21S 
TI1E SCIENCE OP FRONTIER DELIMITATION. 
passes from range to range by low cols, and often takes a line' of 
country which appears to be the lowest in the district, though when 
actually tested the general level must be relatively the highest. The 
effect, however, is probably produced by the line so frequently avoiding 
the high peaks and ridges. In delimiting and survey work, when no 
previous reconnaissance has been made, there is always a danger of 
adopting a direction for observing stations and for sending on carriers 
to fix camp, which, on examination, proves to be quite wrong, the 
watershed having doubled back or struck a new line, which it is im¬ 
possible to guess at till the ground is examined. The one rule to be 
followed is not to trust the watershed to take any particular direction 
beyond the point to which the eye can actually trace it. Though it 
may be in front of you and apparently going straight forward, it is 
quite possible that it will double back and be found on your left or 
right rear. The more mountainous the country, the slower the progress 
you make in the general direction of the watershed, on account of its 
twisting and breaking back. There is this compensating advantage 
about following a watershed in a mountainous country, that the survey 
work covers a wider strip of ground than if the line was a straight 
one. The one great advantage, however, of delimiting a watershed in 
a mountainous country is that you have no difficulty in seeing how it 
runs, and, wind and twist as it may, it is not possible to make a 
mistake about it. But when the watershed is followed in a flat 
country, more particularly if it is a bush covered country, the work 
becomes very laborious and errors may easily be made. If the ground 
is covered with the high elephant grass, so common in tropical 
countries, it is impossible to tell whether it is rising or falling a few 
yards away, and you may easily cross small valleys without discovering 
them. In these circumstances the grass should be burned, if it is dry 
enough, although bush fires seriously interfere with survey work, and 
make it impossible to see distant points. The watershed must be felt 
for step by step, and as soon as you find yourself in a valley, you must 
try back. Guides should always accompany the commissioners, and 
they should be constantly asked where the different waters run to. I 
have noticed that natives, however ignorant and stupid, and I have met 
many phenomenally distinguished in these respects, are generally able 
to give correctly the direction of valleys and water-lines. 
In a flat country it is difficult to decide on the exact spot for setting 
up a beacon, if the ground is covered with tall herbage or bush. A 
clinometer is of some assistance, but it is not easy to use it. A little 
practice in examining the features carefully, however, and in tracing 
the watershed between two elevations, will soon prevent errors. 
In a flat country the watershed does not bend much, and follows the 
general direction more consistently than in a mountainous country. 
(2) A parallel of latitude .— 
In following a parallel of latitude the first thing is to fix accurately 
is the latitude of the starting point, if it is not known. This is no less 
required if the line to be followed is the latitude of a certain place; the 
