Institutions
The Hand Crank at Spurn
The clock on the instrument panel read 22:47 when I held my pass to the reader. It beeped once, the door clicked, and that was my shift started. Same as always. I had been doing it for four years. Spurn Head Coastal Monitoring Station, Spurn Peninsula, overnight rotation, 23:00 to 07:00.
The clock on the instrument panel read 22:47 when I held my pass to the reader. It beeped once, the door clicked, and that was my shift started. Same as always. I had been doing it for four years.
Spurn Head Coastal Monitoring Station, Spurn Peninsula, overnight rotation, 23:00 to 07:00. You would think that after four years the routine would feel automatic, and mostly it did. But I still ran the checklist the same way every time. Top to bottom. No skipping.
No, I'll get it on the way back. That is how things get missed. I had seen technicians skip steps and spend the rest of a shift chasing a number that turned out to be a loose connector they had walked past an hour earlier. It was not worth it.
So I started in the instrument room. Same as always. The pressure sensor was reading about 1,010 hPa, normal for November on the Holderness coast. The wind gauge was working, showing a north-easterly at around 11 knots.
The humidity logger had a sticky display. It had been doing that for two weeks. I had already raised a maintenance ticket, but the underlying data was transmitting properly. I tapped it twice and the number unstuck.
I wrote it down anyway. The ceilometer showed broken cloud near 670 metres, with another layer near 1,220 metres, matching the regional forecast I had checked before leaving my flat. That was a good sign. When the observed and forecast cloud bases are far apart, it is often the first indication that something is genuinely developing offshore.
The radio uplink was green. The satellite link was green. The Met Office data relay was green.
I initialled each line in the equipment log and moved on. The main instrument room takes perhaps fifteen minutes if you are thorough. After that, the checklist takes you downstairs to the antenna array and lower-deck systems, then back up and outside to the catwalks. I always did the catwalks last because they were cold, and I saw no point in being cold for longer than necessary.
It was November. The wind coming across the Humber estuary had a particular bite to it that I can only describe as personal, as though it were specifically annoyed with you. The lower deck was fine. The antenna array was fine. The tidal-services access hatch beneath the lower deck was sealed, and its indicator showed green. I noted that and moved on.
I did not think about the hatch again for a while.
The south catwalk was fine. The anemometer mount was solid. There was no corrosion on the fittings, and the casing of the wind-direction sensor was intact. I did find a loose bolt on the railing bracket near the southern end. I tightened it with the 10 mm ring spanner I keep in my coat pocket.
I wrote it down anyway, so maintenance would have a record. That is the job.
Then I went to the north catwalk. The Secomak Type 447 electromechanical coastal warning siren was fixed to the north railing, which I had always thought an odd place for it. The estuary lay to the west of the peninsula, whilst that catwalk overlooked the exposed beach and the North Sea. The mounting put the unit in an awkward position beside the railing, but its horn pointed along the seaward beach and towards the adjacent waters where it was meant to carry.
I had not installed it, and I was not going to move it. So that was where it lived.
It was a large unit, a grey housing roughly the size of a cabin bag mounted sideways, with the horn projecting over the railing and the motor-drive housing at the back. The manual crank folded flat against one side when the motor was operating. You would use the crank only if the motor failed.
I ran the standard test, entered the activation sequence at the panel and waited for the drive to spin up.
Nothing.
I tried again. The panel acknowledged the command, but the motor did not respond. I could hear the relay clicking, which meant the signal was reaching the unit, but the drive itself was not turning.
I removed the housing cover and looked inside. I probably should not have done that without a maintenance technician present, but I only wanted to confirm what I already suspected.
The drive was dead. The brushes were worn almost to nothing, and one lead had pulled free of its terminal. It was not a quick repair.
I entered it in the log.
Secomak Type 447, north catwalk. Motor-drive failure. Inoperable under powered activation. Flagged for maintenance. Unit functional by manual crank only.
I initialled the entry and moved on. It was a minor note. Equipment fails at coastal installations all the time. Salt air eats everything eventually. I assumed maintenance would deal with it within a day or two.
I did not think about it again.
Graham Pritchard arrived at 23:24. I know the exact time because I was updating the hourly log when the door beeped and I glanced at the panel clock out of habit. He had been due at 22:45, which made him thirty-nine minutes late.
I wrote that down as well.
He came in with his coat half unzipped and his keys still in his hand, which meant he had been hurrying somewhere and had ceased to hurry by the time he reached the door. He smelt of blended Scotch whisky. Not overwhelmingly, but it was there, the sort of smell that had been settling into a coat for a couple of hours.
