orse on the
road below,"--a legend which shows it must have attracted _their_ notice
too.

I selected as the special scene of exploration this morning, a deep
ravine of the boulder-clay, which had been recently deepened still more
by the waters of a mill-pond, that had burst during a thunder-shower,
and, after scooping out for themselves a bed in the clay some twelve or
fifteen feet deep, where there had been formerly merely a shallow drain,
had then tumbled into the ravine, and bared it to the rock. The
sandstones of the district, soft and not very durable, show the
scratched and polished surfaces but indifferently well, and, when
exposed to the weather, soon lose them; but in the bottom of the runnel
by which the ravine is swept I found them exceedingly well marked,--the
polish as decided as the soft red stone could receive, and the lines of
scratching running in their general bearing due east and west, at nearly
right angles with the course of the stream. Wherever the rock had been
laid bare during the last few months, _there_ were the markings;
wherever it had been laid bare for a few twelvemonths, they were gone. I
next marked a circumstance which has now for several years been
attracting my attention, and which I have found an invariable
characteristic of the true boulder-clay. Not only do the rocks on which
the deposit rests bear the scratched and polished surfaces, but in every
instance the fragments of stone which it incloses bear the scratchings
also, if from their character capable of receiving and retaining such
markings, and neither of too coarse a grain nor of too hard a quality.
If of limestone, or of a coherent shale, or of a close, finely-grained
sandstone, or of a yielding trap, they are scratched and
polished,--invariably on one, most commonly on both their sides; and it
is a noticeable circumstance, that the lines of the scratchings occur,
in at least nine cases out of every ten, in the lines of their longer
axes. When decidedly oblong or spindle-shaped, the scratchings run
lengthwise, preserving in most cases, on the under and upper sides, when
both surfaces are scratched, a parallelism singularly exact; whereas,
when of a broader form, so that the length and breadth nearly
approximate,--though the lines generally find out the longer axis, and
run in that direction,--they are less exact in their parallelism, and
are occasionally traversed by cross furrows. Of such certain occurrence
is this longitu