Supplementary Methods 1. U-Pb isotope SIMS analysis of zircons The U-Th-Pb data (Supplementary Table 2) was obtained from polished, gold-coated zircon grain mounts (epoxy resin) with a SHRIMP II housed at the Curtin University of Technology, Perth. Operating conditions and analytical protocols are essentially the same as described by ref. 1. The Pb/U calibration was performed relative to the in-house standard zircon CZ3 (564 Ma), which was analysed repeatedly throughout each session. Apparent magmatic ages in Supplementary Table 2 include uncertainty in the Pb/U calibration. All errors are quoted at 1 precision. 2. Oxygen isotope analysis of zircon Oxygen isotope data (Supplementary Table 3) were acquired with a Cameca ims 1270, using a 6 nA primary 133Cs+ beam and charge compensation by normal-incidence electron gun. The method closely followed that outlined by ref. 19. Secondary ions were extracted 18 at 10 kV, and O- and 16 O- were monitored simultaneously on dual Faraday cups. Each analysis involved a pre-sputtering time of 45 seconds, followed by data collection in two blocks of five cycles, amounting to a total count time of 40 seconds. Data were collected over two separate analytical periods using slightly different analytical protocols. In the first period (15-20 November 2004), a fixed primary beam was focussed directly onto the sample, sputtering material from an oval area measuring ~25 m in the longest dimension. The secondary yield of 18O under these conditions was typically between 2.5 x106 and 3.0 x 106 counts. Kohler illumination was employed in the second period (25-27 April 2005), which produced a slightly smaller and more regular, flat-bottomed pit and improved sensitivity for 18O (average 4.0 x106 counts). To correct for instrumental mass fractionation (IMF), all data were normalised to an internal standard, Geostandards zircon 91500 (18O 10.07 ± 0.03‰ VSMOW, 0.66 wt% HfO2, ref. 16), which was assumed to be homogeneous under the analytical conditions employed. Chips of this zircon were embedded into each grain mount and analysed (typically in blocks of ten) to bracket every 10-15 measurements on the sample zircons. For the November 2004 analytical period, the internal precision on each standard analysis based on counting statistics was typically between ±0.3‰ and ±0.8‰ (2 s.e.) and spot-to-spot reproducibility ~0.25-0.40 ‰ (1 s.d.). The internal precision was marginally improved for the April 2005 period (average 0.4 ‰, mostly < 0.5 ‰ at 2 s.e.), but the external precision based on replicate analysis of the same standard chip remained ~ 0.3 ‰ (1 s.d.). A potential source of error with in situ oxygen isotope analysis by ion microprobe concerns a matrix-induced 18O/16O fractionation resulting from different zircon Hf contents (Peck et al. 2001, and see summary in ref. 16). For the Cameca 4f, operating at high voltage offset, this can reach 1 ‰ for every 1 wt % HfO2 (Peck et al. 2001), although this effect has yet to be rigorously appraised for multi-collector ion microprobes using zero off- set. For this reason, in both the November 2004 and May 2005 periods, analyses of the KIM-5 zircon standard (1.23 % HfO2, 18O = 5.09 ± 0.06 ‰, ref. 16) were interposed throughout each session as an independent monitor on data quality. Analyses during November 2004 and April 2005 yielded mean values of 5.14 ± 0.28 ‰ (1 s.d., n = 68) and 5.15 ± 0.28 ‰ (1 s.d., n = 34) respectively, both indistinguishable from the laser fluorination value (18O = 5.09 ± 0.06 ‰) quoted by ref. 16. This demonstrates that matrix-related fractionation is negligible over this HfO2 range, which, fortunately, brackets most of the spread of HfO2 values observed within the detrital and inherited zircons analysed in this study. One grain mount analysed in April 2005 also contained the multigrain zircon standard Temora 2 (1 % HfO2). The average 18O of 23 analyses from these grains was 8.25 ± 0.20 ‰ (1 s.d.), indistinguishable from the laser fluorination value quoted by ref. 16 (8.20 ± 0.01 ‰). Oxygen isotope data for both standards and unknowns are listed chronologically in Supplementary Table 3. The session divisions variously reflect primary beam restarts, specimen exchanges, or major adjustments to magnet calibration, electrostatic lens settings or the primary beam steering or intensity. Note that only analyses