12 BENGAL PLANTS. 
that are not found in either of these areas, but that occur in the 
Garo Hills in Assam or in Tippera, which bounds the deltaic plain 
on the east, and in Chittagong, which continues, but on a more 
extensive scale, the features that characterize Tippera 
é provinces of Tippera and Chittagong are hilly throughout. 
The northern part of Tippera, where the hills are low, is largely 
covered with bamboo jungle. The southern portion is, like the 
higher part of See sa oe with dense, often rather dry 
forest. The lower part of the Chittagong hills is often covered 
with brushwood. "pei oT outer hills themselves lie cultivated 
river-valleys, while between these hills and the sea is a narrow level 
of rice-land with, towards the north, a muddy sea-face, as in 
the adjacent eastern portion of the Sundribuns. More to the south 
a@ series of low flat islands skirt the coast, while the shores have the 
same mangrove vegetation and sea-fence as the western Sundribuns. 
For the purposes of this work, the natural boundaries of the four 
western provinces, Tirhut, Behar, Chota Nagpur, and Orissa, have 
been left unchanged. As regards the first three, this treatment is 
as natural as it is convenient. ‘Tirhut, lying from west to east 
between the Gandak and the Kosi, from north to south between 
the Subhimalayan forest and the Ganges; and Behar extending 
from the Son on the west to the old bed of the Bhagirati on the 
the Upper Gangetic plain. Chota Nagpur, immediately to the 
south of Behar, similarly constitutes a direct north- -easterly ex- 
tension of the highlands of Central India. 
From one point of view it might have been advisable to deal 
with Tirhut and Behar together. It is, however, more convenient 
to separate them because Tirhut is wholly flat, whereas Behar is 
much diversified by hills, outliers from the flanks of the Chota 
Nagpur plateau. Behar, too, is appreciably drier than Tirhut, 
and these two circumstances, greater diversity of surface and less 
humidity, account for the presence in Behar of many species from 
Bandelkand, and some even from the Panjab, that are absent from 
Tirhut. Another and, though an accidental, not less important 
factor in influencing the vegetation of Tirhut is the density of the 
population. So close, in consequence, is the tilth, that throughout 
whole dietriets field is conterminous with field, and the cultivated — 
