a reservoir.  It begins to flow down the Embarrah, which divides into the rivers Tzo and Teoughe.  The Tzo divides into the Tamunak’le and Mababe; the Tamunak’le discharges itself into the Zouga, and the Teoughe into the lake.  The flow begins either in March or April, and the descending waters find the channels of all these rivers dried out, except in certain pools in their beds, which have long dry spaces between them.  The lake itself is very low.  The Zouga is but a prolongation of the Tamunak’le, and an arm of the lake reaches up to the point where the one ends and the other begins.  The last is narrow and shallow, while the Zouga is broad and deep.  The narrow arm of the lake, which on the map looks like a continuation of the Zouga, has never been observed to flow either way.  It is as stagnant as the lake itself.

The Teoughe and Tamunak’le, being essentially the same river, and receiving their supplies from the same source (the Embarrah or Varra), can never outrun each other.  If either could, or if the Teoughe could fill the lake—­a thing which has never happened in modern times—­then this little arm would prove a convenient escapement to prevent inundation.  If the lake ever becomes lower than the bed of the Zouga, a little of the water of the Tamunak’le might flow into it instead of down the Zouga; we should then have the phenomenon of a river flowing two ways; but this has never been observed to take place here, and it is doubtful if it ever can occur in this locality.  The Zouga is broad and deep when it leaves the Tamunak’le, but becomes gradually narrower as you descend about two hundred miles; there it flows into Kumadau, a small lake about three or four miles broad and twelve long.  The water, which higher up begins to flow in April, does not make much progress in filling this lake till the end of June.  In September the rivers cease to flow.  When the supply has been more than usually abundant, a little water flows beyond Kumadau, in the bed first seen by us on the 4th of July; if the quantity were larger, it might go further in the dry rocky bed of the Zouga, since seen still further to the east.  The water supply of this part of the river system, as will be more fully explained further on, takes place in channels prepared for a much more copious flow.  It resembles a deserted Eastern garden, where all the embankments and canals for irrigation can be traced, but where, the main dam and sluices having been allowed to get out of repair, only a small portion can be laid under water.  In the case of the Zouga the channel is perfect, but water enough to fill the whole channel never comes down; and before it finds its way much beyond Kumadau, the upper supply ceases to run and the rest becomes evaporated.  The higher parts of its bed even are much broader and more capacious than the lower toward Kumadau.  The water is not absorbed so much as lost in filling up an empty channel, from which it is to be removed by the air and sun.  There is, I am convinced, no such thing in the country as a river running into sand and becoming lost.  The phenomenon, so convenient for geographers, haunted my fancy for years; but I have failed in discovering any thing except a most insignificant approach to it.