relevant to this paper are tabulated. Fractionation-corrected oxygen isotope data are quoted at 2 standard errors and include the propagated standard deviation of the 91500 standard analyses to account for any instrumental drift. For unknowns, the fractionation factors used to correct raw 18O/16O data were calculated from bracketing analyses of the 91500 standard. When the instrument was stable, assessed by comparison between standard datasets acquired over the course of the session, the fractionation factor was derived from pooled standards. In cases where standard analyses showed significant drift, unknowns were corrected to the average fractionation factor derived from the immediate bracketing standard set. For each block of 91500 standard analyses, the table also lists the per mil deviation of the average measured 18 O/16O from the accepted ratio reported by ref. 16. This gives a convenient measure of the magnitude of the IMF and its variation during and between sessions. 3. Lu-Hf isotope analysis of zircons by laser ablation All laser ablation Lu-Hf isotope analyses (supplementary Table 4) were conducted in the Department of Earth Sciences, University of Bristol, using a 193 nm ArF laser and Finnigan Neptune, with the same Faraday cup configuration employed by Woodhead et al. (2004). The method is outlined by Hawkesworth & Kemp (2005). Data were acquired over two analytical sessions (27-30 November 2004 and 2-3 May 2005) using either a 40 m, 50 m (predominantly) or 65 m beam size and 4 Hz laser pulse repetition rate, producing total Hf beams of ~8-28 V. To counteract ArF degradation the power density at the sample was maintained at around 6-7 J/cm2, which translated into an estimated drilling rate of ca. 0.5 m/sec. Ablation was conducted in He (flow rate ~1.3 l/min, optimised daily), this being combined with argon (~0.9 l/min) in a small glass mixing chamber prior to transport into the ICP. A small (~0.005 l/min) N2 flow was introduced into the Ar carrier gas upstream of the mixing chamber, which suppressed oxide formation in the plasma associated with the use of X-skimmer cones and enhanced sensitivity by a factor of two. Tests on pure REE solutions established that formation of nitride molecules is negligible. The 176Hf/177Hf and 176Lu/177Hf ratios quoted in supplementary Table 4 were derived from a 60 second ablation period, comprising 60 cycles of 1 s integration time. In the case of isotopic zoning or with intersecting cracks/inclusions, only the flattest, most stable portions of the time-resolved signal were selected for integration, performed off-line using a customised Excel spreadsheet. The correction for the isobaric interference of Lu and Yb on 176 Hf was performed in 'real time' as advocated by Woodhead et al. (2004). For Yb, this 171 involved monitoring the interference-free Yb during the analysis and then calculating the magnitude of the 176Yb interference using the 176Yb/171Yb ratio determined by Segal et al. (2003) (176Yb/171Yb = 0.897145). The much smaller Lu correction is performed in the same fashion by measuring 175Lu, and using 176Lu/175Lu = 0.02669 (De Bievre and Taylor, 1993). To correct for instrumental mass fractionation, Yb isotope ratios were normalised to 173 Yb/171Yb = 1.130172 (Segal et al. 2003) and Hf isotope ratios to 179 Hf/177Hf = 0.7325. The mass bias behaviour of Lu was assumed to follow that of Yb. The veracity of this protocol was checked by replicate analysis of a solution of known Hf isotope composition (JMC 475) that was spiked with variable amounts of pure Yb solution. Analyses of pure JMC 475 (diluted to 80 ppb Hf) were interspersed with the Yb-doped JMC 475 solutions, 176 yielding an average Hf/177Hf of 0.282152 ± 6 (2s.d., n = 16), slightly lower than the commonly cited value of 0.282160 (e.g. Söderlund et al. 2004) but within the range quoted for other laboratories (e.g. 0.282144 ± 14, Vervoort et al. 2004; 0.28216 ± 1, ref. 31). The results are summarised in Supplementary Figure 1, where the interference-corrected 176 Hf/177Hf values were calculated using three recently published estimates of Yb isotope composition (Chu et al. 2002, Segal et al. 2003 and Vervoort et al. 2004) and the Yb isotope abundance ratios quoted by IUPAC. This shows that the interference correction using the Yb isotope ratios of Segal et al. (2003) accurately