Eighteen Centimetres Down to Robertson Number Two

7:18 AM — Empty pockets protocol: wallet on desk, keys on hook, watch off, belt buckle (brass, non-ferrous, still off). Phone stays in truck. Gradiometer reads ferrous contamination to 2-3 meter radius. One forgotten coin ruins a morning’s data.

Borrowed Foerster FEREX 4.032 from the same retired geotechnical engineer who lent me the geophones. Vertical gradiometer configuration: two fluxgate sensors on a fibreglass staff, 0.65 m separation. Lower sensor 20 cm off ground, upper at 85 cm. Digital readout box strapped to the staff, sample rate 5 Hz. Zeroed in the driveway away from the garage door (steel rails, massive magnetic signature). Background reading: 51,247 nT. Earth’s field at 53°N.

7:31 AM — The clay layer from last week sits at 1.8 m. Too deep for GPR in this soil. Too deep for the metal detector coil. But if there’s rebar in a footing, or cast iron pipe, or Victorian-era structural iron embedded in that till, the magnetic signature extends upward through everything. Magnetometry doesn’t care about dielectric boundaries or acoustic impedance. Just ferrous mass and its geometry.

The gradiometer subtracts. Top sensor reads 51,247 nT. Bottom sensor reads 51,247 nT. Difference: 0 nT. Walk over a buried nail and the bottom sensor sees 51,251 nT while the top stays flat. Gradient: +4 nT. That’s the anomaly. Solar wind is pushing 30 nT swings across both sensors simultaneously—diurnal variation, ionospheric buffeting—but subtraction cancels it completely. You’re left with spatial gradient only. What’s close to the lower sensor that isn’t close to the upper.

7:44 AM — Grid layout: 8 m × 12 m rectangle in the front yard, starting from the corner stake I pounded for the seismic line. Nylon string every 1 m in both directions. Walking speed dictates resolution. At 5 Hz sample rate and 0.5 m/s pace, I get one reading every 10 cm. Slower than that and my legs cramp. Faster and the grid spacing degrades.

7:52 AM — First transect. North-south line along the western boundary. Staff vertical, sensors level, walk heel-to-toe at a metronome pace. The readout updates every 0.2 seconds: +0.2 nT, -0.1 nT, +0.3 nT, -0.4 nT. Noise floor. Random fluctuations from sensor electronics, quantization error, thermal drift. Magnitude less than 1 nT means nothing’s there.

8:01 AM — Second transect, 1 m east of the first. At the 4 m mark the readout spikes: +8.7 nT. Hold position, check for repeatability. Still reading +8.6 nT. Step backward 0.5 m: +2.1 nT. Step forward past it: +9.2 nT at 4.2 m, then dropping to +3.8 nT at 4.5 m. Step sideways off the grid line: anomaly shifts. It’s real, it’s localized, it’s shallow.

8:09 AM — Mark the spot with a flag. Don’t dig yet. Finish the grid first, map the whole anomaly field, then interpret. Archaeological discipline.

8:14 AM — Third transect. Same anomaly at 4 m, reading +11.3 nT now. Stronger directly above the source. Fourth transect: +7.2 nT. Fifth: +3.1 nT. By the sixth it’s gone. The anomaly is elongated north-south, maybe 2 m long, peak at transect 3.

8:41 AM — Twelve transects complete. Data logged by hand on graph paper: X = position along grid, Y = transect number, Z = gradient in nT. Contour the highs. The pattern is a dipole. Positive lobe on the south, negative on the north. That’s diagnostic. Ferrous object aligned with Earth’s field, magnetized by induction. South end is the induced north pole, north end is the induced south pole. Closer to the bottom sensor on the south, farther on the north. Geometry tells you orientation.

8:58 AM — Second anomaly at the eastern edge. Smaller, +3.4 nT peak, symmetrical. Probably round. Nail, bolt, rebar stub.

9:03 AM — The large anomaly is 15 cm deep. I know that without digging. Depth estimate from the width of the positive lobe: magnetic fields fall off with the cube of distance, so if the anomaly reads half-peak at 0.8 m lateral offset, the source is roughly 0.8 m below the lower sensor. Lower sensor is 20 cm off the ground. 20 cm air + 80 cm soil ≈ 1 m total depth. But the clay starts at 1.8 m and the square-cut nail was in fill above it. This is shallower. Recent fill, not Victorian.

9:12 AM — Dig the large anomaly. Trowel down through sandy loam. At 18 cm: steel. Screwdriver, Robertson #2, red plastic handle half-disintegrated. Dropped during construction, probably 1978 when the house went up. Non-archaeological.

9:17 AM — Dig the small one anyway. At 11 cm: bolt, hex head, galvanized, 8 mm × 40 mm. Also modern.

9:23 AM — Lesson: magnetometry finds ferrous. Doesn’t discriminate age. A nail from 1890 reads identically to a nail from 1990 unless you dig and check the cut style. The method is fast, non-invasive, high-coverage—but every anomaly might be trash.

9:31 AM — Extend the grid south toward where the seismic line ran. If there’s a foundation footing with rebar, or cast iron drainage, or wrought-iron structural elements, it should read stronger than a screwdriver. Footings are large, continuous, deep. Different anomaly morphology.

9:47 AM — Nothing. Gradient stays within ±1 nT across the entire southern extension. Either the foundation is too deep (unlikely—Victorian footings rarely went below 1 m in this area), or it’s constructed without ferrous materials (stone and lime mortar, possible), or it was never here to begin with and the nail came from something else. A fence. A shed. Debris dumped as fill.

10:02 AM — Survey complete. Logged 142 data points. Two modern ferrous anomalies confirmed and excavated. No structural signatures. The Victorian question remains open.

Gradiometer back in its case. Boots checked for embedded nails before getting in the truck. Screwdriver and bolt sitting on the tailgate. Subtract the background, keep the gradient. Two sensors, one question, zero foundations.