Trees or shrubs with usually deciduous opposite leaves, the only exceptions being C. capitata, more or less evergreen in mild districts; and C. alternifolia and C. controversa both of which have alternate leaves. Flowers usually white, sometimes greenish or yellowish, always small, and produced in terminal corymbs or cymes, or clustered densely in heads; the parts of each flower are in fours. Fruit a drupe containing a two-celled stone. Many of the cornels are characterised by having the hairs of the leaf flattened to the surface and attached to it by their centres.
The species of Cornus treated in this work can conveniently be classified into four units, each of which should, in the opinion of some botanists, be given generic rank. The dismemberment of Cornus is not accepted in this work but in the following analysis the names of the segregate genera are given, together with their authors and synonymy.
1. Flowers borne in corymbose inflorescences which lack both bracts and bracteoles. This is the largest group and contains all the species described in this work except those mentioned by name under the other three groups. Considered as a genus, this group would take the name Swida Opiz (1838) (syn. Cornus L. emend. Hutch. (1948); Thelycrania (Dumort.) Fourr.). The generic name is also spelt’Swyjda’ or ‘Svida’.
2. Flowers borne in dense umbels with a yellowish involucre which falls as the flowers open. Here belong C. mas and its allies, namely C. officinalis and chinensis from E. Asia and C. sessilis from western N. America. Although Linnaeus did not designate a type-species for the genus Cornus, the name had been restricted to C. mas by Opiz in 1838 and this species and its allies would therefore constitute the genus Cornus in the narrow sense (syn. Macrocarpium Nakai).
3. Flowers in dense clusters, surrounded by large and conspicuous bracts. This group may be subdivided as follows:
a) Fruits clustered but free from each other. Here belong C. florida and nuttallii (also the Mexican C. urbaniand).
b) Fruits completely united into a fleshy compound fruit (syncarp). To this group, confined to E. and S.E. Asia, belong C. capitata, C. kousa and a few other species.
The correct name for this group, if considered to rank as a genus, would be Benthamidia Spach (1839) (syn. Benthamia Lindl. (1833), not Lindl. (1830), nor A. Rich. (1828); Cynoxylon Raf. ex Small (1903); Dendrobenthamia Hutch. (1942)). If a further split were made, and the two sub-groups each given generic rank, then C. florida and its allies would retain the name Benthamidia Spach and C. capitata and its allies would take the name Dendrobenthamia Hutch.
Species of osier-like habit, like C. alba, can be increased by cuttings of naked wood put in the open ground like willows, during the winter. Others with a stoloniferous habit can be propagated by offsets, and the rest by layers, when seed is not available. The following may be recommended as the best for general cultivation:
For Flower. – C. mas, C. rugosa, C. racemosa, C. kousa, C. florida, C. nuttallii (not for dry gardens), C. capitata (in mild localities).
For Colour of Leaf. – C. alba ‘Spaethii’, C. alba ‘Elegantissima’, C. mas ‘Aurea Elegantissima’.
For Beauty of Stem. – C. alba, C. alba ‘Sibirica’, C. stolonifera ‘Flaviramea’.
For Habit. – C. controversa, C. macrophylla, C. rugosa, C. hessei.
Rafinesque published the name Cynoxylon in 1838 but it is a little uncertain whether he intended it to represent a genus or a subgenus; for this reason it is best passed over in favour of Benthamidia.
From the Supplement (Vol. V)
The Chinese species in cultivation are surveyed by William Gardener, in The Plantsman, Vol. 1, pp. 85-105 (1979).