32 M. Jules de Guerne on a 



the obstacles of its banks, spreads out into a labyrinth of 

 * marigots,' which penetrates south of the major current to a 

 distance of 150 kilometres into the interior. During seven 

 months of the year this district of n)arigots, which extends to 

 the south of the river, below the meridian of Timbuctu, is cut 

 off from free communication with the Niger : there is nothing 

 to be seen but scattered lakes, pools of stagnant water, 

 obliterated channels, dead water, and damp hollows filled 

 with reed-beds ; but the river-floods, which overflow in a 

 lateral direction along depressions in the ground, transform 

 these into a network of flowing streams, and during four or 

 five months large boats can be navigated freely in an immense 

 maze of lakes and channels, in which the currents change 

 without apparent rule, according to tlie force and the direction 

 of the winds, the abundance of I'ain, or the heat of the sun. 

 In this part of Central Africa nature assumes an aspect Avhich 

 resembles in certain respects that of Sweden, with its lakes 

 of every variety of form, its creeks, its firths, and its 

 rivers 



" In one channel the water is open and deep, 



unimpeded by vegetation ; in another it is encumbered with 

 weeds, through which it is difficult for boats to force a passage, 

 and which occasionally become detached in the shape of 

 floating islands, which eventually form barriers at the bends 

 of the stream which are impassable for boats : every year 

 traders are compelled to change their route through the 

 tortuous intricacies of this vast labyrinth " 



But even if we exclude the possibility that the higher 

 waters above the rapids may be reached at last in an indirect 

 manner by means of the floods, the rapids themselves might 

 still be directly surmounted by the Hydroids ; for it has been 

 found at Hamburg that the water-mains are invaded by 

 colonies of Cordylojyhora lacustris, Allman, which develop 

 there in all directions in spite of the intensity of the current*. 

 Besides, if the actual rocks in the torrent do not furnish these 

 animals with a base upon which to grow, they will attach 

 themselves to dead or living shells f- This is the habit of 



* Kraepeliu, " Die Fauna der Hamburger Wasserleitung," Abliandl. a. 

 d. Geb. der Naturwissenseb., beravisg. v. Nat. Vereiu Hamburg, ix. 1886, 

 pp. 5, 6. Allman bad previously recorded tbe presence of Covch/lophora 

 in tbe water-maius of London (Allman, * A Monograpb of tbe Gymno- 

 blastic or Tiibularian Hydroids,' Kay Society, London, 1871-72, p. '25'^). 



t It lias long been known tbat Cordijlophora lacustris is frequently met 

 witb upon tbe sbell of Dreissena poli/morpha, Pallas. In tbis situation 

 it was found in Paris itself, in tbe great reservoir of tbe Jardin des Plantes 

 (E. Perrier, " Sur I'existeuce a Paris du Cordi/lopkora lacustris, Allman," 

 Arcb. zool. experim. ii. 1873, Notes et Revue, p. xvii). 



