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.

Magnolia kobus DC.

Modern name

Magnolia kobus DC.


M. thurberi Parsons ex W. Robinson

A deciduous tree, ultimately 30 or 40 ft high, with a trunk 3 ft in girth, of quick growth and pyramidal form when young, but eventually round-headed. Young branches aromatically fragrant when crushed; winter leaf-buds downy. Leaves obovate, 3 to 6 in. long, often contracted at the apex to a short point, tapering at the base to a short stalk. Flowers amongst the smallest in the genus, often under 4 in. in diameter when fully expanded; petals six, pure white, obovate; sepals small, soon falling; flower-stalk downy. Fruits pinkish, seeds bright red. Flowers in April. Bot. Mag., t. 8428.

Native of Japan. Although one of the less attractive of magnolias when young, when it does not flower freely, this species is an interesting addition to cultivated Japanese trees on account of its vigorous constitution. It is much used by Japanese gardeners as a stock on which they graft M. stellata, and it blooms profusely when older. It was probably first introduced to England by Maries; there was a fine tree in the Coombe Wood nursery sent home by him in 1879. This tree is said to have first flowered around 1909, when thirty years old, and this no doubt has helped to support the belief that M. kobus is slow to flower. But other introductions from Japan have flowered when quite young and the truth may simply be that this species is variable in the time taken to reach the flowering stage. Since it is one of those species that can be propagated quite easily by cuttings, forms that take an unconscionable time to reach flowering age will no doubt disappear from commerce.

var. borealis Sarg. – This appears to be a more robust version of the species, found occasionally up to 80 ft high, with stouter branchlets, leaves 6 to 7 in. long and flowers up to 5 in. wide. Sargent described it from specimens that he had collected near Sapporo, in the north island of Japan (Hokkaido), in 1892 and cited other specimens from that area and from the northern part of the main island. This variety does not, however, entirely replace the type in northern Japan; in other words, a form of M. kobus from Hokkaido or the northern part of the main island is not necessarily var. borealis.

In 1876 the Arnold Arboretum had received seeds of M. kobus from Sapporo, and a tree raised from these, growing in his garden at Boston, was considered by Sargent to belong to his var. borealis. In 1908, when almost thirty years old, this tree was still sparse-flowering, but fifteen years later it was making a fine display and bearing more flowers every year (E. H. Wilson, Gard. Chron., Vol. 73 (1923), p. 301). In 1922 the late Charles Eley of East Bergholt Place, Suffolk, obtained seeds which almost certainly came from this tree and distributed a share of these, and later seedlings, to many gardens. Mr Maxwell Eley tells us that of the seedlings planted at East Bergholt all but two were removed because they were so slow to flower. Of those retained, one first bore flowers when twenty-five years old and to quote Mr Eley’s words ‘never flowered properly till 1964, when it was so dense with flowers that it looked as if a white sheet had been draped over it’. Since then it has flowered every year but not so well as in 1964. This tree measures 35 × 214 ft (1966). The sister tree is smaller and has not yet flowered. In other gardens too the Eley introduction has been slow to flower, but evidently there is still hope for them.

From the Supplement (Vol. V)

The name M. kobus DC. has been so consistently used for this species ever since its introduction to the West that it may come as a surprise to learn that it rests on insecure foundations. Its general acceptance is no doubt due to the noun-epithet kobus used by de Candolle, for this links it with plate 42 of Banks’ Icones Selectae (1791), a collection of engravings based on drawings by the German traveller Engelbert Kaempfer, and indeed de Candolle cited this plate in his original description. The plate in question is captioned ‘Kobus’, which is a rendering of the Japanese name for the species, ‘kobushi’. Unfortunately, de Candolle also cited as a synonym M. gracilis Salisb., which is a form of M. quinquepeta (liliiflora). The name M. kobus DC. was in fact challenged by the Japanese botanist Koidzumi, who pointed out in 1929 that the specimen in de Candolle’s herbarium, named M. kobus, is M. quinquepeta. He accordingly published a new name for the species – M. precocissima – but so far as can be ascertained this was not taken up by any other botanist.

There is worse to follow. In 1794, Thunberg published the name Magnolia tomentosa as part of an attempt to rectify the treatment of Magnolia in his Flora Japonica. Now it has never been doubted that M. tomentosa Thunb. is in part M. kobus, since Thunberg too cited the Banks/Kaempfer plate 42. But his description was drawn not from M. kobus but from a flowerless specimen of Edgeworthia papyrifera, and the Japanese name ‘mitsmata’ which he cites (together with ‘Kobus’) also belongs to that species. Unsurprisingly, M. tomentosa has been rejected as a confused name.

Unfortunately, the rules of nomenclature as they now stand (1986) make no provision for the rejection of confused or doubtful names such as M. tomentosa Thunb., nor for the ‘cleaning up’ of a name by, in the case of M. kobus DC., adding to the author citation ‘excl. syn. M. gracilis Salisb.’.

This matter would not have been mentioned here but for the fact that the Japanese botanist K. Ueda has recently proposed the rejection of the name M. kobus DC. in favour of M. precocissima Koidz. To complicate matters still further, he considers that the Banks/Kaempfer plate 42 represents not M. kobus sens. strict. but the related M. stellata (M. kobus var. stellata) and that the name M. tomentosa Thunb. should be used for this magnolia. See further in Acta Phytotaxonomica et Geobotanica, Vol. 36(4-6), pp. 149-158 (1985) and Taxon, Vol. 35(2), pp. 344-7 (1986).

In view of the likelihood that the names M. kobus and M. stellata will be proposed for conservation under Article 14.2 of the ICBN (1983) it is inadvisable for either of the names mentioned to be adopted in horticultural literature pending a final decision on the matter.

specimens: Kew, pl. 1915, 40 × 512 ft (1979) and another 48 × 612 ft (1978); Winkworth Arboretum, Godalming, Surrey, 42 × 312 ft (1979); Petworth House, Sussex, 52 × 312 ft (1983); Borde Hill, Sussex, Azalea Ring, 34 × 534 ft (1981); Trewithen, Cornwall, 44 × 534 ft (1979); East Bergholt Place, Suffolk, pl. 1923, 33 × 3 ft (1972).

For M. kobus var. stellata (Sieb. & Zucc.) B. C. Blackburn, see M. stellata, page 669. For M. kobus var. loebneri (Kache) Spongberg, see M. × loebneri, page 657.

M. ‘Wadas Memory’. – The original plant of this clone was received at the University of Washington Arboretum, Seattle, a few years before 1940 and named in 1959. The flowers are larger than in M. kobus and very freely borne over a period of some three weeks in April. It is considered to be a hybrid with M. salicifolia as the other parent, but the name M. × kewensis, based on a plant once thought to be of this parentage, is not available (see M. salicifolia in this supplement).



Other species in the genus