Compilation
Agathisanthemum quadricostatum

A. bojeri, with its usually rather open inflorescence of heterostylous flowers, short calyx-lobes, pubescent disc and mostly coastal distribution, is clearly distinct from A. globosum with isostylous flowers in very compact globose heads, longer calyx-lobes, finely papillate disc and mostly upland distribution. The “species” described above is in many ways intermediate between these two and, although undoubtedly not just direct hybrids since the two do not now occur together, it is very likely that it has arisen by hybridisation in the past. In Kenya and Uganda where the two could not have been in contact during any time relevant to the problem, no intermediate populations of any kind exist but in Tanzania, particularly in the west where it is very likely that A. bojeri could have occurred, it may have been eliminated through hybridisation with A. globosum to give rise to the ill-defined “species” here recognised for convenience. Heterostyly is not at all constant e.g. in Milne-Redhead & Taylor 8491 (Tanzania, 12 km. W. of Songea, by Kimarampaka stream) the inflorescence is as open as in many specimens of A. bojeri, the calyx-lobes are 4.6 mm. long, the disc has some hairs and the flowers have both anthers and style well exserted; it clearly has characters of both A. bojeri and A. globosum but the general appearance is that of A. quadricostatum. Another intermediate specimen is Lunan H 86/46 (Tanzania, E. of Tabora, Igalula), a poor specimen made the type of a supposed new species, A. assimile, by Bremekamp. This has an open inflorescence, calyx-lobes 4 mm. long, the disc distinctly hairy and the flowers with both anthers and style exserted. He distinguished it from A. bojeri by having the leaf-nerves all springing from the lower half of the leaf-blade (not true so far as can be seen from the specimen), the stipule-sheath, calyx-lobes and capsule all longer. Further it is distinguished from A. angolense Bremek. by not having a cork layer which peels, although in the type specimen the bark is clearly peeling off; in the diagnosis it is among other things distinguished from A. quadricostatum by its hairy disc and in not having a 4-angled stem, but the first character is variable in A. quadricostatum and the type of A. assimile has some parts of the stem very distinctly 4-ribbed. The genus is much commended to local workers in East Africa for field and experimental studies; only by such means will the confusion be resolved.