narrowest part to a width of only twenty-five toises, bounded on each margin by lofty perpendicular cliffs, at the end of which the Andes are fairly passed, and the river emerges on the great plain."[FN#31] Thus the Yellala belongs to the class of obstructed rapids like those of the Nile, compared with the unobstructed, of which a fine specimen is the St. Lawrence.  It reminded me strongly of the Busa (Boussa) described by Richard Lander, where the breadth of the Niger is reduced to a stone-throw, and the stream is broken by black rugged rocks arising from mid-channel.  It is probably a less marked feature than the Congo, for in June, after the “Malka” or fourteen days of incessant rain, the author speaks of whirlpools, not of a regular break.

I thus make the distance of the Yellala from the mouth between 116 and 117 miles and the total fall 390 feet, of which about one half (195) occurs in the sixty-four miles between Boma and the Yellala:  of this figure again 100 feet belong to the section of five miles between the Vivi and the Great Rapids.  The Zambeze, according to Dr. Livingstone ("First Expedition,” p. 284), has a steeper declivity than some other great rivers, reaching even 7 inches per mile.  With 3 to 4 inches, the Ganges, the Amazonas, and the Mississippi flow at the rate of three knots an hour in the lowest season and five or six during the flood:  what, then, may be expected from the Nzadi?

According to the people, beyond the small upper fall where projections shut out the view, the channel smoothens for a short space and carries canoes.  Native travellers from Nkulu usually take the mountain-path cutting across an easterly bend of the bed to Banza Menzi, the Manzy of Tuckey’s text and the Menzi Macooloo of his map.  It is situated on a level platform 9 miles north of Nkulu, and they find the stream still violent.  The second march is to Banza Ninga, by the First Expedition called “Inga,” an indirect line of five hours = 15 miles.  The third, of about the same distance, makes Banza Mavunda where, 20 to 24 miles above the Yellala, Tuckey found the river once more navigable, clear in the middle and flowing at the rate of two miles an hour—­a retardation evidently caused by the rapids beyond:  I have remarked this effect in the Brazilian “Cachoeiras."[FN#32] Above it the Nzadi widens, and canoeing is practicable with portages at the two Sangallas.  The southern feature, double like the Yellala, shows an upper and a lower break, separated by two miles, the rapids being formed as usual by sunken ledges of rock.  Two days’ paddling lead to the northern or highest Sangalla, which obstructs the stream for 22 miles:  Tuckey (p. 184) makes his Songo Sangalla contain three rapids; Prof.  Smith, whose topography is painfully vague, doubles the number, at the same time he makes Sanga Jalala (p. 327) the “uppermost fall but one and the highest.”  Finally, at Nsundi (on the map Soondy N’sanga), which was reached on Sept. 9, a picturesque sandy cove at the opening of a creek behind along projecting point, begins a lake-like river, three miles broad, with fine open country on both banks:  the explorer describes it as “beautiful scenery equal to anything on the banks of the Thames.”