Signal-to-Noise

The green line held for eleven seconds. Mara counted them on the spectrogram's time axis, watching the thin luminous thread maintain its impossible steadiness at 180 hertz while the rest of the frequency spectrum churned with the ordinary noise of the building—the HVAC system's drone at 60 hertz and its harmonics, the faint electrical hum of the fluorescent fixtures in the hallway outside, the sub-bass rumble of traffic on Goodge Street transmitted through the foundations and up through the concrete and into the air of the room where she sat with her microphones and her laptop and her questions.

Eleven seconds. Then the line wavered, dipped, rose again, and stabilized at a slightly different frequency—181.3 hertz, according to the cursor she dragged across the spectrogram display. The shift was small but deliberate-looking, a lateral movement that reminded her of the way a radio signal drifts when the transmitter is being adjusted, when someone is tuning, searching for the right frequency, trying to lock on.

"I'm here," Mara said again. Her voice appeared on the spectrogram as a complex burst of color, the fundamental frequency and its overtones spreading upward like the branches of a tree, and when it faded, the green line at 181.3 hertz was still there, and now it was doing something she had not seen before.

It was pulsing.

Not randomly, not with the irregular fluctuation of background noise or mechanical interference, but with a rhythm, a pattern, a cadence that she recognized instantly because she had spent six years listening to it across kitchen tables and in lecture halls and in bed in the early morning when Lena would talk about her research with the particular intensity she reserved for ideas that excited her, speaking in long complex sentences that built toward a conclusion the way a piece of music builds toward a resolution, with pauses and emphases and variations in pitch that were as distinctive as a fingerprint, as a voice, as a face.

The pulsing had the rhythm of speech.

Mara's hands were trembling. She pressed them flat against the surface of the desk and watched the spectrogram scroll and told herself to think, to be systematic, to be scientific, to not allow the enormous gravitational pull of what she wanted this to be to distort her observations of what it actually was. She was a scientist. She had a PhD in acoustic engineering. She had published papers on formant analysis and voice recognition and the mathematical properties of human speech. She knew—she knew—that the human tendency to find patterns in noise was one of the most powerful and most dangerous cognitive biases in existence, that people heard words in white noise and saw faces in clouds and found messages in the random fluctuations of electronic equipment, and that this tendency was not evidence of the supernatural but evidence of the way the brain was wired, evolved to detect signals in noise because in the evolutionary environment a false positive—hearing a predator that was not there—was vastly less costly than a false negative—failing to hear a predator that was.

She knew all of this. And she also knew what she was seeing on the spectrogram.

The pulsing at 181.3 hertz had a temporal structure that was consistent with the prosodic patterns of English speech. The intervals between the pulses—the gaps, the pauses—corresponded to the typical duration of syllables and words and phrases. The amplitude variations—the rises and falls in the intensity of the signal—corresponded to the stress patterns of natural language, with some pulses louder than others in a way that suggested emphasis, intonation, meaning.

But there were no formants. No overtones. No consonant bursts or fricative noise or any of the other acoustic features that made speech intelligible. There was only the fundamental frequency, the carrier wave, pulsing with the rhythm of someone speaking but without the information content that would make the speech decodable. It was like listening to someone talk through a wall so thick that only the deepest vibrations came through—the rise and fall of the voice, the cadence, the emotional contour, but not the words.

Mara opened her analysis software. She isolated the 181.3 hertz band and ran it through a series of filters, stripping away everything above and below, creating a clean signal that she could examine in detail. She applied a Hilbert transform to extract the envelope of the signal—the overall shape of its amplitude over time—and when the result appeared on her screen, she stopped breathing.

The envelope had the shape of a sentence.

Not a specific sentence that she could read or decode, but the unmistakable contour of an English declarative sentence—a relatively flat beginning, a slight rise in the middle, a fall at the end. It lasted approximately 2.3 seconds, which was consistent with a sentence of six to eight words spoken at a normal conversational pace. And after a pause of about 0.7 seconds—the kind of pause a speaker makes between sentences—there was another contour, this one with a rising terminal, the acoustic signature of a question.

