At a number of locations along the Pacific margin of the Americas and west Antarctica, small volumes of alkalic basalts were erupted following successive ridge crest-trench collisions. The basalts were generated as a result of upwelling of asthenosphere through windows in the subducted plate. There is no evidence for local high temperature mantle plumes or significant lithospheric extension associated with these basalts. For the best sampled area, the Antarctic Peninsula, mean values for fractionation-corrected iron content (FeO*) vary from 6.9 to 10.6 wt.%, and for Na20 (NAB.0) from 3.25 to 4.6 wt.%, implying generation of small meltfractions at variable mean pressures. The results of rare earth element inversion modelling yield a melt generation interval of 100 to 52 kin, with a maximum melt fraction of c. 7% generated from a MORB-like source at Tp 1300~ Trace element and isotope systematics are also consistent with the generation of the basalts from a MORB-like source.Mean pressures of melt generation increase with increasing distance from the original trench, but trace element and Nas.0 data suggest that there is no systematic variation in extent of melting with distance from the trench. The data are consistent with a model whereby a MORB melting column, initially intersecting the peridotite solidus at between 15 and 30 kbar, is truncated by a lithospheric cap which thickens from c. 15-20 km (~ 5-8 kbar) up to a maximum of c. 50 km (~ 15 kbar), such that the mean pressures of melting increase with increasing distance from the palaeo-trench. The MORB-like major element geochemistry of basalt samples closest to the ancestral trench are consistent with initial intersection of the peridotite solidus at low pressures (c. 15 kbar). For areas of thickest lithosphere, mean pressures of melt generation are higher, but extents of melting are lower than for a MORB melting column with a similar initial pressure of intersection of the peridotite solidus. All these basalts therefore potentially represent analogues for the small melt-fraction precursors to the generation of MOR tholeiites.Thermal constraints suggest that these low volume, small melt-fractions, were generated with CO2 on the solidus, because mean pressures of melt generation are greater than the pressure of intersection of the T v 1300~ mantle adiabat and the dry peridotite solidus. Potentially, all MORB may be generated initially with CO2 on the solidus, and if this is correct, it does not require thermal anomalies to generate large extents of melting at high mean pressures.