One Point Eight Metres Where the Ice Left Clay

The square-cut nail from March has been sitting on my desk for twelve weeks now. Pre-1900 construction, rusted down to a stub, pulled from the backyard of a house built in 1978. Someone was here before. Something stood here. Metal detecting answers what—the nail exists, it’s ferrous, it’s old—but not where exactly or how deep the foundations went or what else might be six metres down where the coil can’t reach.

Surface detection has limits. So does coring trees—you get eighty years of climate record if you extract carefully, but only what’s recorded in wood. The ground keeps older records. Bedrock, fill layers, buried footings, water tables. Acoustic waves can read them.

Seismic refraction mapping: plant geophones in a 24-meter line, strike a steel plate with a sledgehammer, record the millisecond-scale arrival times of compression waves refracted through subsurface layers, calculate depths from velocity contrasts. It’s what civil engineers use before breaking ground. What archaeologists use to map ruins without digging. What Apollo astronauts deployed on the Moon with explosives and thumper devices to map lunar crust.

I borrowed twelve 10 Hz geophones from a retired geotechnical engineer two neighbourhoods over. He did foundation surveys in the ’90s, hasn’t touched them in a decade. They came in a foam-lined case with a 12-channel data logger the size of a hardcover book and a steel striker plate that weighs 9 kg by itself.

A geophone is electromagnetic induction in reverse. Spring-mounted wire coil, case-mounted permanent magnet. When ground moves, the coil cuts through the magnetic field and generates voltage proportional to velocity—about 30 volts per metre per second. It’s a microphone for the earth. Same physics as a guitar pickup, same physics as the tuning coil in the crystal radio, same discipline: amplify millivolt signals without drowning in noise.

I set them up this morning in my backyard. Twelve geophones spaced two metres apart, 24-meter array running north-south parallel to the fence. Each one gets pushed into the soil until the spiked base seats firmly—loose coupling kills the signal. BNC cables back to the logger. The logger triggers on impact: sledgehammer hits the striker plate, shock wave propagates, first arrivals get timestamped to the nearest 0.1 millisecond.

The striker plate went down at the south end, flat on bare soil. Not on grass—I learned that on the first swing. The plate has to couple cleanly or the energy scatters. I cleared a 30 cm square patch, tamped it flat, centered the plate.

Overhead swing. Two-handed grip. 9 kg head dropping through a full arc. The impact is loud—a flat crack that echoes off the garage and rings in my ears for five seconds after. The ground moves. All twelve geophones register.

The data logger screen shows twelve traces, each one a wiggling line plotting voltage over time. The first arrival on geophone 1—the one closest to the striker—comes at 4.2 milliseconds. Geophone 2: 7.8 ms. Geophone 3: 11.1 ms. By geophone 6 the times stop increasing linearly. Geophone 7 arrives at 19.3 ms, but geophone 8 comes in at 19.0 ms. Faster. The wave took a shortcut.

That’s refraction. Compression waves travel slower through loose topsoil (maybe 400 m/s) than through denser clay or bedrock (2000-4000 m/s). When a wave hits the boundary at the critical angle—Snell’s Law, same optics that bends light through a prism—it refracts along the interface at the higher velocity, then bends back up to the surface. Beyond a certain distance, the refracted wave beats the direct wave, arriving first even though it traveled deeper.

I plot the data by hand on graph paper: distance along the X-axis, arrival time on the Y-axis. The first six points fall on a steep slope—direct wave through topsoil. Points 7 through 12 fall on a shallower slope—refracted wave through a faster layer. Two lines. Two slopes. The break between them is the crossover distance, and from that I can calculate the depth to the layer boundary.

The math is trigonometry. Velocity of the top layer comes from the slope of the first line: distance divided by time, about 380 m/s. Velocity of the second layer from the second slope: 2100 m/s. The depth to the interface is the intercept time multiplied by the top-layer velocity, divided by twice the cosine of the critical angle.

I get 1.8 metres.

There’s a clay hardpan 1.8 metres down under my backyard. Denser than the sandy loam on top, loose enough that it’s not bedrock. Probably glacial till from the last ice age, compacted under a kilometre of ice and then exposed when it melted. That’s what stopped my hypothetical 19th-century foundation from going deeper. That’s the layer the nail sits above.

Second shot from the north end. Same setup, reversed geometry. If the layer is continuous, the arrival times should invert symmetrically. They do. Depth calculation comes out to 1.7 metres. Close enough—ground isn’t uniform, and I’m swinging a hammer by hand, not dropping a controlled weight.

The method breaks if a slower layer sits under a faster one. Velocity has to increase with depth or the refracted wave never comes back to the surface. You’d read two layers, miss the third entirely. A “hidden layer,” undetectable by refraction but sitting there anyway. The only clue would be drilling and finding it physically, or running electrical resistivity, or using reflection seismology, which measures echoes instead of refractions.

I don’t know what’s under the clay. Bedrock, probably, but maybe not. Maybe water-saturated silt. Maybe another till layer, softer, deposited during an earlier glaciation and then buried. The data won’t tell me. Refraction has limits.

But now I know the clay is there. I know how deep it is. I know the nail came from fill above it, not structure embedded in it. And I know that if I wanted to map whether the foundations of whatever stood here in 1890 extended under my garage, I’d need to run a longer line, swing harder, and hope the footings were shallow enough to refract a signal back.

Twelve strikes. Twenty-four arrival times. Two layers resolved. The rest is math and inference.

The geophones are coiled in their case. The striker plate is leaning against the garage. My hands are sore from the impacts traveling back up through the handle. Tomorrow I’ll see if the clay layer is continuous under the front yard, or if it slopes, or if there’s a buried anomaly where the Victorian-era structure might have poured footings.

Today, though, I’ve mapped 1.8 metres of history I couldn’t reach with a metal detector or an auger.