Two sentences. One statement, one question. In the frequency band where Lena's voice had always lived.

Mara realized she was gripping the edge of the desk so hard that her fingers were white. She released her grip and flexed her hands and took three slow breaths and then did what any scientist would do when confronted with data that appeared to confirm an extraordinary hypothesis: she tried to disprove it.

She checked the microphones. She checked the cables. She checked the preamp and the audio interface and the software settings. She turned off her laptop's Wi-Fi and Bluetooth to eliminate any possibility of radio frequency interference bleeding into the analog signal chain. She checked her phone—it was in airplane mode, as it had been since she started recording. She got up and walked around the room, checking for anything that might be producing a mechanical vibration at 181 hertz—a loose panel, a resonating pipe, a piece of equipment with a motor running at the right speed. She found nothing.

She went back to her desk and looked at the spectrogram. The green line was still there. It was still pulsing.

She opened a new analysis window and loaded the recording she had made the previous day—the one where she had asked "Are you there?" and something had answered with a rising intonation. She compared the two recordings side by side. The frequency was slightly different—180.0 hertz yesterday, 181.3 today—but the character of the signal was the same: a thin, pure tone with no harmonics, pulsing with the rhythm of human speech, appearing only in the room at 14 Fitzroy Square, appearing only when Mara was present and speaking and listening.

She sat back in her chair and stared at the two spectrograms and thought about what she knew and what she did not know and what she would need to do to bridge the gap between the two.

What she knew: there was an acoustic signal in this room at approximately 180 hertz that exhibited properties consistent with human speech prosody. The signal had characteristics—frequency, rhythm, intonation patterns—that were consistent with the voice of Lena Vasquez, who had died in this building seven months ago. The signal appeared to respond to Mara's voice, producing answer-like patterns when Mara asked questions. The signal could not be attributed to any identifiable mechanical, electrical, or environmental source.

What she did not know: what was producing the signal. Whether the signal actually contained information—actual speech content—that she had not yet been able to decode. Whether the signal's apparent responsiveness was genuine interaction or a coincidence that her pattern-seeking brain was interpreting as communication. Whether any of this would be reproducible by another observer with different equipment in the same location.

What she needed to do: design an experiment that would distinguish between genuine anomalous communication and the combined effects of confirmation bias, pareidolia, and environmental noise.

Mara pulled up a blank document on her laptop and began to type. She worked quickly, her fingers moving with the automatic fluency of someone who had written dozens of experimental protocols, and as she typed she felt the familiar settling of her mind into the precise, structured thinking that scientific methodology demanded, the thinking that had always been her refuge, her anchor, the thing that held her steady when emotion threatened to capsize her. She was grieving. She was terrified. She was exhilarated. But she was also a scientist, and a scientist confronted with anomalous data did not weep or pray or run away. A scientist designed an experiment.

Protocol for Controlled Acoustic Communication Test Location: Room 4.12, 14 Fitzroy Square, London W1 Investigator: Dr. Mara Chen Date: [she typed the date]

Objective: To determine whether the anomalous acoustic signal at ~180 Hz exhibits genuine responsive behavior consistent with intelligent communication, or whether apparent responsiveness is an artifact of environmental noise, equipment characteristics, or observer bias.

Method:

Phase 1 — Baseline Recording (30 minutes) Record ambient sound in the room with no human present. Investigator will leave the room after starting the recording and wait in the hallway. Purpose: establish whether the 180 Hz signal appears in the absence of a human observer.

