From a detailed survey and sampling study of corrugated massifs north of the Fifteen-Twenty Fracture Zone on the Mid-Atlantic Ridge, we demonstrate that their surfaces are low-angle detachment fault planes, as proposed but not previously verified. Spreadingdirection-parallel striations on the massifs occur at wavelengths from kilometers to centimeters. Oriented drill-core samples from the striated surfaces are dominated by fault rocks with low-angle shear planes and highly deformed greenschist facies assemblages that include talc, chlorite, tremolite, and serpentine. Deformation was very localized and occurred in the brittle regime; no evidence is seen for ductile deformation of the footwall. Synkinematic emplacement of diabase dikes into the fault zone from an immediately subjacent gabbro pluton implies that the detachment must have been active as a low-angle fault surface at very shallow levels directly beneath the ridge axis. Strain localization occurred in response to the weakening of a range of hydrous secondary minerals at a very early stage and was highly efficient.

[1] Major problems in experimental studies of the flow of partially molten granitic rocks are (1) maintaining control of the grain size of the matrix of solid grains, (2) controlling the melt fraction, and (3) controlling the melt viscosity (closely linked to water content). To overcome these problems, we used a synthetic ''granitoid'' comprising a solid matrix of quartz grains (50 mm grain size) mixed with an albite-quartz melt prepared from oxides, with added water (2.5 wt %) dissolved in the melt phase (viscosity 9 Â 10 4 Pa s). Experiments were performed undrained, with melt fractions of 0.1, 0.2, 0.25, and 0.3. Constant strain rate, creep, and stress relaxation experiments were carried out in a gas medium apparatus at 1273 and 1173 K, mostly at 300 MPa confining pressure, up to $15% shortening strain. Strain rates ranged between 10 À4 to less than 10 À7 s À1 , and fully ductile mechanical behavior was observed. Melt fraction has a more profound effect on strength than water content, temperature, or total confining pressure. Low strain rate (<10 À6 s À1 ) data were fitted to an empirical flow law of the form de/dt = A exp(B f m ) exp(ÀH/RT) s n with the parameters logA = À1.39, m = 3, n = 1.8, B = 192, H = 220 kJmol À1 , where de/dt is strain rate (s À1 ), s is flow stress (MPa), f is melt fraction, and T is temperature (K). Strength is expected to fall rapidly above about 0.4 melt fraction to levels characteristic of flow of the melt containing suspended crystallites. Extrapolation to geological strain rates using this flow law shows that in nature, migmatites bearing granitic melt will be extremely weak, much weaker than silicate rocks deforming by intracrystalline plasticity. Microstructural study shows grains remain equant at all strains, with no discernable formation of a grain flattening fabric nor a large strain contribution from microfracturing; thus bulk intracrystalline plasticity is of minor importance, and intergranular sliding is implied. The nonlinear flow observed is tentatively attributed to a combination of sintering of grain contacts and sliding failure of such contacts, with a substantial contribution from diffusive transfer processes.

The provenance of the Numidian Flysch in the western Mediterranean remains a controversial subject which hinders understanding of this regionally widespread depositional system. The Numidian Flysch is a deep marine formation dated as Oligocene to Miocene which outcrops throughout the Maghreb and into Italy. Evidence that is widely used for provenance analysis has not previously been reviewed within the context of the Maghrebian Flysch Basin as a whole. The structural location within the Alpine belt indicates deposition proximal to the African margin, while the uniformity of the Numidian Flysch petrofacies suggests a single cratonic source, in stark contrast to heterolithic and immature flysch formations from the north of the basin. Detrital zircon ages constrain a source region with Pan-African and Eburnian age rocks, unaffected by either Hercynian or Alpine tectonic events, which precludes the European basement blocks to the north of the basin. Palaeocurrent trends which suggest a northern source are unreliable given foreland basin analogues and observed structural complications. An African craton source remains the only viable option once these data are reviewed in their entirety, and the Numidian Flysch therefore represents a major Cenozoic drainage system on the North African margin. Deposition is concurrent with regional Atlas uplift phases, and coincidental with globally cooling climates and high sea levels. The Numidian Flysch is therefore interpreted to represent a highstand passive margin deposit, with timing of deposition controlled primarily by hinterland uplift and climatic fluctuations.

No abstract

Cosmogenic 10 Be surface exposure ages for bedrock sites around Torridon and the Applecross Peninsula in Wester Ross, northwest Scotland, provide new insights into the Lateglacial transition. Accounting for postglacial weathering, six statistically comparable exposure ages give a late Younger Dryas (G-1) exposure age of 11.8 AE 1.1 ka. Two further outliers are tentative preYounger Dryas exposure ages of 13.4 AE 0.5 ka in Torridon, and 17.5 AE 1.2 ka in Applecross. The Younger Dryas exposure ages have compelling implications for the deglaciation of marginal Loch Lomond Stadial ice fields in Torridon and Applecross. Firstly, they conflict with predictions of restricted ice cover and rapid retreat based on modelling experiments and climate proxies, instead fitting a model of vertically extensive and prolonged ice coverage in Wester Ross. Secondly, they indicate that >2 m of erosion took place in the upper valleys of Torridon and Applecross during the Younger Dryas, implying a dominantly warm-based glacial regime. Finally, the exposure ages have clarified that corrie (cirque) glaciers did not readvance in Wester Ross, following final deglaciation.