small lake called Bemba. As we proceeded west, we
passed over the sources not only of the Loangwa, but of another stream,
called Moitawa or Moitala, which was represented to be the main feeder of
Lake Bemba. This would be of little importance, but for the fact that
the considerable river Luapula, or Loapula is said to flow out of Bemba
to the westward, and then to spread out into another and much larger
lake, named Moero, or Moelo. Flowing still further in the same
direction, the Loapula forms Lake Mofue, or Mofu, and after this it is
said to pass the town of Cazembe, bend to the north, and enter Lake
Tanganyika. Whither the water went after it entered the last lake, no
one would venture an assertion. But that the course indicated is the
true watershed of that part of the country, we believe from the unvarying
opinion of native travellers. There could be no doubt that our
informants had been in the country beyond Cazembe's, for they knew and
described chiefs whom we afterwards met about thirty-five or forty miles
west of his town. The Lualaba is said to flow into the Loapula--and
when, for the sake of testing the accuracy of the travelled, it was
asserted that all the water of the region round the town of Cazembe
flowed into the Luambadzi, or Luambezi (Zambesi), they remarked with a
smile, "He says, that the Loapula flows into the Zambesi--did you ever
hear such nonsense?" or words to that effect. We were forced to admit,
that according to native accounts, our previous impression of the
Zambesi's draining the country about Cazembe's had been a mistake. Their
geographical opinions are now only stated, without any further comment
than that the itinerary given by the Arabs and others shows that the
Loapula is twice crossed on the way to Cazembe's; and we may add that we
have never found any difficulty from the alleged incapacity of the negro
to tell which way a river flows.

The boiling-point of water showed a descent, from the edge of the plateau
to our furthest point west, of 170 feet; but this can only be considered
as an approximation, and no dependence could have been placed on it, had
we not had the courses of the streams to confirm this rather rough mode
of ascertaining altitudes. The slope, as shown by the watershed, was to
the "Loangwa of the Maravi," and towards the Moitala, or south-west,
west, and north-west. After we leave the feeders of Lake Nyassa, the
water drains towards the centre of the contine