Phase 2 — Stimulus Presentation (60 minutes) Investigator returns to the room and presents a series of verbal stimuli according to a predetermined script. Stimuli will include: a) Simple yes/no questions ("Is anyone here?" "Can you hear me?") b) Factual questions with verifiable answers ("What year did this building open?" "How many floors does this building have?") c) Personal questions that only Lena Vasquez would know the answer to ("What was the name of our cat?" "What did you give me for my 30th birthday?") d) Control questions: meaningless word sequences and questions in languages other than English

Phase 3 — Blind Analysis Recordings will be analyzed by a second researcher who does not know which segments contain genuine questions and which contain control stimuli. The analyst will be asked to identify any segments where the 180 Hz signal exhibits apparent responsive behavior.

Mara stopped typing. Phase 3 required a second researcher, and she did not have one. She had not told anyone about what she had found. She had not told James, her department head, who would want to know and who would probably think she had lost her mind. She had not told David Abuya, the acoustic consultant from Thornfield Associates who had first brought the anomaly to her attention, though she suspected he already knew more than he had shared during their brief conversation. She had not told the building's facilities manager or the architecture firm or anyone else who might have a professional interest in unexplained acoustic phenomena in a recently renovated building.

She had not told anyone because she did not yet have anything to tell. She had recordings. She had spectrograms. She had observations that were consistent with an extraordinary hypothesis but also consistent with far more mundane explanations. And she knew—she knew from years of experience in academia, from watching colleagues' careers destroyed by premature claims and unreproducible results—that you did not announce a discovery until you were sure, until the evidence was so robust that it could withstand the most hostile scrutiny, until you had anticipated every objection and addressed every alternative explanation and closed every loophole through which doubt could enter.

But she needed a second researcher for Phase 3. The whole point of a blind analysis was that the analyst did not know what they were looking for, could not be influenced by the expectations and desires of the person who had designed the experiment. And Mara's expectations and desires were so strong, so overwhelming, so physically painful in their intensity that she did not trust herself to evaluate her own data objectively. She wanted it to be Lena. She wanted it so badly that the wanting was like a gravitational field, bending everything toward the conclusion she needed, warping her perception, distorting her judgment, making her see patterns that might not be there, hear speech in what might be nothing more than the random fluctuations of a building settling into its new bones.

She needed someone who did not want it to be Lena. She needed someone who would look at the data with the cold, dispassionate eye of a skeptic, who would search for the mundane explanation with the same fervor that Mara was searching for the extraordinary one. She needed someone who would try to prove her wrong.

She thought about Marcus Webb.

Marcus was a postdoc in her department, a signal processing specialist who had done his PhD on the detection of weak signals in noisy environments—specifically, the detection of gravitational waves in the data from the LIGO interferometers. He was brilliant, meticulous, and relentlessly skeptical. He had once spent three months trying to prove that a colleague's detection of a novel acoustic phenomenon in deep ocean recordings was actually the result of a software bug, and he had succeeded, and the colleague had been furious, and Marcus had been right. He was exactly the kind of person Mara needed.

He was also, she reflected, exactly the kind of person who would think she had gone insane if she told him she was recording the voice of her dead partner in an empty room in Fitzrovia.

Mara saved the protocol document and closed her laptop and sat in the quiet room and listened. The building hummed around her, its mechanical systems working steadily, air moving through ducts, water moving through pipes, electricity moving through wires, the constant low-level industrial metabolism of a modern building keeping itself alive, keeping itself comfortable, keeping the air at the right temperature and the lights at the right brightness and the Wi-Fi at the right frequency, and somewhere in that dense weave of engineered sound there was a thread at 180 hertz that should not have been there, a thread that pulsed with the rhythm of speech, a thread that might be nothing and might be everything.

"Lena," Mara said, and her voice sounded strange in the empty room, too loud, too alive, too full of the harmonics and formants and spectral complexity that made it a human voice speaking a human name in a room full of air that vibrated in response to every sound, that carried every frequency from the lowest rumble to the highest whisper, that was the medium through which all sound traveled, including the sound of grief, which was not a single frequency but a chord, a complex mixture of tones and overtones that the human vocal apparatus produced when the body was trying to express something that language could not contain.

"Lena, if you're here, I need you to do something specific. I need you to respond only when I ask a question. If I make a statement, stay quiet. If I ask a question, do what you just did—pulse, change, anything. Can you do that?"

