officer never expects his shells actually to kill or disable any very
large number of the enemy if they are protected by deep and
well-constructed earthworks. Of course, if a shell falls plump into a
trench it is pretty certain to play havoc with the defenders, but, when
one considers that the mouth of a trench is some five or six feet wide,
it is easy to realise the difficulty of dropping a shell into the narrow
opening at a range, say, of 4,000 yards. Moreover, some of the more
elaborate Boer trenches are so cleverly constructed in a waving line
like a succession of S's, that even if a shell does succeed in pitching
into one bit of the curve it makes things uncomfortable only for the two
or three men who occupy that portion of the earthwork. No, the real
value of artillery in attack is to shake the enemy and keep down his
rifle fire. If shells are accurately fired the tops of trenches may be
swept by a constant rain of shrapnel bullets, under which the enemy's
riflemen will of necessity suffer when they expose their heads and
shoulders to take aim over the parapet. But even in this case the shell
fire must be extremely accurate if it is to be of any great use. If
shrapnel shells burst well, some thirty yards in front of the enemy, the
force of the bullets released by the explosion is terrific; if, on the
other hand, the shells burst high up in the air, 150 yards in front, you
might almost keep off the bullets with an umbrella; and one sometimes
hears of these missiles being actually found in the pockets of
combatants. At Omdurman our shells played tremendous havoc with the
dense masses of the enemy; but here the Dervishes advanced to the attack
in broad daylight and over a flat plain absolutely devoid of cover, and
with its "ranges" well known and marked out beforehand.

In one of our southward journeys with a load of wounded men we passed, a
little below Graspan, through the midst of a swarm of locusts. We pulled
up the windows and so kept the wards free from these clumsy insects. At