"All right," he said.
"All right."
He hung his coat on the hook beside the door and dropped into his chair. He did not explain the time or offer an excuse. Graham Pritchard had been working in coastal monitoring since before I finished school, and he operated on the assumption that his presence was sufficient. Timing was optional.
I want to be clear that I liked Graham. Liked. He was blunt and occasionally insufferable, and he had a habit of explaining things I already knew, but he genuinely understood the work, particularly the buoy systems. Twenty-two years of watching marine data will do that. At fifty-four, he had seen more instrument failures than most of our engineers.
I trusted his judgement of equipment in a way I did not trust most people's.
I looked at the log entry. 23:24. Thirty-nine minutes late. I considered reporting it.
I had the Coastguard duty officer's number in my phone. I knew the procedure and what the form looked like. One call, one report, and it would go into his file.
He had been written up before, I was fairly sure, although I did not know the details. Another report might actually have consequences. But it was one bad night, and twenty-two years was twenty-two years.
I closed the log and did not pick up my phone.
It felt like mercy at the time.
I have thought about it since.
Graham brought up the Outer Dowsing data without being asked, which was the one part of the job about which he was always prompt. Within three minutes of sitting down, he had opened the station list and begun scrolling through the latest readings. I was still finishing my catwalk notes.
He went quiet. I noticed, but did not look up immediately.
"Mercer."
"What?"
"Come and look at this."
I crossed to his console. He had monitoring buoy OD-17 on the main display. It lay about twelve nautical miles east-north-east of the station, covering the Outer Dowsing shoal and adjacent North Sea waters. The shoal and its surrounding banks had troubled shipping for centuries. The buoy mattered.
The data on the screen was wrong.
Not corrupted or missing. Wrong in a particular, unsettling way. The last four readings were identical.
Not close. Not within rounding tolerances. Identical.
The same wave height. The same water temperature. The same pressure. The same wind speed, down to the decimal place.
The timestamps were 21:15, 21:30, 21:45 and 22:00. Four readings, the same figures repeated.
"How long has it been doing that?" I asked.
"Since 21:15. Four readings. It might have started earlier, but that is as far back as this display goes without pulling the archive."
I studied the rows. "Sensor failure."
Graham shook his head. "Sensors fail. They go quiet or send rubbish. Random values, nulls, error codes. They don't lock."
He tapped the screen.
"This is locked. Something is holding these values and transmitting them on schedule as if everything is normal."
"So the transmitter is working, but the sensors aren't updating."
"The transmitter is working. Whether the sensors are updating is another question. Something upstream of transmission is stuck."
I stood looking at the four identical rows. He was right that it was strange. I had seen buoy malfunctions before. We all had. They usually looked like noise: random spikes, missing packets, a sensor dropping to zero and remaining there.
This did not look like noise.
It looked deliberate. That was the wrong word for a piece of equipment, but it was the word that came to mind.
"I'm calling Bridlington," I said.
Graham shrugged. That was his way of saying he thought it unnecessary but would not stop me. I was already reaching for the telephone.
The HM Coastguard Operations Centre in Bridlington had a direct line for reports from the joint Met Office and MCA monitoring sites. I had called it perhaps a dozen times in four years: equipment flags, data anomalies, the occasional bird strike that knocked a sensor out of alignment. It was not normally a dramatic call.
You gave your station identifier, described the fault, and they logged it. They either escalated it or they did not. Usually they did not. Usually you received an incident reference and a promise that the day shift would follow up. That was the end of it.
I dialled from memory. It rang four times before somebody answered.
"HM Coastguard Operations Centre. Martin Shaw speaking."
His voice was flat. Not unfriendly, merely the voice of a man who had answered the same telephone in the same manner a thousand times.
"Daniel Mercer, Spurn Head Coastal Monitoring Station. Station identifier Spurn Echo Four. I'm reporting a data anomaly from monitoring buoy OD-17 at Outer Dowsing."
I heard him typing, slow and steady.
"Nature of anomaly?"
"Locked transmission loop. Four identical readings at fifteen-minute intervals beginning at 21:15. The transmitter is functioning, but the data have not changed for more than two hours. It doesn't match any failure pattern I've seen. No noise and no nulls, just the same values repeating."
More typing. Then a pause.
"Received. I'm assigning incident reference November 318 Kilo. A technician will review the archived data during the day shift."
"Understood."
