Do icelandic alkali basalts really have normal mantle d 18 O? CHRISTINA J. MANNING,MATTHEW F. THIRLWALL, DAVID LOWRY Department of Geology, Royal Holloway University of London, Egham, Surrey TW20 0EX, UK A low d 18 O mantle source has recently been proposed to gen- erate at least some of the low d 18 O in Icelandic basalts (Thirlwall et al., 2006). Using laser flourination (LF) on olivine, normal mantle d 18 O (+5.2 ± 0.3&, (Mattey et al., 1994)) was only observed in highly incompatible element depleted lavas. However using conventional flourination, normal mantle values have also been reported for the only Icelandic alkaline lavas (Vestmannaej- yar and Snaefellsnes, (Sigmarsson et al., 1992, 1993)). d 18 O WR has been shown to decrease along the Southern Volcanic Zone towards the centre of Iceland (Sigmarsson et al., 1992), thought to reflect mixing between alkali basalts with normal mantle d 18 O, produced at the tip in Vestmannaeyjar, and low d 18 O crus- tally contaminated tholeiites from the rift zone in central Iceland. Within Snaefellsnes, a decrease in 87 Sr/ 86 Sr and increase in 143 Nd/ 144 Nd towards the centre of Iceland has been reported (Sig- marsson et al., 1992, 1993), consistent with a contribution from an old enriched normal d 18 O source at the tip of the peninsula. This is not easily reconciled with the low d 18 O ol (+4.2& (Thirlwall et al., 2006)) in primitive Reykjanes Peninsula samples with equal- ly low 143 Nd/ 144 Nd. Vestmannaejyar lavas yield normal mantle LF d 18 O ol values (+5.0 ± 0.2&, 2sd, N = 13). Compared with low d 18 O Fe–Ti bas- alts on the nearby mainland, their alkaline character is only visible in elevated K and Na. Other incompatible elements are similar at constant MgO, resulting in high K/Nb ratios (260  normal mantle). Low K/Nb in most basalts from Iceland cannot be a con- sequence of crustal contamination as K/Nb decreases northward along the Reykjanes Ridge. Low K/Nb probably reflects recycled ocean crust in the mantle source (Thirlwall et al., 1997). Vestman- naejyar lavas are thus not derived from normal Icelandic mantle, and their normal d 18 O ol cannot be used to support a contamina- tion origin for low d 18 O elsewhere in Iceland. Snaefellsnes lavas yield d 18 O ol (+4.6 ± 0.13&,2sd, N = 9) in the west where the lavas were regarded to be uncontaminated mantle melts (Sigmarsson et al., 1992). Despite being alkali bas- alts Snaefellsnes basalts have Sr–Nd signatures very similar to low d 18 O enriched Reykjanes tholeiites (Thirlwall et al., 2006). Like most Icelandic lavas, they have low K/Nb, indicating that they are derived by small degree partial melting of low d 18 O enriched Icelandic mantle. References Hemond et al., 1993. J. Geophys. Res. 98, 15833–15850. Mattey et al., 1994. Earth Planet Sci. Lett. 128, 231–241. Sigmarsson et al., 1992. Earth Planet Sci. Lett. 110, 149–162. Thirlwall, 1997. Chem. Geol. 139, 51–74. Thirlwall et al., 2006. Geochim. Cosmochim. Acta 70, 993–1019. doi:10.1016/j.gca.2006.06.783 What’s so super about supercritical fluids in subduction zones? C.E. MANNING Department of Earth and Space Science, University of California, Los Angeles, CA 90095, USA (manning@ess.ucla.edu) It is commonly assumed that fluids in subduction zones have properties intermediate between hydrous silicate melt and H 2 O. Such supercritical, intermediate fluids are potentially important because they could dominate mass transfer to the mantle wedge. However, the existence and longevity of such fluids is problematic (Manning, 2004). Albite-H 2 O is typically used as an analogue sys- tem, but the favorable position of its critical curve (Shen and Kep- pler, 1997) is not representative; critical curves for basalt and peridotite lie at substantially higher P (Kessel et al., 2005,). In addition, low-porosity natural systems only coexist with interme- diate fluids over a restricted PT interval. Finally, intermediate flu- ids, if formed, will exist over short length scales as composition shifts during reactive flow in the mantle wedge. Although supercritical fluids probably do not play a major role in subduction-zone metasomatism, their chemistry holds a clue to understanding high-P mass transfer by water-rich solutions. Full miscibility can only occur by progressive polymerization of dis- solved Si, Al, Na, and other metals. This behavior yields, e.g., aqueous Si–Si, Si–Al, and Si–Na–Al oxide polymers of varying stoichiometry. High-P experimental studies indicate the presence of polymeric complexes even in subcritical, dilute, H 2 O-rich fluid (e.g., Newton and Manning, 2003; Manning, 2006). Silicate poly- mers in these solutions enhance the solubility of minor elements such as Ti, P, and Zr in high-P fluids (Antignano and Manning, 2005), probably by substitution into their more energetically favorable oxygen-coordinated sites. Subcritical silicate polymeri- zation thus affords a mechanism for mobilization of nominally low-solubility components. Because they form over a wider PT and bulk-compositional range, subcritical silicate polymers in dilute solutions are likely responsible for more mass transfer in subduction zones than intermediate, supercritical fluids. References Antignano, A., Manning, C.E., 2005. Eos 86 (52), V31C-0620. Kessel, R., Ulmer, P., Pettke, T., Schmidt, M.W., Thompson, A.B., 2005. Earth Planet Sci. Lett. 237, 873–892. Manning, C.E., 2004. Earth Planet Sci. Lett. 223, 1–16. Manning, C.E., 2006. J. Geochem. Explor. 89, 251–253. Mibe, K., Kanzaki, M., Kawamoto, T., Matsukage, K.N., Fei, Y., Ono, S., 2005. Eos 86 (52), V33C-03. Newton, R.C., Manning, C.E., 2003. Contrib. Mineral Petr. 146, 135–143. Shen, A., Keppler, H., 1997. Nature 385, 710–712. doi:10.1016/j.gca.2006.06.784 A388 Goldschmidt Conference Abstracts 2006