The occurrence of andalusite and sillimanite in unaltered igneous rocks is, according to the orthodox view expressed in most standard textbooks (see, e.g., Tyrrell, 1934, pp. 50, 164; Shand, 1927, pp. 62, 146; Grout, 1932, p. 230), always to be ascribed to contamination of magmas by highly aluminous sedimentary or metamorphic rocks. Having been given cause to doubt the correctness of this view by the recognition at Pyramid Hill, Victoria, of andalusite-bearing granites and aplites in which evidence of assimilation is lacking, I was then very interested to discover that the opinion has often been expressed, and evidence adduced in support of it, that both andalusite and sillimanite may be normal pyrogenetic constituents of igneous rocks.1 That is, they may under certain conditions crystallize from uncontaminated magmas. Some authors, while admitting that andalusite and sillimanite may crystallize from magmas, regard such pyrogenetic occurrences of these minerals as caused by the development of local excess of alumina, due to the assimilation of shales (e.g. Wells, 1931; Shand, 1927, p. 62). Others do not make their position clear, merely classing andalusite and sillimanite as assimilation minerals (sic), but Tyrrell goes so far as to state that they are “never of pyrogenetic oiigin” (1934, p. 50). Because of the reliance that is placed upon accessory minerals in igneous rocks as indicators of consanguinity of magmas and of the role of assimilation and other processes in pedogenesis, it is important that the status of each mineral should be thoroughly understood. In most classifications of accessory minerals andalusite and sillimanite are either classed as “contamination accessories” (Wells, 1931) or grouped with minerals that are commonly due to contamination (Wright, 1932), and Wright regards them as “of little value for correlation purposes”. Chatterjee, however, was able to use andalusite as an indicator, on the one hand, of relationship between the Falmouth and Bodmin Moor granites, both of these containing a purple variety in fair amount, and, on the other, of the distinction of these granites from those of Dartmoor and St. Austell, in which andalusite is colourless and rare. The rare, sporadically developed andalusite in the Dartmoor granite is considered by Brammall and Harwood (1923) to be a contamination mineral, but Teall suggested (1887) that the andalusite in the Cheesewring granite is probably an “original constituent” (i.e. not mechanically incorporated with the granite, as strew from xenoliths or wall rocks), and the relative abundance and uniformity of distribution of andalusite in the normal type of the Bodmin Moor granite, as exhibited at the Cheesewring (see Ghosh, 1927), lend support to this suggestion.