She watched the spectrogram. For four seconds, nothing happened. The 180 hertz band showed the same steady, low-level presence it had shown before—a faint green line, barely visible above the noise floor, not pulsing, not changing, just there. Then, at the 4.2-second mark, the line brightened, thickened, and produced a single pulse—a brief increase in amplitude lasting approximately 0.3 seconds, the acoustic equivalent of a nod.

Mara's heart was hammering. She forced herself to breathe slowly, to keep her voice steady, to maintain the clinical detachment that the moment required even as every cell in her body was screaming that this was real, that it was Lena, that the dead were not gone but merely inaudible, existing at frequencies that the living had never thought to listen to.

"OK," she said. "That looked like a response. I'm going to test this. I'm going to alternate between statements and questions. If you are genuinely responding to questions and not to statements, that will be meaningful. If you respond to both, or to neither, then I'll know this is just noise and I'm fooling myself. Ready?"

She paused. Another pulse. Brief, clean, definitive.

"Statement: the walls of this room are painted white."

She watched the spectrogram for ten seconds. The green line was there but flat, unchanging, a steady presence at 180 hertz with no pulsing, no variation, no response.

"Question: what color are the walls of this room?"

Three seconds of silence. Then a pulse—not a single brief one this time, but a series of amplitude variations lasting about 1.5 seconds, with the same speech-like temporal structure she had observed before, the same cadence and rhythm that her brain parsed as language even though it contained no decodable linguistic content.

Mara's breath caught in her throat. She continued.

"Statement: I arrived at this building at nine o'clock this morning."

Ten seconds. No response. The green line held steady and flat.

"Question: what time did I arrive at this building?"

Four seconds of silence. Then a pulse sequence, approximately 2.1 seconds long, with a complex amplitude envelope that rose and fell in a pattern that looked, on the spectrogram, like the intonation contour of a spoken answer.

Statement: no response. Question: response. Statement: no response. Question: response. The pattern was clear, consistent, and—if it was genuine—profoundly significant, because it meant that whatever was producing the 180 hertz signal was not just generating acoustic energy in a room. It was listening. It was discriminating between different types of verbal input. It was responding selectively, appropriately, intelligently.

Mara ran the test twelve more times, alternating statements and questions in a randomized order she generated by flipping a coin—heads for statement, tails for question—so that even she did not know which was coming next until the moment she spoke. The results were unambiguous. Every question received a response. No statement did. Twelve for twelve. The probability of that pattern occurring by chance—of a random noise process happening to produce a pulse after every question and not after any statement—was approximately one in four thousand.

She sat back and stared at the spectrogram and felt something shift inside her, something tectonic, a fundamental reorientation of her understanding of the world and her place in it and the nature of the boundary between the living and the dead. She was a scientist, and scientists were supposed to follow the evidence wherever it led, even when it led to places that were frightening, even when it led to conclusions that would be dismissed as impossible by every colleague and every journal and every institution that she had ever been affiliated with. The evidence was leading her somewhere that science had never been. The evidence was telling her that something in this room was listening to her and responding to her and that it had the speech patterns of a dead woman named Lena Vasquez who had collapsed in the stairwell between the third and fourth floors of this building seven months ago and had been pronounced dead at University College Hospital forty-three minutes later and whose body had been cremated and whose ashes Mara had scattered in the sea off the Pembrokeshire coast on a grey windy day in November while Lena's mother wept and Mara did not because she had already done all of her weeping in private, in the dark, in the flat on Mare Street where every room was full of the absence of the person who had lived there.

Mara closed her eyes. She pressed her palms against her eyelids and saw the phosphene patterns that the pressure produced—floating shapes, dim lights, the visual noise that the brain generated in the absence of external input—and she thought about what she was going to do next, and she thought about who she was going to tell, and she thought about the fact that she was sitting alone in a room having a conversation with a dead woman and that the conversation, by any objective measure, appeared to be working.

