BITTER WATERS 295

therefore, that the music was produced by Sa'dan's movements'-
from one undisturbed zone to another and, when he came
down, having had enough of that form of amusement, I
went up in his place armed with a bottle (to collect a sample
of the sand), note-book and watch.

I found a very light north wind (or north-north-east)
blowing from behind the dune-range which enclosed the
hollow. The summit was about 200 feet above the well,
and the slope of the sand, converging on the depression in an
almost perfect arc representing perhaps two-thirds of a
circle, was exceedingly steep and soft, though by no means
uniformly steep all round. And it seemed to me that the
music only arose when the sand was sliding down the more
steeply inclined parts. I could not therefore produce the
phenomenon absolutely at will unless I happened to be on
the brink of a section suitable for the performance. On
three separate occasions, however, music followed on my
disturbance of the loose sand at the summit—a loud, har-
monious, organ-like booming.

I stood on the brink of the amphitheatre and pushed
the thick soft sand of the summit downwards with my
feet. Thereupon one or more broad shute-like bands of
sand began to move steadily down the slope, setting up
a distinctly audible frictional sound—just such a sound
indeed as one would expect in the circumstances, a loud
sound of rubbing or grating as of a rough body sliding over
a sandy floor. This sound—in which was no suspicion of
music and which would not be audible at any considerable
distance—increased in a steady crescendo until the moving
mass seemed to have progressed about 50 feet down the slope,
whereupon the quality of the sound changed abruptly from
the grating to a booming.

The loudness of the sound seemed to depend on the
quantity of the sand in movement, but in all cases there was
a crescendo of sound which only diminished slightly before
stopping altogether quite abruptly—as abruptly as the
engine of a motor car—and simultaneously, as it seemed to
me, with a sudden cessation of all movement on the slope.
Each item of actual booming lasted between two and three