Bufumbira is the name given to the part of the Virunga volcanic field (see Congo No. 6) lying within Uganda. It covers about 380 km2, but this is only some 10% of the total Virunga field. The southern margin of Bufumbira is marked by its three most lofty volcanoes between the summits of which lies the Uganda-Rwanda border. These volcanoes are Muhavura (4127 m), Mgahinga (3420 m) and Sabinyo (3590 m), but in addition there are another 34 smaller ones together with two vents from which lavas issued, as well as numerous explosion craters. Many of the volcanoes lie on north-northwest-trending lines, that are parallel to the general strike of the underlying Karagwe-Ankolean basement rocks, as illustrated by Holmes and Harwood (1937, Fig. 8). Combe and Simmons (1933) include numerous photographs illustrating the volcanic landscapes. Both Muhavura and Mgahinga are relatively symmetrical cones with summit craters, there being a lake in the former. Sabinyo is much more heavily dissected, however, with deep gorges running into the mountain. All three are probably built essentially of lava flows. Lavas also issued from most of the smaller volcanoes but they contain a higher proportion of pyroclastic rocks, and a few appear to comprise solely such materials. Predominantly, however, the Bufumbira field is one of lava flows as is apparent from the geological map of Combe and Simmons (1933) on which 49 flows are distinguished, principally according to the vent from which they originated. In many cases, particularly on the larger volcanoes such as Muhavura, there are certainly many more individual flows than are indicated on the map. The flows mapped by Combe were named and classified initially on field appearance and preliminary microscopic study. The petrography was then taken further (Combe and Simmons, 1933), after more collecting and microscope work, with the erection of the following main groups: leucite basanite, trachytic leucite basanite, leucite basalt, limburgite and related rocks, tephrite and trachytic tephrite, olivine basalt and related rocks, andesite and trachytic basalt, and some relatively minor rock types. Each of these groups is described by occurrence, with considerable detail being presented on the locality, nature and petrography. Many of the rocks described by Combe and Simmons (1933) were then sent to Holmes and Harwood for more detailed petrological and chemical investigation, which resulted in a now classic memoir (Holmes and Harwood, 1937). Holmes divided the rocks into two types: plutonic rocks, which included ejected blocks and inclusions, and volcanic rocks, which includes all the lavas. Because the material sent to Holmes and Harwood was not confined to specimens from Bufumbira but included samples from the whole Virunga field, the classification adopted included rock types and series not represented in the Bufumbira area. In the following account reference will only be made to rock types represented in the Bufumbira field; for the petrology of the Virunga field as a whole, as classified by Holmes, reference should be made to the memoir. Unfortunately, because of the differences between the classification systems adopted by Combe and Simmons (1933) and Holmes and Harwood (1937) Holmes' types do not correspond to the 49 flow types distinguished on the map made by Combe. All the rocks of the Bufumbira field have K2O>Na2O, although this does not apply to the whole Virunga field. Holmes divided them into olivine leucitite, basanite, potash trachybasalt and potash trachyandesite (latite) series. Mafic members of the olivine leucitite series are called ugandite and contain abundant olivine and clinopyroxene phenocrysts in a groundmass of dominant pyroxene, a little olivine, very abundant leucite, subordinate alkali feldspar and/or plagioclase, opaque phases and perovskite. These rocks pass into even more olivine-rich varieties and, with decrease in olivine and pyroxene, into leucitite, some varieties of which contain significant alkali feldspar and nepheline. With the introduction of plagioclase olivine leucitite grades into leucite basanite, referred to by Holmes as kivite, and these are the dominant rock types of the Muhavura and Mgahinga volcanoes. In these rocks olivine and pyroxene are generally more sparse as phenocrysts, but are plentiful in the groundmass together with labradorite, sometimes