Any genuine hydrothermal fluid passing out of fracture conduits will be principally mixed within the Leg 193 boreholes by remnant drilling fluid (surface seawater) and near-seabed seawater drawn into the holes after drilling or already added by circulation deeper within the PACMANUS hydrothermal system. Our end-member calculations treat these as a single component. Other sources of contamination include material leached from wallrocks by the borehole fluid and material dissolved from sepiolite or barite muds used to flush the holes. Leaching of Mg from sepiolite would affect our estimates of seawater components but cannot be evaluated, although we assume the effect is negligible. Low Ba contents of the samples preclude significant contamination by barite.

The high Ca contents calculated for the end-member borehole fluids, around 6 or 7 times those of the end-member chimney vent fluids, suggest there has been some leaching of anhydrite from wallrock veins and breccias at the measured temperatures of collection. This would also contribute S and Sr to the borehole fluids. The apparent excess of S in the two more successful end-member borehole fluids (confirmed as sulfate by shipboard analyses and the lack of H2S odor), relative to total S (SO42– and H2S) in the end-member chimney vent fluids, is indeed only slightly below the stoichiometric equivalent of Ca in anhydrite. Trace Sr in Leg 193 anhydrites varies from 207 to 4568 ppm, averaging 2630 ppm for 59 samples (Roberts et al., 2003). To explain the excess Sr in the two end-member borehole fluids this way would require dissolution of anhydrite containing 1000–1400 ppm Sr, well within the measured range.

The 87Sr/86Sr ratios measured on vein anhydrites at Sites 1188 and 1189 together with depth profiles implying higher degrees of seawater mixing for anhydrite-crystallizing fluids close to the seafloor (Roberts et al., 2003) allow further inferences if indeed anhydrite dissolution has contributed to the borehole fluids. The calculated end-member 87Sr/86Sr (0.7062) for the borehole fluid sampled at 130 mbsf in Hole 1189B (CSIRO 142727) is close to the average value of 0.7065 measured in vein anhydrites from 118 to 158 mbsf within the lower sequence (Shipboard Scientific Party, 2002b) of Hole 1189B. While this might be coincidental considering anhydrite variability in this interval (87Sr/86Sr = 0.0006; n = 16), it hints that more than just the excess Ca and Sr could have dissolved from anhydrite. For the sample collected at 3 mbsf in Hole 1188B (CSIRO 142724), however, the end-member 87Sr/86Sr estimate (0.7064) is distinctly lower than the ratios in anhydrites collected from the upper 50 m of nearby Hole 1189A (average = 0.7078; = 0.0009; n = 6), but closer to ratios for anhydrites below 50 mbsf in Holes 1188A and 1188F (average = 0.7061; = 0.0005; n = 51). This strongly suggests that borehole fluid sampled near the seabed in Hole 1188B was ascending from far greater depths.

The REE abundances and chondrite-normalized patterns of anhydrites from Sites 1188 and 1189 are extremely variable (Bach et al., 2004). Concentrations of Nd range from 0.08 to 28.3 ppm, and the patterns include light-REE enrichment and depletion or mid-REE enrichment and depletion, with some negative but mostly positive Eu anomalies. Again using apparent excess Ca, even the maximum Nd measured in anhydrite would contribute only 0.24–0.30 ppm Nd if dissolved into the end-member borehole fluids, distinctly below their computed concentrations of this element (Table T2). The abundances and chondrite-normalized patterns of REE in the borehole fluids, relative to chimney vent fluids, remain unexplained. Leaching from altered wallrocks is unlikely since these mostly show pronounced Eu depletion (see Fig. F19B in Binns, this volume [Manuscript 211]).

Aside from the anhydrite dissolution and REE issues, the computed end-member composition of the borehole fluid sampled at 130 mbsf in Hole 1189B (CSIRO 142727; Roman Ruins) is remarkably similar to those of the end-member vent fluids from Satanic Mills chimneys for most elements, exceptions being its lower Al, somewhat lower Fe and Mg, and, among the chalcophile elements, higher Cd and lower Mo and Pb. Contents of Cu and Zn are similar. The general resemblance suggests that the hydrothermal component within the borehole fluid is dominated by a substantial portion of vent-type fluid. However, the lack of H2S in the former requires there to have been some oxidation of the latter. The particularly low U content of the original borehole fluid relative to seawater indicates this element has been extracted from its hydrothermal component. Although U is normally mobile under oxidized conditions and deposited by reduction to UO2, this element is relatively abundant in low-temperature Fe-Mn-Si oxide deposits at PACMANUS (see Table T2 in Binns, this volume [Manuscript 211]), possibly as a result of adsorptive scavenging. Postdrilling oxidation and deposition of oxides within or near the borehole might cause lower Fe, Mn, and U contents in the collected sample.

Similarity of the "hydrothermal" component to chimney vent fluids is less evident for the sample from 3 mbsf in Hole 1188B (CSIRO 142724). As calculated, with inherent errors arising from the effect of dilution by deionized water on the estimated hydrothermal component, it is exceptionally rich in Mn, Zn, Mo, Cd, Sb, and La; depleted in K, Fe, and Pb; and again deficient in U. Some of these characteristics, as well as negative Ce anomalies, are shared by Mn-rich oxide deposits dredged at and near Snowcap (Site 1188), to which the Hole 1188B borehole fluid may have a greater affinity than sulfide chimneys, as a consequence of subsurface reactions that can not at present be quantified.

The Mn/Fe ratio in the borehole fluid sampled at 3 mbsf in Hole 1188B (CSIRO 142724) is exceptional. Figure F3 plots abundance of these elements in end-member components for two sets of samples from vents 1 and 2 at Satanic Mills (Douville, 1999; Gamo et al., 1996; T. Gamo, pers. comm., 1996), together with samples (more diluted by seawater) collected at a third Satanic Mills chimney and one at Roman Ruins (Gammo et al., 1996; T. Gammo, pers. comm., 1996), plus a sample from a Tsukushi chimney (C.J. Yeats, pers. comm., 2001). All these show limited variation in Mn/Fe ratio, although the Tsukushi composition is somewhat richer in Fe. Also plotted in Figure F3, without extrapolation to zero Mg, are total dissolvable Mn and Fe in samples collected by VUNL at Snowcap warm seeps and by hydrocast from the "eye" of the particulate plume directly above PACMANUS. These latter are highly diluted by seawater but are considered, respectively, to reflect the Mn/Fe ratios in diffusely vented fluids at Snowcap and the residue of chimney fluids vented at Satanic Mills and Roman Ruins. They possess Mn/Fe ratios comparable with the chimney vent fluids, as do the highly diluted borehole fluids collected at 107 and 207 mbsf in Hole 1188F (CSIRO 142725 and 142726). The near-seabed borehole sample from 3 mbsf in Hole 1188B (CSIRO 142724) falls well off the general trend.

The above discussion is clearly based on imperfect samples of borehole fluids from PACMANUS and on perhaps ambitious extrapolation from analytical data for the raw samples. The results provide additional information requiring consideration in overall interpretations of the PACMANUS hydrothermal system, but they do not permit quantitative assessments regarding subseafloor fluid-rock interactions or input of magmatic fluids, as we hoped prior to conducting Leg 193. They establish a need to develop improved technology for collecting higher-temperature fluids during future ocean drilling focused on submarine hydrothermal systems.