I wrote the reference in the corner of the log sheet.
"Is there any previous record of OD-17 showing anomalous behaviour? Anything in the system?"
The pause that followed was longer than the typing pauses.
"I'm not seeing anything in the active file," Shaw said.
That was not quite the same as no. I noticed.
"All right. I'll continue monitoring and log any changes."
I was about to thank him and end the call. The entire exchange had taken perhaps ninety seconds, and I was already thinking about the next item in the checking rotation.
Then he said, "Mercer."
"Yes?"
"Do not open the lower-deck hatch tonight."
For a moment I said nothing.
The tidal-services access hatch beneath the lower deck. Down the internal stairs and past the antenna junction box. I had opened it perhaps twice in four years. It was not something you opened without a reason.
"Can you explain-"
The line went dead.
It was not interference or a call breaking up. It simply ended in the middle of my question, as though he had replaced the receiver the instant he finished speaking.
I stood with the handset against my ear for a second before accepting that nobody was there. Then I rang back.
Four rings, followed by a recorded message.
"You have reached the HM Coastguard Operations Centre in Bridlington. For out-of-hours administrative enquiries-"
I hung up and tried again.
The same recording.
That was all. An incident reference, an unsolicited warning and then nothing.
I set down the telephone and looked at what I had written on the log sheet.
November 318 Kilo.
I copied it into the main log beneath the OD-17 entry, with the timestamp and Martin Shaw's name. Proper documentation. That is me. The station could be on fire and I would still complete the correct fields.
The thing was, duty officers did not volunteer information. Those calls did not work that way. You reported, they logged and they assigned a reference. If you needed anything else, you asked a specific question and they either answered it or said it lay outside their remit.
They did not simply add things. They especially did not say, "Do not open the lower-deck hatch tonight," without context, then hang up before you could ask a single question.
I decided that Shaw either had the wrong station, had seen something in the file he was not meant to disclose, or was having a very strange night. I could not think of a fourth explanation that made more sense.
Graham was leaning against the far console with his arms folded, watching me.
"What did they say?"
"Day shift will review the buoy archive. Incident reference November 318 Kilo." I paused. "And the duty officer told me not to open the lower-deck hatch tonight."
Graham stared at me. "He just said that?"
"Unprompted. Then the line went dead."
He was quiet for a moment, which I appreciated. Then he unfolded his arms.
"Duty officers say odd things sometimes. Probably a misunderstanding. Wrong station, or he was looking at the wrong screen. Did you get a name?"
"Martin Shaw."
He shrugged. "Don't know him. Could be new."
He pushed away from the console.
"I'm going down to check the antenna array. The floodlights on the south side were flickering when I arrived. Might be a relay."
"I'll monitor from here."
He took his radio from the charger and clipped it to his coat. He said nothing else, just went through the internal door and descended the stairs. Then I was alone on the upper deck.
The station has a particular kind of quiet at night. The instruments hum. Wind moves around the exterior cladding. You can hear the surf when it is heavy enough, which it usually is, a low constant beneath everything else.
It is not silence, but a texture of sound I know well enough to stop hearing after the first hour.
That night it seemed quieter. I noticed without being able to explain why. The instruments were still humming and the wind was still there, but something in the mixture was wrong. Or something was missing, and I could not identify it.
I looked at the incident reference.
November 318 Kilo.
I thought about calling Bridlington a third time. I did not. There was nothing more to tell them, and if anyone answered, it was unlikely to be Martin Shaw.
I reopened the OD-17 display. The buoy was still transmitting, still cycling the same four readings on schedule.
I watched it for a minute, then brought up the adjacent buoys for comparison. They were normal: random variation, sensors updating, the familiar texture of live data. OD-17 sat in the middle of them like a held breath.
The console radio crackled.
"Mercer."
I picked it up. "Go ahead."
"The exterior floodlights are out." A pause. "All of them."
I checked the lighting-control panel. Every status indicator showed green, meaning the system believed the lamps were on.
"The panel reads nominal up here."
"They aren't on. I'm standing on the lower deck looking towards the beach, and it's dark."
I turned to the camera controls. An ageing Watec low-light coastal camera was mounted on the eastern exterior, principally for cloud and visibility records, but it also covered the approach across the North Sea beach. I brought its feed up on the secondary monitor.
The picture was dark, though not black. There was enough ambient light from the overcast sky and distant reflections off the water to make out the beach beneath the station: the pylons, the surf line and the figures.
They were pale and roughly human in shape and proportion.
