of the snow, at the top of the slope of 3 or 4 feet which led down the
surface of the ice, and then stood to watch the effect of the current
on the flame. The experiment proved that the currents alternated, and,
as I fancied, regularly; and in order to determine, if possible, the
law of this alternation, I observed with my watch the exact duration
of each current. For twenty-two seconds the flame of the _bougie_ was
blown away from the entrance, so strongly as to assume a horizontal
position, and almost to leave the wick: then the current ceased, and
the flame rose with a stately air to a vertical position, moving down
again steadily till it became once more horizontal, but now pointing
in towards the cave. This change occupied in all four seconds; and the
current inwards lasted--like the outward current--twenty-two seconds,
and then the whole phenomenon was repeated. The currents kept such
good time, that when I stood beyond their reach, and turned my back, I
was enabled to announce each change with perfect precision. On one
occasion, the flame performed its semicircle in a horizontal instead
of a vertical plane, moving round the wick in the shape of a
pea-flower. The day was very still, so that no external winds could
have anything to do with this singular alternation; and, indeed, the
pit was so completely sheltered by its shape, that a storm might have
raged outside without producing any perceptible effect below. It would
be difficult to explain the regularity of these opposite currents, but
it is not so difficult to see that some such oscillation might be
expected. It will be better, however, to defer any suggestions on this
point till the glacière has been more fully described.

[Illustration: GROUND PLAN OF THE GLACIÈRE OF MONTHÉZY. Note: The
candle stood at this point.]

We passed down at length through the low archway, and stood on the floor