Axial bathymetry, major/trace elements, and isotopes suggest that the Pacific-Antarctic Ridge (PAR) between 56 ¿S and 66 ¿S is devoid of any hotspot influence. PAR (56--66 ¿S) samples have in average lower 87Sr/86Sr and 143Nd/144Nd and higher 206Pb/204Pb than northern Pacific midocean ridge basalts (MORB), and also than MORB from the other oceans. The high variability of Pb isotopic ratios (compared to Sr and Nd) can be due to either a general high μ (HIMU) (high U/Pb) affinity of the southern Pacific upper mantle or to a mantle event first recorded in time by Pb isotopes. Compiling the results of this study with those from the PAR between 53 ¿S and 57 ¿S gives a continuous view of mantle characteristics from south Pitman Fracture Zone (FZ) to Vacquier FZ, representing about 3000 km of spreading axis. The latitude of Udintsev FZ (56 ¿S) is a limit between, to the north, a domain with large geochemical variations and, to the south, one with small variations. The spreading rate has intermediate values (54 mm/yr at 66 ¿S to 74 mm/yr at 56 ¿S) which increase along the PAR, while the axial morphology changes from valley to dome. The morphological transition is not recorded by the chemical properties of the ridge basalts nor by the inferred mantle temperature which displays few variations (30--40 ¿C) along the PAR. Contrary to what is observed along the South-East Indian Ridge, PAR morphology appears to be controlled more by spreading rate rather than by mantle temperature. Much of the major and trace element variability results from segmentation control on the shallowest thermal structure of the mantle. The cold edge of a fracture zone seems to be more efficient when occurring in an axial dome context. It is expressed as an increase of the magnitude of the Transform Fault Effect along the valley-dome transition, resulting in a clear increase of trace element ratio variability (such as Nb/Zr). There is no strong evidence for the previously proposed southwestward asthenospheric flow in the area. However, this flow model could explain the intrasegment asymmetric patterns. ¿ 2000 American Geophysical Union