26. J. P. Cleveland, S. Manne, D. Bocek, P. K. Hansma, Rev.Sci. Instrum. 64, 403 (1993).27. W. A. Ducker, T. J. Senden, R. M. Pashley, Langmuir 8,1831 (1992).28. D. K. Newman, R. Kolter, Nature 405, 94 (2000).29. T. J. DiChristina, E. F. Delong, J. Bacteriol. 176, 1468(1994).30. A. S. Beliaev, D. A. Saffarini, J. Bacteriol. 180, 6292(1998).31. I. Sokolov, D. S. Smith, G. S. Henderson, Y. A. Gorby,F. G. Ferris, Environ. Sci. Tech. 35, 341 (2001).32. W. Stumm, J. J. Morgan, Aquatic Chemistry: ChemicalEquilibria and Rates in Natural Waters (Wiley, NewYork, ed. 3, 1996), pp 534–540.33. S. Brown, Proc. Natl. Acad. Sci. U.S.A. 89, 8651(1992).34. N. Ohmura, K. Kitamura, H. Saiki, Appl. Environ.Microbiol. 59, 4044 (1993).35. T. J. Beveridge, J. Bacteriol. 181, 4725 (1999).36. On the basis of transmission electron microscopy andfreeze-substitution analyses, extracellular polysac-charide, a common macromolcule on the surface ofmany bacteria, has not been detected on the cell wallof S. oneidensis (A. Korenevsky and T. Beveridge,unpublished data).37. P. J. Flory, Statistical Mechanics of Chain Molecules(Hanser, New York, 1989), pp. 401–403.38. M. Rief, M. Gautel, F. Oesterhelt, J. M. Fernandez, H. E.Gaub, Science 276, 1109 (1997).39. H. Mueller, H.-J. Butt, E. Bamberg, Biophys. J. 76,1072 (1999).40. D. J. Muller, W. Baumeister, A. Engel, Proc. Natl. Acad.Sci. U.S.A. 96, 13170 (1999).41. J.-J. Karlsson, A. Kadziola, A. Rasmussen, T. E. Rostrup,J. Ulstrup, in Protein Folds: A Distance-Based Ap-proach, H. Bohr, S. Brunak, Eds. (CRC Press, BocaRaton, FL, 1996), pp. 56–67.42. J. M. Myers, C. R. Myers, Appl. Environ. Microbiol. 67,260 (2001).43. Future studies will use biological force microscopywith mutants incapable of producing and/or secret-ing cell wall biomolecules such as the 150-kD protein.44. G. V. Bloemberg, G. A. O’Toole, B. J. J. Lugtenberg, R.Kolter, Appl. Environ. Microbiol. 63, 4543 (1997).45. N. P. D’Costa, J. H. Hoh, Rev. Sci. Instrum. 66, 5096(1995).46. We thank B. Lower, C. Myers, J. Banfield, and theanonymous reviewers for constructive comments; G.O’Toole for providing plasmid pSMC2; and J. Tak forsupport. This manuscript is dedicated to B. Diehl.Mineral samples were provided by the Virginia TechGeological Sciences Museum. Financial support wasprovided by the U.S. Department of Energy and theNSF.5 February 2001; accepted 4 April 2001Ultrafast Source-to-SurfaceMovement of Melt at IslandArcs from226Ra-230ThSystematicsSimon Turner,1,2* Peter Evans,1,3Chris Hawkesworth1,2Island arc lavas have radium-226 excesses that extend to higher values thanthose observed in mid-ocean ridge or ocean island basalts. The initial ratio ofradium-226 to thorium-230 is largest in the most primitive lavas, which alsohave the highest barium/thorium ratios, and decreases with increasing mag-matic differentiation. Therefore, the radium-226 excesses appear to have beenintroduced into the base of the mantle melting column by fluids released fromthe subducting plate. Preservation of this signal requires transport to the surfacearguably in only a few hundreds of years and directly constrains the averagemelt velocity to the order of 1000 meters per year. Thus, melt segregation andchannel formation can occur rapidly in the mantle.The velocity of melt ascent from its sourcethrough Earth’s mantle and crust to thesurface is extremely hard to determine,even though such measurements wouldplace important constraints on the mecha-nisms of melt transport and the physicalbehavior of the mantle during partial melt-ing (1–5). Disequilibria between the short-lived U-series isotopes can provide esti-mates of melt velocities and have recentlybeen used to show that mantle melt veloc-ities are too fast for transport to occur bygrain-scale percolation mechanisms all theway to the surface. Instead, melt must, atsome critical stage, separate into discretechannels (1–5). What remains to be deter-mined is how quickly melt moves and atwhat melt fraction such channels form. Theshort-lived U-series isotope226Ra has ahalf-life of only 1600 years and can be usedto estimate melt velocity and residual po-rosity (1–5). However, recent models formelt formation and transport beneath mid-ocean ridges and ocean islands emphasizethat226Ra-230Th disequilibria is createdthroughout the melting column and the dis-equilibria measured at the surface only re-flects that produced in the upper portions ofthe melting column (3–5). Thus, although226Ra disequilibria in mid-ocean ridge andocean island basalts (MORB and OIB, re-spectively) provide important constraintson the residual porosity, they do not placetight constraints on the melt velocity be-cause the depth of origin of the disequilib-ria is debatable.No global226Ra study has been conductedon island arc lavas since the pioneering in-vestigation of Gill and Williams (6 ), whosuggested that226Ra excesses might be relat-ed to fluid addition. Accordingly, we haveundertaken high-precision, mass spectromet-ric measurements of226Ra disequilibria in40 historic lavas from seven island arcsthat, combined with recent data from theTonga-Kermadec arc (7 ), allow us to eval-uate the global226Ra-230Th systematics inisland arc rocks. The (226Ra/230Th)ovalues(where the parentheses denote activity ra-tios) extend to226Ra excesses of over 600%and show a significant variation above eacharc (Table 1), ranging from 0.87 to 2.66 inthe Lesser Antilles, 1.25 to 4.96 in Vanu-atu, 1.51 to 1.59 in the Philippines (twosamples only), 1.06 to 5.44 in the Marianas,1.39 to 4.16 in the Aleutians, 1.00 to 3.53in Kamchatka, 0.94 to 3.71 in Indonesia,and 0.93 to 6.13 in Tonga-Kermadec (7 ).Only one sample has (226Ra/230Th)o⬍ 1,and five samples are within analytical errorof 1.The (226Ra/230Th)oversus (238U/230Th)diagram (Fig. 1) illustrates that the U-seriesdisequilibria in island arc lavas are distinctfrom MORB and OIB, extending to higher(226Ra/230Th)oratios and having the reversesense of U-Th fractionation. Thus, the islandarc lavas with the highest (226Ra/230Th)oarecrudely correlated with the highest (238U/230Th) (8–11). Some aspects of these differ-ent tectonic regimes are similar, in that melt-ing is initiated at depths of 100 km and thetotal extent of melting is probably in therange 8 to 15%. However, for MORB andOIB, melting occurs in response to decom-pression. The230Th excesses [i.e., (238U/230Th) ⬍ 1] reflect the greater compatibilityof U, relative to Th in residual aluminousclinopyroxene and