reproduce the Hf isotope composition of pure JMC 475 over a large range of Yb/Hf ratios that greatly exceeds that encountered in zircons of this study (Yb/Hf < 0.077). Supplementary Figure 1. Analyses of JMC 475 solutions spiked with variable amounts of Yb, where the interference corrected 176Hf/177Hf ratios are calculated using the Yb isotope compositions of IUPAC (De Bievre and Taylor, 1993; grey diamonds), Chu et al. (2002) (open circles), Segal et al. (2003) (black squares) and Vervoort et al. (2004) (open triangles). The shaded band bound by dashed lines depicts the Hf isotope composition of pure JMC 475 solutions measured during the same analytical session (176Hf/177Hf = 0.282152 ± 6, 2 s.d. n = 16), and the solid bar shows the range of Yb/Hf ratios encountered within zircons of this study (Yb/Hf 0.01-0.077). 176 Supplementary Table 5 summarises the mean interference-corrected Hf/177Hf ratios of the zircon standards utilised at Bristol, measured between August 2004 and May 2005. Internal precision routinely varies between ± 0.000010 and ± 0.000030 (at 2 s.e. level), being consistently the best for Sri Lankan zircon CZ3, reflecting the greater signal intensity and higher Hf concentration. External precision was ± 0.000019 for 91500, ± 0.000016 for Mud Tank (66 and 58 ppm, respectively) and ± 0.000024 for Temora 2 (86 ppm, all at 2 s.d.). The laser ablation data on these three zircon standards are 176 indistinguishable to the solution Hf/177Hf data of Woodhead et al. (2004). Solution data are as yet unavailable for CZ3, but our data are similar to the laser ablation data quoted by Ping et al. (2004) using the same instrumentation. To monitor data quality, analyses of standards (especially Temora 2, which has the highest Yb/Hf ratio) were regularly interposed with 'unknowns' during the course of the analytical session. Mass bias, peak shape/centre characteristics and Hf sensitivity showed little variation throughout each analytical session. 176 Standard zircon Hf/177Hf meas (laser) 176 Hf/177Hf meas (solution) 91500 0.282299 ± 19 (n = 71) 0.282306 ± 4 Temora 2 0.282687 ± 24 (n = 95) 0.282686 ± 4 Mud Tank 0.282496 ± 16 (n = 49) 0.282507 ± 3 CZ3 0.281697 ± 12 (n=15) 0.281703 ± 24 (n=16)# Supplementary Table 5. Summary of Hf isotope data of standard zircons using the 176 instrumentation at the University of Bristol, and corrected for Yb interference using the Yb isotope compositions of Segal et al. (2003). All data were acquired over 60 seconds ablation, using a 50 mm spot and 4 Hz laser pulse rate. Solution data are from Woodhead et al. (2004). Data are quoted at 2 standard deviations. # laser ablation data from Ping et al. (2004) References De Bievre P. & Taylor, P.D.P. Table of the isotopic composition of the elements. Int. J. Mass Spectrom. Ion Processes 123, 149 (1993). Hawkesworth, C.J. & Kemp, A.I.S. Unravelling the record of crustal evolution using Hf and O isotopes in zircon. Chem. Geol. (in press). Peck, W. H., Valley, J. W., Wilde, S. A. & Graham, C. M. Oxygen isotope ratios and rare earth elements in 3.3 to 4.4 Ga zircons: Ion microprobe evidence for high 18O continental crust and oceans in the Early Archaean. Geochim. et Cosmochim. Acta 65, 4215-4229 (2001). Ping, X., Fuyuan, W., Liewen, XIE & Yueheng, Y. Hf isotopic compositions of the standard zircons for U-Pb dating. Chinese Sci. Bull. 49, 1642-1648 (2004). Segal, I., Halicz, L. & Platzner, I.T. Accurate isotope ratio measurements of ytterbium by multi-collector inductively coupled plasma mass spectrometry applying erbium and hafnium in an improved double external normalisation procedure. J. Anal. At. Spectrom. 18, 1217-1223 (2003). 176 Söderlund, U., Patchett, J.P., Vervoort, J.D., & Isachsen, C.E. (2004). The Lu decay constant determined by Lu-Hf and U-Pb isotope systematics of Precambrian mafic intrusions. Earth Planet. Sci. Lett. 219, 311-324. Woodhead, J.D., Hergt, J.M., Shelley, M., Eggins, S. & Kemp, R. Zircon Hf-isotope analysis with an excimer laser, depth profiling, ablation of complex geometries, and concomitant age estimation. Chem. Geol. 209, 121-135 (2004). Vervoort, J.D., Patchett, P.J., Söderlund, U. & Baker, M. The isotopic composition of Yb and the precise and accurate determination of Lu concentrations and Lu/Hf ratios by isotope dilution using MC-ICPMS. Geochem Geophys Geosyst. DOI 2004GC000721RR (2004).