A modern reference to temperate woody plants, including updated content from this site and much new material, can be found at Trees and Shrubs Online.

Davidia involucrata Baill.

Dove or Handkerchief Tree

Modern name

Davidia involucrata Baill.


D. laeta Dode, in part

A deciduous tree reaching a height of 40 to 65 ft in the wild, in habit resembling a lime tree (Tilia); young branches glabrous, covered more or less with a glaucous bloom, afterwards turning very dark. Leaves vivid green, broadly ovate or roundish, 3 to 6 in. long, and about three-fourths as wide, heart-shaped at the base, the apex drawn out into a long fine point; margins set with coarse triangular teeth; upper surface furnished with silky hairs, the lower one felted with a thick grey down; stalks slender, 112 to 3 in. long. About eight pairs of nearly parallel veins proceed from the midrib at an angle of 45°. Flowers produced in May with the strongly scented young leaves from the buds of the previous year’s shoots; they are crowded in a rounded head about 34 in. diameter, borne at the end of a drooping stalk about 3 in. long; male flowers composed of numerous long stamens with white filaments and reddish anthers, forming a brush-like mass; female flower reduced to an egg-shaped ovary, with a short six-rayed style and a ring of abortive stamens at the top. It is not, however, in the flowers themselves where the remarkable beauty of the davidia lies, but in two (rarely three) enormous bracts by which each flower-head is subtended. These bracts are white or creamy white, hooded, oblong, long-pointed, and of unequal size, the lower one being the larger, and sometimes nearly 8 in. long and half as wide; the upper bract is about half the size, and stands above the flower-head like a canopy. Fruit a solitary drupe about 112 in. long, green with a purplish bloom but becoming russet-coloured and speckled with red when completely ripe; it contains a single, hard, ridged nut with three to five seeds. For the shape of the fruit, see below.

Native of China in W. Szechwan and parts of W. Hupeh; first discovered near Mupin in 1869 by the Abbé David, after whom the genus is named. The first introduction of this species to Europe belonged to the var. vilmoriniana (see below). Seed of typical D. involucrata was first sent by Wilson in November 1903 and again in 1904, during his second expedition for Messrs Veitch. It is a curious fact that the young trees raised from this seed at Veitch’s Coombe Wood nursery first showed very little of the hairy character of the parent but became conspicuously hairy later (when about seven years old).

The type seems to be rarer in garden than the glabrous-leaved trees (var. vilmoriniana). This scarcity is no doubt due in part to the relatively small number of plants that were raised and distributed (‘several hundreds’ only, according to Wilson) compared to the thirteen thousand raised from the seed of var. vilmoriniana. A further reason is that it seems in many gardens to have proved (in the form introduced) less easy to establish than the variety, less vigorous, and more subject to damage by late spring-frost. Wilson remarks (Pl. Wilsonianae, Vol. 2, p. 257) that in typical D. involucrata the shoots on young plants are dark red, but dull grey or slightly purplish in var. vilmoriniana.

var. vilmoriniana (Dode) Wanger. D. vilmoriniana Dode; D. laeta Dode, in part. – Undersides of leaves yellowish green or somewhat glaucous, slightly downy on the veins at first but otherwise quite glabrous. According to Wilson this variety is common in N. W. Hupeh and also occurs, though more sparingly, in E. Szechwan; it is rare in W. Szechwan, where the type predominates, though the two are sometimes found growing commingled.

The var. vilmoriniana is commoner in cultivation than the type and introduced some years earlier. In 1897 a parcel of thirty-seven seeds was sent to Maurice de Vilmorin by the French missionary Farges. Of this sending only one seed germinated, and that not until June 1899. The plant grew and flourished, flowering for the first time at Les Barres in May 1906. This plant and a few cuttings from it were the only representatives of Davidia in Europe until Wilson’s first journey in China for Messrs Veitch, 1899-1901, during which a large supply of seeds was sent home from which some thirteen thousand plants were raised. At about the same time Vilmorin received a further consignment of seeds from Père Farges which this time germinated successfully. Thus the abundant representation of Davidia in European gardens was assured. Wilson’s introduction first flowered in Veitch’s Coombe Wood nursery in 1911.

As seen in cultivation this variety is rather variable in the colouring and texture of the leaves. Some trees have them dark green, rather thick and some­what glaucous beneath, while in others they are a lighter green, thinner, and yellowish green beneath. The second form agrees with Dode’s D. laeta but Wilson observed that in nature both kinds of leaf are to be found on the same tree. This is true also of many cultivated trees which, if carefully examined, will be found to have the darker, glaucescent leaf on the outer part of the crown and the lighter green laeta-type leaf on the inside. There are certainly no grounds, in our present state of knowledge, for distinguishing more than one variety of D. involucrata. It must be added that Dode’s D. laeta is a thoroughly confused entity, based mainly on young nursery plants some of which were, in fact, typical D. involucrata raised from seed of Wilson’s 1903 sending; in 1908, when Dode described D. laeta, these had not yet produced the adult, hairy leaf.

