2() with irregular elongated blebs within Inclusion Types 00 = none 70 = straight laminar voids (<1() 71 = straight laminar voids (1-5() 72 = straight laminar voids (>5() 73 = flaser voids 74 = lensoid voids (<5() 75 = lensoid voids (>5() 76 = rod-shaped voids 77 = irregular voids or pits 80 = any silicate 90 = apatite 91 = spinel 92 = ulvospinel 93 = ulvospinel + straight laminae 94 = ulvospinel + lenses 95 = ulvospinel + flasers 96 = ulvospinel + wedges 97 = ulvospinel + voids 98 = other inclusions 99 = unknown inclusions