e most frequent, and
they cause the least damage to the slightly-built habitations. Vertical
shocks are most severe; they rend the walls, and raise the houses out of
their foundations. The greatest vertical shock I ever felt was on the
4th of July, 1839, at half-past seven in the evening, when I was in the
old forests of the Chanchamoyo territory. Before my hut there was an
immense stem of a felled tree, which lay with its lower end on the stump
of the root. I was leaning against it and reading, when suddenly, by a
violent movement, the stem rose about a foot and a half, and I was
thrown backwards over it. By the same shock the neighboring river,
Aynamayo, was dislodged from its bed, and its course thereby changed for
a considerable length of way.

I have had no experience of the rotatory movements of earthquakes.
According to the statements of all who have observed them, they are very
destructive, though uncommon. In Lima I have often felt a kind of
concussion, which accords with that term in the strictest sense of the
word. This movement had nothing in common with what may be called an
oscillation, a shock, or a twirl: it was a passing sensation, similar to
that which is felt when a man seizes another unexpectedly by the
shoulder, and shakes him; or like the vibration felt on board a ship
when the anchor is cast, at the moment it strikes the ground. I believe
it is caused by short, rapid, irregular horizontal oscillations. The
irregularity of the vibrations is attended by much danger, for very
slight earthquakes of that kind tear away joists from their joinings,
and throw down roofs, leaving the walls standing, which, in all other
kinds of commotion, usually suffer first, and most severely.

Humboldt says that the regularity of the hourly variations of the
magnetic needle and the atmospheric pressure is undisturbed on
earthquake days within the tropics. In seventeen observations, which I
made during earthquakes in Lima with a good Lefevre barometer, I found,
in fifteen instances, the position of the mercury quite unaltered. On
one occasion, shortly before a commotion, I observed it 2.4 lines lower
than it had been two hours before. Another time, I observed, also on
the approach of the shock and during the twelve following hours, a
remarkable rising and sinking in the column. During these observations
the atmosphere was entirely tranquil.

Atmospheric phenomena are frequent, but not infallible prognostics of
an eart