There remains the question whether the glabrous-leaved trees might not better be regarded as an independent species (D. vilmoriniana). The status of variety given them in the present revision is based on the assumption that the glabrous underside of the leaf is the only consistent and substantial difference between them and typical D. involucrata; this was Wilson’s own conclusion from his study of the species in its natural habitat. However, if it could be shown that the two forms of Davidia also differ constantly in the shape of their fruits this judgement would have to be reconsidered. From a study of a number of samples from cultivated trees K. A. Beckett (Journ. R.H.S., Vol. 87, 1962, p. 24) concluded that the fruit in D. involucrata is oblate, whereas in the glabrous-leaved trees it is broad- to long-ovoid. He tells us, however, that two samples received subsequently from Westonbirt suggest that the difference is not a constant one; these on their leaf-characters were referable to typical D. involucrata but bore ovoid, not oblate, fruits. Mr Beckett’s observations are of great interest, but no definitive conclusion can be reached until the variations shown by wild trees have been more thoroughly studied. For further observations on Davidia fruits see H. Cocker in Gard. Chron., Dec. 6, 1952, p. 226, and H. J. Ivens, ibid., Jan. 17, 1953, p. 22.

In the half-century that has passed since the first edition of this work was published the davidia has become one of the best known of all hardy exotic trees. Wilson considered it to be ‘the most interesting and beautiful of all trees of the north-temperate flora’ and likened the white bracts to ‘huge butterflies hovering among the trees’. It thrives best in a moist soil and a sheltered situation. Although quite winter-hardy those forms which start into growth early are subject to damage by late spring-frosts, the typical, hairy-leaved form being, as already remarked, particularly vulnerable in this respect. Given favourable conditions it grows rapidly and reaches the flowering stage in ten years or even less. It is easily propagated by cuttings, or by the seed, which in most years is abundantly borne.

D. involucrata is represented in most of the larger tree collections, usually by the var. vilmoriniana. The largest are 45 to 57 ft high, 312 to 5 ft in girth.

Davidia involucrata

Davidia involucrata


It is, of course, possible that the garden trees with the predominantly darker green, glaucescent leaves are not from Wilson’s seed but belong all to one clone descended from the original tree at Les Barres which Dode took as the type of his D. vilmoriniana. This tree is known to have been multiplied by cuttings and layers when still only a few years old (Frut. Vilm., p. 146, 1904) and some of its vegetative progeny may well have been imported by British nurseries other than Veitch’s and used as stock plants. A tree of the Les Barres clone was obtained for Kew around 1904 and grown in the Temperate House; this too was propagated vegetatively.

From the Supplement (Vol. V)

As remarked on page 26, the typical variety is less common in gardens than the var. vilmoriniana, but the following have been recorded: Westonbirt, Glos., 56 × 614 ft, recovering from loss of branches during the 1976 drought (1983); Bulstrode Park, Bucks., 54 × 312 ft (1983); Woodhouse, Devon, 59 × 412 ft, grafted on var. vilmoriniana (1982); Tregrehan, Cornwall, 62 × 712 ft at 2 ft (1979).

The following are var. vilmoriniana: Wakehurst Place, Sussex, 72 × 614 ft (1984); South Lodge, Horsham, Sussex, 62 × 414 + 212 ft (1985); Tilgate Park, Sussex, 82 × 414 + 414 + 414 ft (1984); Borde Hill, Sussex, Warren Wood, 72 × 412 ft (1983); Emmetts, Ide Hill, Kent, 42 × 512 ft (1984); Frensham Hall, Shottermill, Surrey, 65 × 514 ft (1981); Fosbury Manor, Wilts., 70 × 434 ft (1984); Stourhead, Wilts., 56 × 534 ft (1984); Batsford Park, Glos., 64 × 412 ft (1983); Hergest Croft, Heref., 52 × 614 ft and 75 × 5 + 434 ft (1985); Wayford Woods, Som., 70 × 414 ft (1981); Burncoose, Cornwall, 56 × 5 ft (1984); Werrington Park, Cornwall, 56 × 634 ft (1977); Bulkeley Mill, Gwyn., 66 × 5 ft (1984); Bodnant, Gwyn., 50 × 512 ft (1981); Hafodunos, Gwyn., 66 × 5 + 4 ft (1984); Dalmeny House, nr Edinburgh, 46 × 6 ft at 3 ft (1984); Kilmacurragh, Co. Wicklow, Eire, 62 × 514 ft at 3 ft (1980).

The propagation of the davidias by cuttings is discussed by J. G. D. Lamb and F. J. Nutty in The Garden (Journ. R.H.S.), Vol. III, p. 189 (1986).



Other species in the genus

[No species article available]