The sand collected by the hammer drill (CSIRO 142807) provides our only tangible indication of what occurred in the topmost cased interval of Hole 1189B, though it is not known whether it represents the entire interval or only its lower section. During the hammer-in casing operation, penetration was fast (7 m/hr) from seabed to 7 meters below seafloor (mbsf), and then slow (1 m/hr) between 7 and 8 mbsf. It sped up from 8 to 10 mbsf, but slowed to <1 m/hr between 10 and 15 mbsf before suddenly increasing remarkably to >30 m/hr until the casing operation ended at 30 mbsf (Shipboard Scientific Party, 2002). Although it seems reasonable to infer that the hard layers were unaltered dacite, no such material is present in the sand. Rather, the sand contains three main kinds of particle: (1) soft altered dacite fragments, (2) compact sulfide aggregates dominated by coarse pyrite euhedra that more closely resemble massive sulfide veins from the stockwork zone than the seafloor chimneys (where pyrite is scarce and more commonly fine grained and botryoidal), and (3) aggregates of coarse anhydrite. This suggests that whatever part of the cased interval the sample represents was basically similar to the underlying stockwork zone cored from 31 to ~100 mbsf in Hole 1189B. However, the bulk geochemistry of the sand indicates that sphalerite and barite, as well as Pb and Au, are significantly more abundant somewhere in the cased interval than in the semimassive sulfide core sample from immediately below the casing.

From an expedition undertaken after Leg 193 at Roman Ruins, in the vicinity of Hole 1189B, Petersen et al. (2003) report a variety of massive sulfide types in shallow diamond cores drilled as far as 5 mbsf, all highly enriched in base and precious metals. These include chimney fragments and apparently resedimented sulfide material plus nodular breccias with chalcopyrite-pyrite clasts set in an anhydrite matrix. The latter are remarkably similar to and likely cogenetic with the semimassive sulfide sample (CSIRO 142703) from immediately below the casing of Hole 1189B. Some holes bottomed in weakly to intensely clay-altered dacite, locally with stockworklike sulfide veining. Although a "missing link" remains, these results and the data for sample CSIRO 142807 collectively suggest that in Hole 1189B the sulfide-veined stockwork zone with fragments of altered dacite persists almost to the seafloor, perhaps with some thin intervals of harder, fresh, or less altered dacite and pods of semimassive sulfide resembling sample CSIRO 142703, and possibly with an upward-increasing abundance of sphalerite and barite. Rapid lateral change is indicated, however, by the cores of Hole 1189A drilled only ~35 m west-southwest from Hole 1189B adjacent to a chimney ~8 m lower on the Roman Ruins mound. Here, a thin interval of fresh dacite (represented by only 17 cm of recovered core) occurred directly below the 3-m jet-in test interval and passed gradationally downward into altered dacite lacking the sulfide stockwork (Shipboard Scientific Party, 2002). Geophysical logging of uncored Hole 1189C, drilled ~31 m east-southeast of Hole 1189B and 7 m lower on the opposite side of the mound crest from Hole 1189A, indicates only limited sulfide occurrence relative to the stockwork zone of Hole 1189B (Bartetzko et al., 2003). Taken together, results of the deep and shallow drilling establish that the stockwork zone in the upper part of Hole 1189B and related but unidentified subsidiary fractures represent the conduit for upward passage of high-temperature hydrothermal fluid responsible for growth of sulfide chimneys at Roman Ruins. However, the main conduit is clearly limited in lateral extent, at least in the direction of a section linking Holes 1188A, 1188B, and 1188C.

If account is taken of the predominance of pyrite in the Site 1189 subsurface massive and semimassive sulfides, their elemental constitution relative to Roman Ruins chimneys is consistent with a genetic relationship between the two groups. This is further indicated by isotopic data presented above, lending support to the interpretation (Shipboard Scientific Party, 2002) that the pyritic stockwork zone intersected in the higher part of Hole 1189B represents the subsurface plumbing system for hydrothermal fluids venting at the seafloor and forming the Roman Ruins chimneys. The mineralized pumice breccia of Hole 1189A (CSIRO 142701) occurs outside this stockwork, but its mineralogy, chemistry, and Pb and S isotopic composition suggest it also belongs to the conduit system, perhaps formed at its fringe by subhalative sulfide deposition in a formerly permeable volcaniclastic horizon.

Given exceptionally poor core recovery in the stockwork zone, the samples are too few for a thorough assessment of these relationships. Some samples from the stockwork zone possess thin quartz veins of uncertain temporal relationship to the more common massive to semimassive pyrite veins and breccia matrixes. One such vein in Sample 193-1189B-8R-1 (Piece 15, 68–70 cm; CSIRO 142710) contains significant chalcopyrite, in excess of pyrite, and disseminated chalcopyrite is also common in its adjacent selvage of silicified wallrock. In the lower sequence of Hole 1189B intersected below the stockwork zone, Sample 193-1189B-13R-1 (Piece 7, 35–38 cm; CSIRO 142717) contains a thin quartz vein with pyrite, sphalerite, lesser chalcopyrite, and significant barite. Pinto et al. (this volume) report additional examples of thin veins with chalcopyrite and sphalerite from the lower sequence, including one with native gold and one with galena. Mineralogically, these have a closer prima facie affinity with the Roman Ruins chimneys than the dominant massive to semimassive pyrite veins of the stockwork zone. Possibly, however, they represent an earlier phase in the hydrothermal history, or alternatively they were deposited in the fringes of the main Roman Ruins hydrothermal system. If the massive and semimassive sulfide veins of the stockwork indeed constitute the main subsurface conduits for the Roman Ruins system, then the results and discussion presented here suggest a pronounced vertical zoning arising from fractional crystallization, whereby pyrite precipitates at depth and the ascending fluid becomes progressively enriched in Cu, Zn, and Pb. Chalcopyrite then sphalerite commence precipitation at levels just below and within the cased interval of Hole 1189B, whereas Pb remains primarily in solution until formation of galena and sulfosalts immediately below the seafloor (lower mound) and in the chimneys. This is a similar order of mineral precipitation to that modeled by Bowers et al. (1985) for East Pacific Rise hydrothermal fluids during either conductive cooling or mixing with cold seawater. Subsurface phase separation as indicated by fluid inclusions in Leg 193 anhydrites (Vanko et al., 2004) will complicate this simplistic model which, nevertheless, offers an alternative explanation to "zone refining" for the uncommonly high contents of base and precious metals in Roman Ruins and other PACMANUS chimneys. A similar contrast between chemistry of chimneys and that of a subsurface stockwork zone was established by Ocean Drilling Program Leg 158 at the basalt-hosted Trans-Atlantic Geotraverse hydrothermal site (Herzig et al., 1998; Petersen et al., 2000).