Transition to the High Mountains of Tropical Africa . 529 
lobata , Marss. In higher altitudes at the summit of M. Silke, M. Bachit, 
M. Dedjen at 3,000 m. to about 4,500 m., the plant does not grow taller 
than to 10-30 cm., the panicle being more contracted. It is the species of 
grass which ascends to the highest altitudes. On the Kilimanjaro, from 
3,700-3,200 m., we find a robust form, 50-70 cm. high, with a many-flowered 
panicle 10 cm. long. At the altitude of 3,600 m., in the uppermost region 
of Ericinella , we find specimens of nearly the same height, with a pubescent 
rhachis and a long but condensed panicle, which, shading into violet, recalls 
Anthoxanthum nivale , K. Schum., to some extent. Very similar, somewhat 
smaller, specimens were collected in Abyssinia at Guna at 3,900 m. Finally, 
on the Kilimanjaro, being the last grass on lava-fields at 4,500 m., a variety 
15-18 cm. high makes its appearance and is called var. supina by Dr. Pilger. 
This variety forms thick, unusually dense tufts, having rigid, much con- 
volute basal leaves, flatter cauline leaves, and a narrow panicle 5~7 cm. long, 
with frequently glabrous rhachis. On the Cameroon Peak, only tall forms 
of Koeleria cristata have been collected so far, which have rather loose 
or more contracted panicles and flat or convolute leaves. In conclusion, 
it may be mentioned that a form of K. cristata , (L.) Pers., somewhat 
different from the above, having leaves partly flat, partly convolute, and a 
contracted panicle, var. gracilis, Hack., grows on Devil’s Peak, near Cape- 
town. We see, then, again a plant, which, having reached Africa from 
Europe, appears in varieties and forms somewhat different from those 
produced in Europe. 
Among the species which came from the northern temperate zone to 
the mountains of Africa, Arabis albida , Stev. (1812), being known also as 
A. caucasica, Willd. 1 (1813), a later synonym, is a very remarkable instance. 
While holding this species sufficiently distinct from A. alpina , I have no 
doubt that the forms attributed as varieties to A. albida by Boissier in 
Flor. Orient., i, p. 174, are only local modifications. A. albida, being widely 
distributed in the vertical direction from the forest-region upwards nearly 
to the uppermost limit of siphonogamic vegetation, presents quite a number 
of varieties completely corresponding to the conditions of localities : longer 
internodia, larger leaves, reduced hairiness in the woods, deep-reaching 
roots, much-branched rhizomes, shorter, mostly very hairy, often tomentose 
leaves, also pale pink petals in the debris of the mountain-slopes. Under 
similar conditions, the same forms appear even at far separated localities. 
Specimens gathered by me in Algeria in the forest of cedar near 
Teniet el Haad at 1,400 m. completely agree with plants from the Curral 
grande, Madeira. Abyssinian specimens from the Hedscha at 3,000-4,000 m. 
approach very much the var. Billardieri of Cyprus. On the other hand, 
1 Halacsy in his Flora Graeca has preferred A. caucasica, Willd., quoting Willd. Enum. Suppl., 
1809; ^ ut Supplement of the Enumeratio of Willdenow was issued not earlier than 1813, 
viz. one year after the publication of A . albida , Stev. 
