I lO 
NOTES OF A BOTANIST 
CHAP. 
the rainy season are the only ones which produce 
those two grasses, and sometimes in such abun- 
dance that they become periodically choked up. 
The same thing happens also in some of the 
parana-min's. Whilst the waters are falling, the 
belt of Caapi'm extends inwards, wherever it finds 
that shallow water in which it most luxuriates ; and 
thus increases vastly in breadth. But when the 
next flood comes, the earth is gradually washed 
away from the roots of the Caapim, until, having 
no longer anything to retain it in its place, the 
loosened mass is detached from the shore and 
floats down the stream. In some cases the lower 
part of the stem is actually decayed, and thus has 
so slight a hold on the ground as to be readily dis- 
lodged by the swelling stream ; and as the stems 
are much entangled, it is only in masses they can 
be liberated. The circular Grass-islands are mostly 
the product of lakes, whose outlet has become silted 
up during the ebb of the river, and is not reopened 
until the waters, having already risen considerably, 
burst the barrier and rush like a cataract into the 
lake, liberating the Caapi'm, whirling it round and 
round, and finally carrying it ofl' to the Amazon. I 
have been in no small peril from the irruption of the 
Amazon into one of these closed channels, as I shall 
have occasion to relate shortly. 
Grass- islands are often of immense thickness. 
One which I examined on the upper Amazon con- 
sisted entirely of Paspalum pyramidale. After 
many futile attempts, I succeeded in drawing up an 
entire stem of the grass, which measured 45 feet in 
length and possessed 78 nodes ; so that, making all 
allowance for the tortuosity of the stems, the island 