alkali feldspar, leucite, opaque phases and accessory apatite; varieties with nepheline and others with biotite occur, while olivine-free types are essentially tephrites. The trachybasalt series, like the others, is potassic and was found to be similar to the type absarokite-shoshonite-banakite series. Leucite is absent from these rocks, or present in minor amounts. The Bufumbira absarokites contain olivine and clinopyroxene phenocrysts in a groundmass of pyroxene, abundant opaques, labradorite, alkali feldspar and a little leucite. These rocks pass into shoshonitic rocks, many referred to by Holmes trippingly as shoshonitic absarokites, in which plagioclase may be a phenocryst phase, as well as pyroxene and olivine, in a groundmass which features alkali feldspar, leucite and opaque minerals. In the banakites olivine and pyroxene phenocrysts occur only rarely, although sodic plagioclase or biotite may form phenocrysts; the groundmass is generally trachytic and of oligoclase-andesine, opaques, pyroxene and alkali feldspar. These rocks in turn grade into trachytes. The potash trachyandesite series was also called the shoshonite-latite series by Holmes, and is confined to Sabinyo, only about a quarter of the products of which lie within Uganda. The Sabinyo lavas are unique in the field in containing orthopyroxene; they are also oversaturated in silica. The shoshonites contain phenocrysts of augite, olivine and andesine, sometimes with alkali feldspar rims, in a groundmass of alkali feldspar, a little oligoclase, greenish pyroxene, opaques, rare biotite and isotropic material. In another variety biotite is abundant as phenocrysts and in the groundmass, and a little orthopyroxene is present. The latites contain the same phenocrysts as the biotite shoshonites but hypersthene is much more abundant and olivine rare; they are quartz normative. It is noteworthy that neither in Virunga, nor in the whole Virunga field, are true basalts found. Ejected blocks and inclusions in the flows of coarse-grained rocks were studied in some detail by Holmes, who tabulated them according to rock type, associated lava and locality (Holmes and Harwood, 1937). They range from dunite through augite peridotite to pyroxenite, biotite pyroxenite and glimmerite. These rocks constitute Holmes' 'transfusion series' which, he proposed, was generated by emanations rising from the depths, the melting of which gave rise to the lavas. A more modern classification of the lavas, following the system based on A (alkali feldspar)- P (plagioclase)- F (feldspathoid) described by Streckeisen (1967), is given by Ferguson and Cundari (1975). In this system Holmes' shoshonitic absarokite, banakite, kivite, leucite basanite and absarokite become latite, tephritic leucite phonolite and phonolitic leucite tephrite; the ugandites and olivine-rich ugandites become leucite tephrite and olivine basanite, and Holmes' various types of leucitite and olivine leucitite become leucitite and phonolitic leucitite. Ferguson and Cundari (1975), using the chemical data of Holmes and Harwood (1937) together with further analyses of some of the original specimens, demonstrated that two chemical series can be distinguished at Bufumbira deriving from basanite; one series differentiates through phonolitic tephrite to trachyte and the other through leucitite to phonolite. Feldspar crystallization trends in these two series are discussed by Ferguson and Cundari (1982) and Ca-enrichment in olivine by Ferguson (1978). Strontium isotope data on many samples originally described by Holmes and Harwood (1937) are given by Bell and Powell (1969); Bell and Doyle (1971) present K and Rb data on many of the same specimens. REE, Ta, Hf, Sc, Co and Th data are given by Mitchell and Bell (1976), and Vollmer and Norry (1983) supply Pb, Nd and Sr isotopic data - again on material described by Combe and Simmons (1933) and Holmes and Harwood (1937). Oxygen isotope and partial chemical analyses will be found in Taylor et al. (1984). The petrogenesis of the Virunga lavas and plutonic inclusions, based mainly on the data of Holmes and Harwood (1937), is considered by Cundari and Le Maitre (1970) using principal vector variation diagrams. The phase relationships of an olivine-rich ugandite described and analysed by Holmes and Harwood (1937) were studied at 10 to 40 kb by Edgar et al. (1980).