What the Bureau Heard — Chapter 2

# Chapter 2 — The Frequency Profile

Over the first week of December, Briggs established three things about AR-1906-C.

First: the signal was coordinated. It didn't originate from a single point. The triangulation array showed multiple simultaneous sources, all within the Financial District, all transmitting on the same fundamental frequency, but each with a slightly different harmonic signature. Briggs mapped seven distinct sources — seven points in the underground conduit network where the signal was strong enough to triangulate independently. The sources formed a rough circle, centered on 101 California Street, with a radius of approximately four blocks.

He overlaid the source map onto the city surveyor's records. The seven points corresponded to seven locations where the post-1906 telephone conduit system had junction boxes — maintenance access points where the copper cables branched to serve different buildings. The conduit system had been rebuilt in 1907-1908, during the city's reconstruction. The plans were on file at the Western Pacific Survey Company, because the Survey Company had been involved in the original surveying work. The plans showed standard telephone infrastructure — copper cables in clay conduits, junction boxes every few blocks, switching equipment in the basements of major buildings.

There was nothing unusual about the infrastructure. The signal was coming through standard equipment. But the signal itself was not standard. And the pattern — seven sources in a circle, centered on a bank — suggested design. Not modification. Not tampering. Design.

Something about the way the conduit system had been built was acting as a resonant structure for the signal. The copper cables weren't carrying the signal the way telephone lines carry voice. They were amplifying it. The conduit network was functioning as an antenna array, and it had been doing so for thirteen years, since the day it was installed.

Second: the signal had a schedule. It appeared only between midnight and 4 AM. It appeared only when the fog was heavy — visibility less than a quarter mile. And it appeared only on nights when — and this was the data point that Briggs filed in the locked box, not the daily log — the barometric pressure dropped below a specific threshold that he couldn't correlate to any known atmospheric phenomenon.

The fog correlation was explainable. The marine layer created temperature inversions that could affect electromagnetic propagation. The Signal Corps had documented similar effects in the Philippines — atmospheric ducting, where radio signals traveled further than expected due to trapped layers of air. Briggs could file this as "anomalous propagation effect, marine layer ducting, telephone frequency band."

The barometric correlation was not explainable. The signal appeared when the pressure dropped below 29.8 inches of mercury, but only at night, only with fog. The pressure threshold was too specific for a random atmospheric effect. It suggested that the signal's appearance was triggered by an environmental condition — or that the signal's source was sensitive to atmospheric pressure.

Or — the thought went in the locked box — that the signal's source was sensitive to the boundary between the physical world and something else. Fog as a medium. Pressure as a threshold. The signal crossing from one side to the other.

Third: the signal was learning.

This was the observation that Briggs was least comfortable with, because it implied agency, and agency implied intelligence, and intelligence implied something that was using the telephone network as a communication system, which was — within the Technocratic paradigm — impossible. The telephone network carried voice. It didn't carry ghosts. It didn't carry spirits. It didn't carry dead women speaking through copper wire.

But the signal was changing. The first reading on November 28 had been simple — a breathing rhythm, a single frequency, no harmonic content beyond the fundamental. By December 3, the signal had developed overtones. By December 5, it had developed a call-and-response structure. By December 7, the audio output contained what Briggs could only describe as phonemes — the building blocks of speech, not yet assembled into words, but present, forming, as if something was learning to speak through a medium that was not designed for speech.

Briggs played the recordings back at half speed. At quarter speed. At one-eighth speed, where the breathing rhythm became a slow pulse and the phonemes stretched into long, wavering tones that sounded like — he checked the logbook twice before writing it down — like someone singing. Very quietly. Very slowly. In a language he didn't recognize.

---

The triangulation array produced its most significant result on December 8.

The signal appeared at 12:03 AM — earlier than any previous reading. The needle on Channel Three jumped to 1.4 millivolts — the highest Briggs had recorded. The breathing rhythm was there, but faster, more urgent, and the harmonic content was dense with what Briggs had classified as phonemes. The signal was active on all seven source points simultaneously. For the first time, the sources were synchronized — the same frequency, the same phase, the same modulation, as if the seven points in the circle were all saying the same thing at the same time.

