"In San Paolo also and Paraná the region of afforestation is not limited by the ridge of the _serra_. Forests and prairies alternate on the plateau. The fires which the Indians used to light in the savannahs have destroyed the forest in places; yet man has played but a little part in the present distribution of vegetation. The forest has persisted wherever the natural conditions were favourable, holding tenaciously to the humid slopes of the hills or to rich and fertile soils. Certain soils, either by reason of their richness or their moisture, particularly favour the forest, while on lighter soils the trees can ill resist the drought. The diabasic soils of San Paolo are always covered with a mantle of forest; so much so that a map of the forests would be equivalent to a geological map. "The forest of the plateau, intersected as it is by stretches of prairie, is less dense and less exuberant than the forest the _serra_; and as we approach the south the difference is yet more evident. Towards the boundaries of San Paolo and Paraná the tropical trees are replaced by resinous varieties. The immense pines of the Paraná, with straight trunks and wide, flattened crests, whose shape is rather reminiscent of that of a candelabrum with seven branches, cover with their sombre grey the wooded portions of the plateau from the Paranápanema to beyond the Uruguay. With their open foliage, pervious to the light, these woods resemble the pinewoods of Europe. "To find the tropical forest once more we must push as far as the Serra Geral, whose southern slopes run down towards the prairies of Rio Grande, as on the east they descend towards the sea. There, on the basaltic flanks of the _serra_, is a last fragment of the tropical woods. In magnificence it almost equals the forests of Rio or of Santos. It is the equatorial forest that makes the continuity of the _serra_, not its geological constitution. When the Brazilians speak of the _serra_ they think of the forest rather than of the mountains. Incautious cartographers, who have worked from second-hand data, which they have not always interpreted correctly, have sown the map of Rio Grande with a large number of imaginary ranges. One seeks them in vain when traversing the country; but one finds, in their place, the forests which the inhabitants call _serras_; the term for mountain has become, by the latent logic of language, the term for forest. Nothing could better emphasize the importance of vegetation in the Brazilian landscape; it effaces all other characteristics.