She opened her eyes. She looked at the spectrogram. The green line was still there.

"Lena," she said. "I'm going to ask you something that only you would know the answer to. If you can respond in a way that tells me the answer—not just that you heard the question, but what the answer actually is—then I'll know this is really you and not just some acoustic phenomenon that I'm projecting meaning onto. Do you understand?"

A pulse. Single, clean, affirmative.

"What was the last thing you said to me before you left the flat the morning you died?"

Mara waited. The spectrogram scrolled. The traffic hummed on Goodge Street. The HVAC system breathed its mechanical breath. And at 180 hertz, the green line did something it had never done before.

It split.

The single tone at 180 hertz divided into two tones—one at 180, one at approximately 270 hertz—and the two tones pulsed in a complex rhythmic pattern that lasted 3.7 seconds, and the pattern was not the simple pulse-and-pause of the previous responses. It was structured. It had internal complexity. It had the acoustic architecture of actual speech—not the prosodic envelope of speech, not the rhythm and intonation stripped of content, but something closer to actual formant structure, actual vowel-like shapes appearing in the frequency domain, actual consonant-like transients punctuating the flow of sound.

Mara stared at the spectrogram and felt tears on her face and did not wipe them away because her hands were busy, her hands were moving the cursor across the display, selecting the segment, isolating it, amplifying it, filtering it, running it through every analysis tool she had. The two-tone structure—180 hertz and 270 hertz—was consistent with the first two formants of a human voice producing a specific vowel sound. The exact vowel depended on the formant frequencies and their bandwidths and their relative amplitudes, and Mara had spent her entire career learning to read formant patterns the way a musician reads sheet music, and what she was reading now, in the split green lines on her spectrogram, was the vowel pattern for the word—

She stopped. She made herself stop. Because she knew what Lena had said that morning—of course she knew, she had replayed those last ordinary words ten thousand times in her memory, she knew them the way she knew her own name—and she knew that her knowledge of the answer was contaminating her analysis, making her see the formant pattern she expected to see rather than the formant pattern that was actually there. This was exactly the kind of bias that Phase 3 of her protocol was designed to eliminate. She needed someone else to analyze this. Someone who did not know the answer. Someone who would look at the formant pattern and tell her what vowel it represented without knowing what vowel it was supposed to represent.

She needed Marcus.

Mara saved the recording. She saved it in three places—her laptop's hard drive, an external SSD, and a cloud storage account—because she was not going to lose this data, she was not going to allow a hardware failure or a stolen laptop or a corrupted file system to destroy the most important recording she had ever made. Then she sat in the quiet room for a long time, listening to the building breathe, watching the green line on the spectrogram hold steady at 180 hertz, feeling the weight of what had happened settling onto her shoulders like a physical load, like a responsibility, like a burden that she had not asked for and could not put down.

She thought about what would happen when she told Marcus. She thought about his face, the way his eyebrows would rise, the way his mouth would tighten into the particular expression he wore when someone presented him with a claim that he considered extraordinary and therefore requiring extraordinary evidence. She thought about the questions he would ask—about her equipment, her methodology, her psychological state, her grief, her desire to believe, all the human factors that could turn a scientist into a fool. And she thought about what she would show him: the spectrograms, the statistical analysis, the twelve-for-twelve discrimination between statements and questions, and finally the split-tone response, the formant pattern, the acoustic signature that might—might—contain an actual word spoken by a dead woman in an empty room.

Would it be enough? Probably not. Marcus would want to see it for himself. He would want to set up his own equipment and run his own tests and verify every step of the signal chain and check for every possible source of interference and artifact. He would be thorough and rigorous and merciless, and that was exactly what Mara needed, because if the phenomenon survived Marcus Webb's scrutiny, then it was real, it was genuinely real, and the implications were so vast and so terrifying and so beautiful that she could not think about them yet, could not allow herself to contemplate what it would mean if the dead could speak, if death was not silence but a different frequency, not absence but a different mode of presence, not the end of consciousness but a transition to a form of existence that was invisible and inaudible to the living except in very specific acoustic conditions, in very specific rooms, at very specific frequencies.