Briggs checked Channel One. The Montgomery Street exchange was showing unusual activity — not on the standard telephone traffic, but on a secondary frequency band that Briggs had added to Channel One's monitoring range specifically because he'd detected anomalous traffic there during earlier AR-1906-C events. The secondary band showed what looked like a coordinated communication session — multiple participants, structured turn-taking, a pattern that matched what the Signal Corps would classify as a "conference call" except that the participants weren't using standard telephone equipment.

He recorded everything. The Dictaphone cylinder was spinning at full speed, the needle scratching the signal into the wax with a fidelity that his earlier recordings hadn't captured. He'd upgraded the recording apparatus three days ago — a new amplifier, a better stylus, a cylinder made of harder wax that could hold finer detail. The upgrade had been justified as "equipment maintenance" in the supply requisition. Gillis had signed it without asking what "finer stylus resolution" meant.

The signal lasted 47 minutes. At its peak — 12:31 AM, barometric pressure 29.6, fog visibility approximately 100 feet — the audio output contained something new. Not phonemes. Not breathing.

Words.

Briggs heard them through the headphones, sitting at his workbench at 12:31 AM on December 8, 1920, with Edison's ears rotated toward Channel Three and the fog pressing against the window like something trying to get in.

The words were in English. The consonants were wrong — too sharp, too hard, as if the speaker was forming words through a medium that didn't allow for the soft fricatives that English required. "Through" became "tr-rrough." "Enough" became "en-nugh." The vowels were long and hollow, like sound bouncing inside a pipe.

He couldn't make out a complete sentence. The signal wasn't strong enough, or the recording apparatus wasn't sensitive enough, or the speaker wasn't close enough. But he heard fragments:

"...the grid... positions... six declining..."

"...bound material... throughput..."

"...harmonic calibration..."

And then, at 12:47, the signal stopped. Not gradually — abruptly, as if a switch had been thrown. The needle dropped from 1.4 to 0.1 in less than a second. The silence was so sudden that Briggs's ears rang with the absence.

He played the recording back. Twice. Three times. The words were there — faint, distorted, buried in the harmonic content, but unmistakably words. English words. Technical vocabulary. "Positions." "Throughput." "Calibration."

Someone — something — was using the telephone network to conduct a meeting. And the meeting was about something with positions and throughput and calibration. Something that had declining elements and bound material. Something that operated on a schedule and used the fog as cover.

Briggs wrote the report at 3 AM, sitting at the workbench with Edison asleep on his shoulder and the fog beginning to thin outside the window. The report was seven pages long, with frequency analysis, triangulation data, barometric correlations, and a phonetic transcription of the words he'd recovered from the recording.

He read it through once. Then he locked it in the box beneath the workbench. The box now contained 40 reports. The daily log at the safehouse would get a two-line entry: "AR-1906-C continued monitoring. Signal strength increasing. No actionable intelligence."

The words "positions," "throughput," and "calibration" sat in the locked box with the other 40 reports, and Briggs sat at his workbench and stared at Channel Three's needle, resting at baseline, and thought about the phonemes that had become words and the words that had become a vocabulary and the vocabulary that described a system he couldn't see.

Edison shifted on his shoulder. The cat's ears twitched — a small, involuntary rotation toward Channel Three's rack, even in sleep. Even now, with the signal gone, the cat was listening.

Briggs envied him. Edison didn't need a paradigm to explain what he heard. He just heard it, and if it was interesting, he paid attention, and if it wasn't, he slept. The cat's epistemology was simpler than the Technocratic Union's, and — Briggs was beginning to suspect — more accurate.

He closed the locked box, checked the sensor calibration one final time, and went to bed.

The forecast for tomorrow called for heavy fog.

---

*End of Chapter 2*