Recent experimental studies have suggested that colloidal silica can form in high-T (300 to >700°C) hydrothermal fluids (Wilkinson et al., 1996). Natural evidence in support of this was found by Williamson et al. (1997) who proposed a colloidal (gel) silica origin for <50 μm irregularly-shaped inclusions of quartz contained in greisen topaz from southwest England. Confocal and microprobe studies, presented here, strengthen this argument although rather than forming a gel in the hydrothermal fluid, it is suggested that the colloidal silica aggregated as a viscous coagulated colloid, with much of its volume (<10 to 30 vol.%) consisting of metal (mainly Fe) -rich particles. This is evident from the largely solid nature of metal-rich shrinkage bubbles contained at the margins of the inclusions of quartz which shows that the material forming the inclusions contained much less liquid than would be expected in a silica gel. These findings may have important implications for models of ore formation since the precipitation of a coagulated colloid could inhibit hydrothermal fluid transport and cause co-deposition of silica and entrained ore-forming elements. The mode of formation of the colloidal silica and further implications of the study are discussed.