EXTREME ND-ISOTOPE HETEROGENEITY IN THE EARLY ARCHEAN - FACT OR FICTION - CASE-HISTORIES FROM NORTHERN CANADA AND WEST GREENLAND

Sm-Nd data on rock suites from early Archaean provinces in northern Ca nada and West Greenland clearly demonstrate that tectonothermal (i.e. igneous, metamorphic, tectonic) processes which affected the rocks lon g after their formation produced open-system behaviour leading to effe ctive resetting of the Sm-Nd system accompanied by complete, or near-c omplete, Nd-isotope homogenisation. This means that extreme caution is required in interpreting highly variable initial epsilon(Nd) values o f ancient rocks in terms of long-standing regional mantle heterogeneit y or of complex mantle-crust interaction processes. In particular, cal culated initial epsilon(Nd), values based on high-precision zircon U-P b dates may be of little or no significance in terms of geochemical ev olution of early mantle and crust source regions. A striking example i s provided by the Acasta gneisses of northern Canada, with published S HRIMP U-Pb zircon dates in the range 3.6-4.0 Ga and apparent, initial epsilon(Nd) values in the range -4.8 to +3.6 (Bowring and Housh, 1995) . A combination of 34 published and new Sm-Nd whole-rock analyses for a wide range of rock types yields a regression (errorchron) age of 337 1 +/- 59 Ma (MSWD = 9.2), with initial epsilon(Nd) = -5.6 +/- 0.7. Whi lst the very negative initial epsilon(Nd) provides strong, independent support for the extreme age of the Acasta gneiss protolith, resetting of the Sm-Nd system at similar to 3.4 Ga renders calculation of initi al epsilon(Nd) based on the zircon U-Pb dates geologically meaningless . Analogous considerations for early Archaean Akilia enclaves and host Amitsoq gneisses of West Greenland suggest that their published range of initial epsilon(Nd) values of similar to -4.5 to +4.5 at similar t o 3.73-3.87 Ca (Bennett et al., 1993) may be unrealistically wide and, therefore, inappropriate for modelling upper-mantle heterogeneity. In an attempt to determine a realistic initial epsilon(Nd) value, we hav e regressed 58 published and new Sm-Nd data for two major rock units o f the Isua supracrustal belt (felsites and mica-schists), regarded as having a short-term crustal history combined with minimal Sm-Nd distur bance. The Sm-Nd regression yields an age of 3776 +/- 52 Ma (MSWD = 8. 2), with initial epsilon(Nd) = +2.0 +/- 0.6 This value is much closer to conventional depleted-mantle models (e.g., DePaolo et al., 1991) th an to the value of similar to +4.5 proposed by Bennett et al. (1993) f or this age range. Our observations suggest that there may have been n o major change in processes between early Archaean and more recent sty les of depleted-mantle evolution.