1 68 JOURNAL OF THE ROYAL HORTICULTURAL SOCIETY. 



spine-pointed bracts, which become very small on the branches. Inflorescence 

 a dense, very leafy, pyramidal or oblong panicle, extending from near the base of 

 the stem to its apex, and about 2 inches wide ; lower branches ascending, upper 

 patent ; ultimate bracts minute, subtending the pedicels, which equal or exceed 

 the flowers, and are thickest under the calyx. Buds ovate-oblong, white tipped 

 pink. Flowers very numerous, \ inch across, starlike. Calyx cup-shaped, green 

 dotted with purple, segments ovate, apiculate, very fleshy, divided to the base. 

 Corolla thrice the calyx, funnel-shaped near the base, patent above, petals J inch 

 long, lanceolate, acute, with ascending tips, white inside, outside keeled and 

 mottled with red and green near the tip. Stamens spreading, slightly shorter 

 than the petals, filaments white, anthers deep purple, the epipetalous ones 

 inserted near the base of the petals. Scales pale yellow, spreading, retuse, oblong, 

 twice as long as broad, equalling the stalk of the carpels. Carpels nearly equalling 

 the petals, white, tapering into slender divergent styles, abruptly contracted 

 below into a slender stalk, turning rosy in fruit. 



Flowers September-October (August in China). Not hardy. 



Habitat. — China : Kansu ; Pe-che-li. 



The sub-globose spiny buds of flat, cuneate-spathulate leaves 

 which, in cultivation, appear to be produced at irregular periods, 

 probably represent a winter state, and are evidently a resting stage. 

 I have had young plants from Mr. Farrer and from Edinburgh ; the 

 fine flowering specimen figured was sent by Mr. E. A. Bowles. 



The quite inadequate description of L^veille [loc. cit.) led me to 

 consider Mr. Farrer's No. 336 from Kansu as distinct from Chaneti, 

 especially when the author of the latter subsequently (Bull. Geogr. 

 Bot., 27, 74, 1917) identified his plant with S. spinosum Thunberg 

 {Cotyledon spinosa Linn.), a quite different plant with a very dense 

 raceme, well known in gardens and in herbaria ; but access to the type of 

 Chaneti in the Le veille herbarium now at Edinburgh, shows the identity 

 of his plant and mine, so his name stands for this remarkable species. 



The following notes supplied by Mr. Reginald Farrer on the 

 plant in its native surroundings are of interest : 



" Though on roofs in other pfeces, as at Lanchow, I saw isolated 

 plants suggesting No. 336 in very poor form, I am certain of my plant 

 only at and about Siku, abounding on the flat roofs in solid sheets 

 of foliage, very beautiful in their glaucous metallic sheen. I remember 

 particularly how it filled every gully between the tiled ribs of the big 

 military yamen, and how, on the roof of my pony-stall, it made in 

 August a dense jungle of its upstanding stocky spires of white stars — 

 no doubt in character far surpassing all that even Bowles' plant was 

 able to produce, and in appearance most suggesting 8-inch spikes of 

 Saxifra^a longifolia, on a small, starved scale of blossom. Really a 

 striking plant, but not, I fear, likely to prove hardy or resistant with 

 us. For it thus loves only the hottest and poorest soils and rocks, 

 in the hot, dry region of the Blackwater River's bed ; and though it 

 ascends from Siku (6,400) another 2,000 feet on the mountains, where 

 it is sporadic on very hot rocks, it nowhere ascends within reach of 

 the alpine zone. Its kindred vegetation is Lilium ienuifolium, Con- 

 volvulus tragacanthoides, Leptodermis virgata, Hedysarum muUijugutn, 

 Incarvillea variabilis, and the Asters hispidus and oreophiliis. I 

 should add that in nature it is certainly not monocarpic, but each 

 plant forms a close and ample agglomeration of rosettes, from which 