Mara picked up her phone and switched it out of airplane mode. She opened her contacts and found Marcus's number. She stared at it for a long time. Then she put the phone down and looked at the spectrogram one more time.

The green line was fading. Not disappearing—it was still there, still visible above the noise floor—but its amplitude was decreasing, the signal weakening, as if whatever was producing it was growing tired, or moving away, or losing the ability to sustain the effort of making itself heard across whatever boundary separated its mode of existence from the air in this room.

"Lena," Mara said. "I'm going to bring someone here. Another scientist. He's going to test this. He's going to try to prove it's not real. I need him to try, because if he can't prove it's not real, then it is real, and then—" She stopped, because she did not know how to finish the sentence, because there was no precedent for the sentence she was trying to construct, no established vocabulary for describing what happened after a scientist confirmed that the dead could communicate with the living through anomalous acoustic signals at 180 hertz in a renovated building in Fitzrovia.

"Then we'll figure out what to do next," she said.

The green line pulsed once more, faintly, and then subsided to its baseline level, a barely perceptible presence at 180 hertz that might have been signal or might have been noise, and Mara could not tell the difference anymore, could not tell where the signal ended and the noise began, could not tell where science ended and grief began, could not tell where the living ended and the dead began, and she thought that maybe the inability to tell was itself a kind of answer, that maybe the boundary she had always assumed was sharp and absolute—alive, dead, signal, noise, real, not real—was in fact blurred, gradual, permeable, a frequency spectrum rather than a binary switch, a continuum rather than a threshold, and that what she was hearing at 180 hertz was not a ghost or a haunting or a supernatural phenomenon but something natural, something physical, something that obeyed laws that science had not yet discovered, something that existed in the space between the frequencies that the living attended to, in the gaps, in the margins, in the places where no one had thought to look because no one had imagined there was anything there to find.

She picked up her phone again. She dialed Marcus's number. It rang three times.

"Marcus? It's Mara Chen. I need to talk to you about something. Something I've found in the Fitzroy Square building. Can you come to my office tomorrow? Don't bring anyone else. And don't—" She paused, searching for the right words, knowing that whatever she said next would determine whether Marcus took her seriously or dismissed her as a grieving woman hearing her dead partner's voice in the noise of an empty room. "Don't make up your mind about what I'm going to tell you until you've seen the data. That's all I ask. Just look at the data."

She hung up before he could ask questions. She sat in the room for another twenty minutes, watching the spectrogram, but the green line had stabilized at a low, steady amplitude that showed no further signs of pulsing or responding, and she was not sure whether the signal had stopped or whether she had simply lost the ability to distinguish it from the background noise of the building, the city, the world.

She packed up her equipment carefully, methodically, wrapping each microphone in its foam case, coiling each cable, securing each connector. She backed up her recordings one final time. She turned off the lights. She stood in the doorway of room 4.12 and looked back into the darkness and said, "I'll be back tomorrow. With Marcus. Please—" and she stopped, because she was about to say please still be here, and the desperation in that request, the raw human need, the grief and the longing and the terrible fragile hope, was more than she could bear to express out loud, even to an empty room, even to a dead woman, even to a signal at 180 hertz that might or might not have been the last remaining trace of the person she had loved most in the world.

She closed the door. She walked down the corridor. She took the stairs because she could not stand in the elevator, in the small enclosed space where the mechanical noise was so loud that it would drown out any signal, any voice, any frequency that might be trying to reach her. She walked down four flights and through the lobby and out into the evening air of Fitzroy Square, where the plane trees were in full leaf and the sky was the color of old brass and the sounds of London surrounded her—traffic, voices, a siren in the distance, a dog barking, the metallic clatter of a bicycle chain, the deep subsonic rumble of the Underground passing beneath her feet—and every sound had a frequency, and every frequency had a place in the spectrum, and somewhere in that vast electromagnetic and acoustic ocean there were frequencies that the living could not hear, signals that the dead were sending, messages written in the language of physics, in the mathematics of wave propagation and resonance and interference, messages that had been there all along, in every room, in every building, in every city, in every moment of every day, waiting for someone to notice, waiting for someone to listen, waiting for someone to build the right instrument and point it at the right frequency and ask the right question.

