The petrological and mineralogical characteristics of the unusual CM2 chondrite, Bells, have been investigated in detail by scanning electron microscopy (SEM), electron microprobe analysis (EPMA), and transmission electron microscopy (TEM). Bells is a highly brecciated chondrite which contains few intact chondrules, a very low abundance of refractory inclusions, and is notable in having an unusually high abundance of magnetite, which is disseminated throughout the fine-grained matrix. Fragmental olivines and pyroxenes are common and, based on compositional data, appear to have been derived from chondrules as a result of extensive brecciation. The fine-grained mineralogy of matrix in Bells differs considerably from other CM chondrites and has closer affinities to matrix in CI chondrites. The dominant phases are fine-grained saponite interlayered with serpentine, and phases such as tochilinite and cronstedtite, which are typical of CM chondrite matrices, are entirely absent. Pentlandite, pyrrhotite, magnetite, anhydrite, calcite, and rare Ti-oxides also occur as accessory phases.Based on its oxygen and noble gas isotopic compositions (Zadnik, 1985; Rowe et al., 1994), Bells can be considered to be a CM2 chondrite, although its bulk composition shows some departures from the typical range exhibited by this group. However, these variations in bulk chemistry are entirely consistent with the observed mineralogy of Bells. The unusual fine-grained mineralogy of Bells matrix can be reasonably attributed to the combined effects of aqueous alteration and advanced brecciation in a parent body environment. Extensive brecciation has assisted aqueous alteration by reducing chondrules and mineral grains into progressively smaller grains with high surface areas, which are more susceptible to dissolution reactions involving aqueous fluids. This has resulted in the preferential dissolution of Fe-rich chondrule olivines, which are now completely absent in Bells although present in other CM chondrites. The formation of saponite in Bells probably resulted from the dissolution of relatively silica-rich phases, such as pyroxene and olivine, that were derived from chondrules. The result of such dissolution reactions would be to increase the activity of silica in the fluid phase, at least on a localized scale, stabilizing saponite in preference to serpentine. An increase in aSiO2 would also have destabilized preexisting cronstedtite which may have reacted to form magnetite and Mg-Fe serpentine under conditions of constant ƒO2 .