An elementary treatment of the theory of spinning tops and gyroscopic motion, by Harold Crabtree.

CHAPTER III.
DISCUSSION OF THE PHENOMENA IESCRIBED IN THE
INTRODUCTORY CHAPTER.
46. If we turn to the Introductory Chapter we shall now
see the reasons for the paradoxical behaviour of the tops there
described. To take the first instance mentioned if a top is
placed on its toe while spinning, the torque, which would cause
it to fall over if it were not spinning, fails in this case to
turn over the axle but rotates it, and so the top precesses.
Also, if we consider the rule given in Art. 38 for the direction
of precession, we see why an ordinary top precesses in the
direction of its spin, when viewed, say, from above (Fig. I.),
while that in Fig. III. (a) precesses in the reverse direction:
for in the latter case the centre of gravity is below the point
of support and the torque acts in the reverse direction.
47. Let us now consider what causes a top that has been spun
at an inclination to the vertical to rise to a more vertical
(though not necessarily quite vertical) position before settling
down to a steady motion. Suppose we take an ordinary top
and spin it on a rough board. When it has settled down to a
motion in which its axle is practically at a constant angle to
the vertical, let us drag the board round in the same direction
as the precession. It will be found that the centre of gravity
of the top rises, and some of us may remember that the method
of resuscitating a dying top is to get it on to one's hand and
drag the hand round in the necessary direction. Similarly, if
we draw the hand round in the opposite direction the top falls;
so that we can sum up our experiment in the two following
statements:
(i) hurry the precession, the top rises;
(ii) retard the precession, the top falls.
Hence, when a top is spun at an inclination to the vertical
and rises, there must be at first something that is hurrying
the precession, and eventually (since the top afterwards falls)
something that retards it.