Older descriptions of the family differ in some respects owing to the inclusion of elements now generally excluded, e.g. somewhat zygomorphic flowers occur in Henriquesiaceae, often considered as a tribe (Henriquesieae) of the Rubiaceae. In East Africa there are 100 genera and about 600 species.The classification of this family into tribes is not really very difficult, but the basic arrangement of these tribes into subfamilies is a matter of contention. The classical division is into two groups, one with the ovary-locules containing one ovule and the other with them containing 2 or more. This is still followed by many authors, e.g. Hallé in Fl. Gabon 12 (1966), and is a partially convenient method, being subject to no more exceptions than most classifications of this type. There is no doubt, however, that this is artificial and the classification proposed by Bremekamp and myself is more natural, though perhaps less practical in some respects, since it is partly based on the presence or absence of rhaphides. Since neither ovules nor rhaphides are easy to see, use has also been made of other characters in the key, as was first done by Hutchinson in the first edition of F.W.T.A. and later followed by Keay & Hepper in the second. A conspectus of the classification followed in this present account is given below; it differs little from that proposed by me in 1958, except for some modifications adopted from Bremekamp’s latest paper.Various pollination mechanisms exist in the family from simple protandry, where the pollen is dispersed before the stigmatic surfaces are revealed, to more sophisticated arrangements such as heterostyly, where there are 2 or even 3 types of flowers. In complete cases the style is well included and the anthers well exserted in one sort of flower and exactly the opposite in the other sort, but in some tribes the differences are far less marked. In many homostylous species (i.e. those not showing heterostyly) the stigma and the anthers are well isolated. The ixoroid pollination mechanism, mentioned quite frequently in the text, is where the usually cylindrical or clavate style-head (which could easily be taken for the stigma) acts as a pollen receptacle; it is in close contact with the anthers and pollen is deposited on it before the flower opens. After the flower opens the style elongates and the pollen is ready for insect transfer; later the ‘stigma-lobes’ divide to reveal the true stigmatic surfaces and are ready for pollen from other flowers. Unisexuality is rare in the Flora area, but occurs in Anthospermum, possibly derived from extreme cases of heterostyly. Most species of Rubiaceae are insect pollinated, many having white flowers strongly scented at night, being obviously pollinated by moths. Anthospermum, with its long styles and dangling anthers is clearly wind pollinated, and a few of the larger flowered Vanguerieae may even be bird pollinated. Doubtless, despite the special mechanisms, quite a good deal of self-pollination exists. There is a wealth of material here for studies to be made in the field by biologists on the spot. ON THE KEYS The family Rubiaceae contains so many genera and species, many of which resemble each other even when not closely related, that it is impossible to make a usable key which does not involve looking at small and difficult characters. If, after a preliminary run through the key, it is found that macroscopic characters will not suffice then it is best to examine the plant in detail and make a list of its essential characters. The aestivation of the corolla-lobes in bud is a very important character indeed and the main types are shown in fig. 1/1–5. This is fairly easy to see in fresh material under an ordinary hand-lens but when only herbarium material is available a bud should be boiled up. The arrangement of the ovules in the ovary is also a crucial character, whether there are one or several to many in each of how many locules and if solitary whether they are attached to the base, apex or middle of the locule. This is admittedly a difficult character to see and dissection under a binocular dissecting microscope is often necessary. A transverse cut across the ovary with a razor blade will often clearly show the number of locules, and in the case of solitary ovules whether they are attached at the base or apex can be ascertained by seeing which half of the cut ovule drops out easily. A good deal can be learnt by cutting gradually into the side of the ovary until the ovule or placentas are revealed. Care must be taken not to mistake a placenta covered with minute ovules for a single ovule; this mistake has been made even by professional botanists and several species have been redescribed in the wrong genus and tribe as a result. The presence or absence of heterostyly can be a very useful character and is worth noting in the field where a population is available. It is of course difficult or impossible to ascertain from a single herbarium specimen. If flowers showing long-exserted anthers and included styles are present on the single specimen then the species is almost certainly heterostylous, but the reverse is of course by no means true. Other microscopic characters are very useful and not difficult to see if adequate equipment is available. The presence of rhaphides restricts the choice of genera considerably. These are easily seen under a microscope and they break up into very numerous fine needle-shaped crystals if teased out under water; they are particularly easily seen if a polarising microscope is available, but in many cases, e.g. those between the pyrenes of Psychotria fruits, they are clearly visible to the naked eye and they break down into fluffy crystals even when seen dry under a × 10 hand-lens. In other cases rhaphides can be seen in leaf and other tissues with the naked eye (fig. 1). Pollen grains vary very considerably in shape in the family and are valuable characters. Even the gross morphology easily visible at low powers of a compound microscope can restrict one’s search for the identity of a difficult specimen. The following main classes may be distinguished so far as the East African tribes are concerned.1) Colpate grains. 3–4(–5)-colpate. The usual kind of grain found in the family.2) Pluricolpate grains. 5–25-colpate. These are divisible into two types:A) disc-shaped grains, usually large and with numerous short colpi on the rim—restricted to Spermacoceae;B) ellipsoidal grains with fewer longer colpi extending over the body—Spermacoceae and Rubieae.3) Porate grains. 3–4-porate. Of common occurrence in the family particularly in Mussaendeae, Hamelieae, Gardenieae, Guettardeae, Vanguerieae, Morindeae, Psychotrieae and Craterispermeae.4) Tetrahedral tetrads of 3-porate grains. Restricted to the Gardenieae. Examples of these types are shown in fig. 1/10–16.Various conventions exist for describing the various parts of the calyx. For this account the part adnate to the ovary is referred to as the calyx-tube and the free joined part as the limb-tube; the limb-tube and the lobes together are referred to as the limb.