A dry-stone wall holds because its pieces
lean on one another. No mortar hides a
poor fit. Broad stones bridge joints,
small wedges prevent rocking, and the wall
5
tilts slightly into the soil it supports.
My uncle Joas explained these principles
while we walked uphill through the
abandoned orchard, but from a distance the
terrace wall looked less like a structure
10
than a gray line interrupted by one muddy
gap.
I had come to help close that gap before
the autumn storms. Joas brought two
hammers, a level, and no drawing. “The
15
wall is a sentence written by gravity,” he
said. “Before changing a word, find out
what the sentence is doing.” The metaphor
made our work seem less like replacing
missing objects and more like interpreting
20
an argument whose author could no longer
answer questions.
At the break, I recorded what we could
see. The wall stood chest-high on either
side, but its lowest stones curved inward
25
toward the opening. Moss was thick above
the gap and absent from its floor. A fan
of pale gravel spread downhill. I called
the pattern storm damage: water had torn
stones out and washed gravel through. Joas
30
agreed that water had traveled there. He
was less certain that it had caused a
failure.
We worked by alternating observation and
interpretation. I uncovered a flat stone;
35
Joas asked whether it had fallen or been
laid level. He found two narrow upright
stones facing each other; I proposed that
roots had rotated them into parallel
positions. Each guess directed the next
40
scrape of our trowels, yet neither of us
treated a guess as a discovery. The ground
kept revising our confidence.
By noon, we had exposed a shallow channel
paved with fist-size pebbles. Its sides
45
met the wall’s inward-curving base
cleanly. Under one cheek stone, Joas found
a square of rusted iron pierced by a round
hole. He threaded a twig through it and
showed me how the piece could once have
50
held a removable wooden board. The
supposed collapse was an overflow gate.
During heavy rain, a farmer could lift the
board and let water leave the terrace
before saturated soil pushed the entire
55
wall outward.
Our new explanation did not make every
loose stone intentional. Several upper
courses had indeed fallen, probably after
the board decayed and the channel clogged.
60
But the finding reversed our original
plan. Filling the opening from foundation
to cap would restore a continuous-looking
wall while sealing its pressure release.
Joas set aside a handsome rectangular
65
stone I had chosen for the gap. “Wrong
place, good stone,” he said, and later
used it as a cap farther uphill. We
rebuilt the two sides, lined the old
channel, and formed a low notch where a
70
new oak board could slide. From the path,
the finished section still looked
incomplete. I disliked that appearance
until Joas poured a bucket onto the
terrace. Water gathered behind the wall,
75
found the notch, and slipped through
without carrying soil.
This account explains how we read and
repaired one wall; it is not a campaign to
preserve every old gap. A breach may be
80
simple neglect, and a historic feature may
no longer suit the land around it. In our
case, the channel discharged into a bed
where my aunt grew rosemary and hyssop.
Keeping the spillway protected the terrace
85
but exposed those plants to occasional
flooding.
We moved the largest herbs to higher
ground and left gravel between the
remaining ones. That solution distributed
90
the cost rather than abolishing it.
Closing the gap would have protected the
herb bed during ordinary rain but
endangered the wall in an extreme storm.
Reopening the channel protected the wall
95
by assigning excess water a route through
part of the garden.
The first hard rain arrived two weeks
later. I watched water darken the orchard
soil and pool against the terrace. For
100
several minutes nothing reached the notch,
and I wondered whether our interpretation
had been merely an elegant story. Then a
narrow stream passed beneath the board.
The wall held. Two remaining hyssop plants
105
bent under muddy water, but the relocated
rosemary stood dry above them.
Joas did not call the repair a success
until we checked the wall the next
morning. No stone had shifted. He
110
photographed the notch, the flattened
hyssop, and the intact terrace in the same
frame. The picture refused a triumphant
ending: evidence of protection occupied
the upper half, evidence of its price the
115
lower. What we had restored was not an
unbroken line. It was an old arrangement
for deciding where water, and therefore
damage, could go.