564 MR. G. BENTHAM ON COMPOSITAE. 
danthus in Chile and Astemma in the Andes. The Psiadia is an 
endemic species of a genus otherwise Mascarene or of eastern 
Africa, presenting a geographical connexion analogous to that of 
the St.-Helena Melhanie of De Candolle with the Mascarene 
Trochetia. 
The remaining Composite of the island are species belonging to 
other countries, and especially South-African. How far they 
have been introduced in modern times through the agency, di- 
rect or indirect, of man, or how far they may be ancient colo- 
nists brought to the island by currents, birds, or other causes 
still in operation or whose action may have ceased, it will perhaps 
not be possible to determine. The following are those which 
are said to be well established :— Gnaphalium indicum and G. syl- 
vaticum ?, Cotula coronopifolia and C. australis, Siegesbeckia orien- 
lalis, Eclipta alba, Senecio vulgaris and S. sylvaticus, Osteosper- 
mum moniliferum, Hypocheris radicata and H. glabra, and Son- 
chus oleraceus, besides an Oligocarpus, which I have been unable 
to match with the known species, but may yet very likely be found 
in South Africa with its congeners. Several planted trees, also 
from various countries, thrive so as to appear wild. 
Our knowledge of the plants of Tristan d'Acunha is derived 
from Dupetit Thouars's Flora of the island in his * Mélanges, 
supplemented by Captain Carmichael in the 12th volume of the 
Linnean Transactions. He enumerates four Composite, of which 
two, Chevreulia stolonifera and Lagenophora Commersonii, growing 
sparingly on the most barren exposed rocks, are Antarctic South- 
American plants ; the other two are weeds of the cleared grounds 
—one the cosmopolitan Sonchus oleraceus ; the other Gnaphalium 
pyramidale, Thou., is described as endemic, which, however, is not 
probable. I know not where any specimen may be preserved to 
determine the point. : 
7. Mascarene Islands. 
The Composite flora of Madagascar, Mauritius, and Bourbon is 
of bigh interest in a phytogeographieal point of view, from its 
evident connexion with the most ancient types of the African 
eontinent. Unfortunately it is as yet too little known to found 
upon it any satisfactory conclusions. The following data are 
gleaned from De Candolle's * Prodromus, and from such speci- 
mens as are preserved in the Kew herbarium. It is probable 
