The southern Pacific Ocean offers the rare possibility to study a situation where a spreading ridge (the Pacific-Antarctic Ridge (PAR)) migrates toward a fixed hot spot (the Louisville hot spot) (Small, 1995). Hollister Ridge is a 450 km long linear structure whose position, between the PAR axis and the most recent edifices of the Louisville hot spot trail, led some authors to suggest that the ridge is genetically related to the hot spot (Small, 1995; Wessel and Kroenke, 1997). Mapping and sampling of the ridge in 1996 revealed, however, that the contribution of the Louisville plume material to its mantle source is minor and suggested that it might be the result of intraplate deformation (G¿li et al., 1998; Vlast¿lic et al., 1998). We report new, highly precise Pb isotopic data from Hollister Ridge, which (1) confirm that the maximal contribution of the Louisville plume, in the centrally, volcanic active part of the ridge, probably does not exceed 20% (15 and 35% for lower and upper limits) and (2) reveal through time an increasing plume influence. The initiation of the Louisville plume involvement in the source of Hollister Ridge is estimated to have occurred between 1.04 and 0.77 Myr ago. It thus followed closely the most recent volcanic activity reported along the Louisville trail (1.11 Ma (Koppers et al., 2004)). This suggests that Hollister Ridge has recorded the dispersion of the Louisville plume as the spreading ridge approached the hot spot. Assuming that the Louisville hot spot is located near the youngest seamount dredged along the Louisville seamount chain, Hollister Ridge lies along the shortest path of pressure release connecting the hot spot to the spreading axis. This path involves, first, an abrupt upwelling across the Eltanin fault system and, subsequently, a more progressive migration toward the spreading axis. Because Hollister Ridge is older than 2.5 Ma, the structure might not be the consequence of the plume-ridge flow. Instead, Hollister Ridge most likely emplaced through a lithospheric crack (G¿li et al., 1998), which, subsequently, may have captured the plume-ridge flow.