I arrived at the Marrow Creek reservoir a little after nine.
Fog clung to the surface, thick enough to dull the light.
Somewhere beneath that still water lay the old agricultural station — WMC-22 — shut down after the flood in the late seventies.
My task was simple: confirm whether any of the original soil telemetry beacons still emitted a signal.
They had been designed to broadcast low-band pulses, measuring moisture across the orchard rows that once stood here. The trees had long rotted away.
Only the towers remained — six rusted spires half-sunk into the bank, panels coated in algae.
When I powered one on, the air changed.
The frequency was supposed to be below the threshold of hearing, yet I felt it in my chest — a steady vibration, about fifty hertz, the same pitch the archive lights make when their ballast begins to fail.
I spoke a calibration phrase into the recorder:
“Field test. Soil telemetry. Finch conducting.”
The transmitter pulsed once. A clean waveform. No interference.
Procedure complete. Or at least, it should have been.
By early afternoon, the data looked wrong.
A second waveform overlapped my test signal — faint, but rhythmic, like something repeating a few miles away.
When I filtered the noise, I heard a woman’s voice threaded through the carrier wave:
“…county weather, reporting clear skies…”
The recording ended in static.
A perfect ten-second fade.
The last documented broadcast from WMC-22 had been the day before the flood.
I told myself it was an atmospheric reflection — an old transmission refracting off the water.
But when I replayed the file, the inflection had changed — the same words, newly spoken.
I checked the beacon.
Its indicator light was pulsing again, though I hadn’t scheduled a second test.
By dusk, interference had spread to nearby bands.
Even my vehicle radio — long dead from disuse — buzzed faintly with the same voice, half-submerged in static.
When I switched it off, the speaker kept vibrating, a breath against the air.
I shut down power to the array.
The sound persisted — not from the equipment, but from the ground itself.
A low hum rising through the soil, like something stirring beneath the surface.
I scraped a bit of earth near the transmitter’s base. The dirt was warm and smelled faintly sweet, like fruit beginning to rot.
When I touched it, the hum grew louder.
I recorded another note:
“Residual feedback loop. Possible piezoelectric discharge. Continue observation until nightfall.”
An hour later, the beacon reactivated.
I hadn’t restored power.
The broadcast was clearer this time — the voice older, slower:
“…county weather, orchard… winds south…”
Then static.
A rhythm pulsing underneath, almost like breathing.
The indicator light blinked in time with it, reflecting across the reservoir in soft green flashes.
For a moment I thought I saw shapes beneath the water — rows of trees shifting as though a current moved through them.
The hum deepened.
Every instrument needle on my console drifted upward in unison — soil saturation, barometric pressure, electrical potential.
All rising together.
I keyed the mic.
“Station WMC-22, this is Finch. Conducting manual override. Please confirm transmission.”
There was a pause, then my own voice came back — flat, filtered, repeating the same words.
But when I compared the timestamps, the echo had spoken first.
I shut everything down again.
Still, the voice lingered in the speaker, fading only when I stepped away from the array.
The moment I turned back, it returned.
Around nine, condensation began forming on the transmitter’s casing.
Not water — something thicker, amber-colored, smelling faintly of apples.
It hissed when it touched metal.
The ground trembled once — not enough to register as seismic, more like breath expanding under the soil.
I packed the recorder, sealed the logs, and prepared to leave.
Before disconnecting the final cable, I noticed the transmitter display flicker to life.
A single line scrolled across the screen:
WMC-22 — LIVE
Then my own voice:
“Cut the signal. You already started it.”
The drive back took less than an hour.
Halfway along the access road, the dashboard radio turned on by itself.
The same weather report played through the static — distant but clear.
In the background, beneath the voice, came a faint sound: leaves moving in slow rhythm, though there was no wind.
I parked at the archive just before eleven.
Even with the ignition off, the radio kept transmitting, low and steady.
The antenna glowed faintly green.
I logged the event as uncontrolled field emission, sealed the equipment, and placed it in Cold Room 3 alongside previous Marrow Creek artifacts.
Temperature stable. Signal quiet.
I should have left it there.
But the hum followed me up the stairwell.
By the time I reached my office, I could hear it through the vents — a pulse just below hearing, patient, familiar.
I thought of the orchard then.
The rows beneath the water.
And the voice that spoke before I did.
I turned on the recorder again, meaning to document the phenomenon.
Instead, I heard breathing on the line.
Not mine.
“…weather’s changing…”
I left it running.
When I came back hours later, the reels were still turning — though the power light was off.
The hum hadn’t stopped.
It had just learned to wait.
[End of recovered material]
Timestamp correlation: 22:22:46 / Field Source WMC-22 / Finch Archive Integrity Log 1022B