DISCUSSION R. Klemd Comment on the paper by Schmidt Mumm et al. High CO 2 content of ¯uid inclusions in gold mineralisations in the Ashanti Belt, Ghana: a new category of ore forming ¯uids? (Mineralium Deposita 32: 107±118, 1997) Received: 28 May 1997 / Accepted: 5 September 1997 The recent paper by Schmidt Mumm et al. (1997) discussing the ore-forming ¯uids in gold deposits of the Paleoproterozoic Ashanti Belt (Ghana) is a welcome addition to the understanding of Paleoproterozoic lode-gold deposits in the Birimian of West Africa. The authors propose on the basis of ¯uid inclusion studies only, that the gold mineralisation of the Ashanti Belt is probably derived from a CO 2 -rich, almost water-free ¯uid. They further imply that the mineralisation in the various gold deposits took place under drastically dierent temperature (150±440 °C) and pressure (<1.7 and 5.4 kbar) conditions. I however disagree with their conclusion that the presently available ¯uid inclusion data indicate a new category of ore-forming ¯uids. As is outlined later, ¯uid inclusion evidence from Birimian-hosted lode-gold deposits may very well imply that the mineralising ¯uid is compatible with H 2 O-CO 2 -salt- rich ¯uids as observed in other lode-gold deposits world-wide. I further feel that the estimation of the P±T conditions during the mineralisation of the dierent deposits of the Ashanti Belt is highly questionable. The Birimian Supergroup hosts structurally-controlled lode- gold deposits which occur in the transition zones between the volcanic belts and the basin sediments (Leube et al. 1990; MileÂsi et al. 1992). Ore-forming ¯uids of mineralised quartz veins from the Birimian volcanic belts of Ghana and Burkina Faso were investi- gated by several workers and are characterised by various signi®- cant similarities (see Klemd et al. 1996 and Klemd and Hirdes 1997 for an extensive reference list). As is shown by these studies, some of the gold-bearing quartz veins and quartz pebbles from gold- bearing conglomerates contain more than 80% CO 2 -rich ¯uid in- clusions without any visible H 2 O, some of which display a very high density (³1.0 g/cm 3 ). H 2 O-CO 2 -salt- and H 2 O-salt-rich inclu- sions only occur in subordinate amounts. On the other hand, many of these structurally hosted Birimian lode-gold deposits such as Larafella in Burkina Faso (Klemd and Ott 1997), Obuom, Mpe- satia, Kubekrom or the Ayanfuri gold deposit in the Ashanti belt (Manu 1993; Schmidt Mumm et al. 1997) have quartz veins which are dominated by H 2 O-CO 2 -salt- and H 2 O-salt-rich ¯uid inclu- sions, while CO 2 -rich inclusions occur only in subordinate amounts. This is even the case in the upper levels of the Ashanti Mine where CO 2 -rich inclusions comprise just 40% of all observed inclusions (Bowell et al. 1990). This further suggests that a single ore-forming ¯uid process operated in each of these deposits, instead of a number of processes. Therefore, if gold mineralisation is due to one common process such as metamorphic devolatilisation or mantle degassing, the CO 2 -dominated ¯uid inclusions are unlikely to be unmodi®ed relics of the mineralising ¯uids as suggested by Schmidt Mumm et al. (1997) and Klemd et al. (1996), respectively. They would appear rather to be the product of selective trapping and/or post-trapping modi®cations. This is supported by the fact that the high X CO2 of the inclusion ¯uid is not in thermodynamic equilibrium with the mineral assemblages of the altered wall rock. With decreasing distance from the quartz veins the intensity of chloritisation, seri- citisation and epidotisation increases (Schwartz et al. 1992; Manu 1993; Klemd et al. 1997), thus requiring the in®ltration of H 2 O-rich ¯uids. In contrast, stable isotope studies by OberthuÈr et al. (1996) indicate equilibration between the CO 2 and the hydrothermal al- teration assemblage. Furthermore, according to f O2 constraints from ¯uid-mineral equilibria, the original ¯uid had to be an H 2 O- CO 2 mixture (Schwartz et al. 1992). Additionally, gold is most likely transported in gold-hydrosulphide or gold-chloride com- plexes which favour H 2 O  CO 2 -rich ¯uids instead of CO 2 -rich ¯uids as a transport medium (see Seward 1989). The SiO 2 -solubility in CO 2 - rich ¯uids is also very limited (Walther and Orville 1983), such ¯uids are consequently unlikely to have been responsible for the precipitation of all the vein-quartz. In contrast I regard another explanation for the origin of the CO 2 -rich inclusions to be probable: although only few detailed geothermobarometric studies have been undertaken, recent results show that the rocks of the Ashanti Belt and the Kibi Winneba Belt have undergone epidote-amphibolite- to amphibolite-facies condi- tions, with peak-metamorphic conditions ranging between 500 and 650 °C at 5 to 6 kbar (Kleinschrot et al. 1994; HuÈnken 1995; HuÈnken et al. 1995; John 1997). In the Ashanti Belt ore deposition is interpreted to be largely synmetamorphic (OberthuÈr et al. 1996), which could be in accordance with the fact that the calculated isochores for the highest density CO 2 -rich ± and H 2 O-CO 2 -salt inclusions of almost all gold deposits intersect the P±T-range for peak metamorphic conditions, possibly indicating a syn-peak (?)- metamorphic trapping of these ¯uid inclusions (HuÈnken 1995; Klemd and Hirdes 1997). However, even if the gold mineralisation occurred during retrograde greenschist-facies conditions at tem- peratures >350 and <500 °C the vein-quartz will still show re- crystallisation features such as grain-boundary migration (e.g. Johnson and Hollister 1995). This is in accordance with the ob- servation, that all investigated quartz veins in the dierent deposits have undergone ductile deformation (e.g. Manu 1993; Schwartz et al. 1992, personal communication 1997) and recrystallisation during the retrograde greenschist-facies overprint (e.g. Manu 1993; Klemd et al. 1997; Klemd and Hirdes 1997). Consequently, most of the CO 2 -rich inclusions were the result of post-trapping deforma- tion-recrystallisation processes during retrograde conditions (Klemd et al. 1997). Grain-boundary migration recrystallisation Mineralium Deposita (1998) 33: 317±319 Ó Springer-Verlag 1998 Editorial handling: DR R. Klemd FB-Geowissenschaften, UniversitaÈt Bremen, P.O. Box 330440, D-28334 Bremen, Germany