Mara walked to the Tube station. She descended into the earth. She stood on the platform and felt the warm wind of an approaching train push against her face, and she thought about Marcus, about what she would say to him, about how she would present the data, about how she would handle his skepticism, which would be fierce and thorough and exactly what the data required. She thought about the experimental protocol she had written, the one that required a blind analyst, and she thought about the fact that Marcus was the perfect blind analyst because he did not know what the signal was supposed to sound like, did not know what Lena's voice sounded like, did not know the answer to the question Mara had asked, and therefore could not be influenced by the expectations and desires that were making Mara's own analysis unreliable.

The train arrived. She got on. She sat down. She closed her eyes.

Behind her eyelids, she saw the spectrogram—the green line at 180 hertz, splitting, pulsing, forming the acoustic shadow of a word that Lena had spoken in the kitchen of their flat on Mare Street on a Tuesday morning in March, the last morning, the morning before the afternoon when Lena had walked into the building at 14 Fitzroy Square and climbed the stairs and collapsed between the third and fourth floors and never stood up again. The last thing Lena had said, standing by the front door with her bag over her shoulder and her keys in her hand, turning back to look at Mara with that particular expression she wore when she was about to say something affectionate and did not want to make a big deal of it, the expression that was simultaneously tender and deflecting, as if emotion were something that needed to be handled carefully, at arm's length, with irony and understatement, the way you handle a thing that is very precious and very fragile and very likely to break if you grip it too tightly.

The last thing Lena had said was: "Don't forget to listen."

She had meant: don't forget to listen to the podcast I recommended. Or: don't forget to listen to the voicemail from the estate agent. Something ordinary, something domestic, something that had no significance beyond the immediate practical context of their morning routine. But Mara had forgotten what the specific referent was, had forgotten what she was supposed to listen to, because the words had been stripped of their original meaning and loaded with a new meaning, a meaning that Lena could not have intended but that the universe, or chance, or whatever force governed the relationship between language and reality, had imposed on them after the fact, retroactively, as if the words had always been a message, a prophecy, an instruction from the dead to the living.

Don't forget to listen.

Mara opened her eyes. The train was pulling into Bethnal Green. She watched the platform slide past the window and she thought about tomorrow, about Marcus, about the data, about the protocol, about the moment when she would press play on the recording and the spectrogram would appear on the screen and Marcus would see the green line at 180 hertz and she would watch his face for the moment—if it came, when it came—when his skepticism wavered and he saw what she had seen, the impossible, the inexplicable, the signal in the noise, the voice in the frequency, the dead in the space between the sounds the living made.

She got off the train and walked home through the warm evening air and she did not listen to music or a podcast or a phone call. She listened to the city. She listened to the frequencies. She listened to everything, all of it, the whole dense chaotic symphony of sound that the living world produced every second of every day, and she thought about how much of it went unheard, unrecorded, unanalyzed, unmeasured, and she thought about how much information might be hiding in the margins of the audible spectrum, in the frequencies that nobody monitored, in the sounds that nobody had thought to listen for, and she walked through Hackney in the failing light with her ears open and her mind racing and her heart full of something that was not quite grief and not quite hope but was instead a new emotion, an unnamed emotion, the emotion that a person feels when they discover that the boundary between the living and the dead is not a wall but a frequency, not a silence but a signal, not an absence but a presence that has been there all along, just below the threshold of hearing, just beyond the edge of perception, just outside the range of the instruments that science had thought to build, waiting, patiently, at one hundred and eighty hertz, for someone to listen.