fluids (e.g., East Pacific Rise, 21°N, Galapagos Spreading Center, 

 and Mid-Atlantic Ridge TAG site, Table 2 and references therein). 

 The chloride concentration of the normal Axial endmember is within 

 the range of variability normally ascribed to mineralogic control 

 (Von Damm, 1988). Silicate levels are consistent with quartz 

 solubility control (Von Damm and Bischoff, 1987) and calcium 

 appears slightly elevated relative to other hydrothermal fluids 

 with similar chloride concentrations. The normal Axial endmember 

 is represented by samples from the Inferno, Mushroom and Hell 

 vents. 



Mixing lines connecting the Virgin Mound data with the 

 seawater endmember define a second data group. All of the non- 

 volatile element concentrations are consistently and significantly 

 lower in this low-salinity endmember than in the normal endmember. 

 The chloride endmember concentration, for example, corresponds to 

 a chlorinity only one-third that of seawater and is significantly 

 lower than reported values for any other ridgecrest spreading 

 center. The Fe and Mn concentrations are also abnormally low and 

 preferentially depleted in Fe relative to Mn (Fig. 4). 



In contrast to the non-volatile elements, the volatile ele- 

 ments are highly enriched in the low-salinity fluids of Virgin 

 Mound. These fluids effervesced vigorously as they were expelled 

 from the titanium syringe samplers which were observed to leak gas 

 bubbles during ascent from dive 1733. As the only gas analysis 

 for the Virgin Mound data group is from one of these syringes, 

 caution must be exercised in the interpretation of the gas data. 

 Nonetheless, if we assume that most of the measured condensable 

 gas (1.28 cc g - *) is CO2, then the Virgin Mound CO2 endmember 

 concentration (173 mmol kg -1 ) is about 15 to 30 times that 

 estimated for the Galapagos Spreading Center and East Pacific 

 Rise, 21°N endmembers, respectively (Craig et al.,1980). The 

 H2S data from Virgin Mound and a related vent extrapolate to a 

 zero-magnesium endmember of approximately 11 mmol kg -1 , while 

 the Inferno and related vents endmember concentration is about 

 6 mmol kg -1 . 



The maximum observed discharge temperature at Axial Volcano 

 is 326°C, compared to the predicted boiling point of 348°C (Bis- 

 choff and Rosenbauer, 1984 ) for the local seafloor pressure. The 

 proximity of measured temperatures to the boiling temperature 

 together with the elevated volatile content and depleted chloride 

 in samples from Virgin Mound are consistent with the hypothesis 

 that the hydrothermal fluid rising through the ocean crust 

 underwent subcritical phase separation, which generated a low 

 salinity vapor phase. This vapor phase then must have become 

 partially segregated from its parent hydrothermal fluid to even- 

 tually exit the seafloor at the anomalous Virgin Mound vent. That 

 the phase separation occurred late in the circulation cycle, 

 temporally and spatially close to venting, is supported by the 

 proximity of the measured temperature and pressure to the boiling 



43 