There were twenty-seven of them.
They lay in evenly spaced rows across the beach, all oriented towards the station pylons, prone and flat against the sand.
Graham's voice came through the radio again, quieter now.
"Can you see them on the camera?"
"Yes."
"What are they?"
"I don't know."
I kept watching. None of the figures moved. They simply lay in their rows, facing the pylons.
Then all twenty-seven stood up.
Not one after another. No ripple passed through the formation. Every one of them rose from prone to standing in the same instant and with the same motion, as though a single signal had gone through them all.
They stood motionless, facing the station.
The camera continued to show me that. I said nothing into the radio. Graham said nothing either.
I kept my eyes on the feed. The camera's low-light mode washed the picture out, but there was enough detail to see that their pallor was not merely an artefact. It was the colour of something that had remained underwater for a long time.
They were roughly human-shaped and human-proportioned, but their stance was wrong. Not dramatically wrong. The distribution of weight was simply a little off, as though they had learnt how to stand from a written description.
I engaged the manual digital zoom. It was poor, blocky and pixelated, but I brought them close enough.
Every face was identical.
Each had the same slack jaw, hanging slightly open. The same milky eyes, not white but the colour of old plastic left in sunlight. Dense, irregular grey barnacles ran from chin to jaw, over the cheeks and up to the hairline. They did not look scattered or newly attached. They looked as though they had grown there over time and been allowed to continue.
I sat with the fact that this was something I was seeing.
Then I pulled the image back. One face at that distance had been enough.
It was not merely that the faces resembled one another. It was the same face repeated twenty-seven times: the same slack mouth, the same clouded eyes and the same crust of barnacles in precisely the same pattern.
That disturbed me more than I expected.
Graham still had not spoken. I was about to call him when something in the second row, just left of centre, caught my attention.
One figure wore something on its chest.
A lanyard.
I zoomed in again. The image remained blocky, but I could distinguish an orange MCA contractor lanyard and a rectangular Met Office-MCA site pass hanging from it. There was a logo at the top and an identity photograph below.
I stared at the screen for perhaps ten seconds, trying to decide whether I was reading it correctly. The resolution was too poor to make out the photographed face clearly, but I recognised the general features. I had seen that identity card before.
At first I could not remember where. It belonged to the sort of peripheral memory formed by glancing at the same thing often enough without consciously studying it.
Then I had it.
The contractor on the overnight maintenance rotation. His pass had been clipped to the handover board in the store cupboard, as contractors often left them between site visits. I had glanced at it a dozen times.
I needed to know who he was and where he ought to be.
The store cupboard stood at the back of the main operations room, beyond the instrument rack and secondary communications panel. I had been inside it a hundred times. A laminated contact sheet fixed to the inside of the door listed contractor names, pass numbers, emergency contacts and scheduled rotation dates. If the card belonged to the overnight contractor, the details would be there.
I crossed the room more quickly than I usually moved. I was not running. I simply was not taking my time.
I gripped the handle and pulled.
The door did not open.
I pulled harder, thinking the frame had swollen in the coastal damp. That happened.
But it was not stuck. It was locked.
Deadbolted.
The door had a single-cylinder deadbolt, with a thumb turn inside and a keyhole outside. It could be secured normally only by someone within the cupboard.
I tried the handle again. The bolt held.
That should not have been possible if the cupboard was empty.
An external override key hung in the key cabinet beside the communications panel, intended for exactly the sort of incident in which somebody accidentally locked themselves inside. It was not intended for whatever this was.
The key was on its hook. I took it, returned to the cupboard and opened the deadbolt.
The smell reached me first.
Salt water and something for which I have no clean word. Not rot, exactly. Closer to low tide, but concentrated.
The cupboard was about 2.5 metres deep, with metal shelving along both sides. Backup batteries occupied the racks to the left; spare equipment and supply boxes lined the right. The overhead fluorescent tube was off. Light from the operations room reached far enough inside.
Peter Wainwright was wedged behind the battery rack on the left, sitting with his back against the wall and his knees raised. He wore standard contractor's work clothes. His head rested at a slight backward angle.
Both ears had bled. The blood had dried dark down either side of his neck. It had come from within.
Both eardrums had ruptured.
I did not approach immediately. I stood in the doorway and looked at him.
His face was still. Not slack like the faces on the beach. Simply still. Whatever had happened, it had happened some time ago.
Then I noticed the shelf above him.
Paint had been worn away from a patch about ten centimetres wide. Not random damage. The same area had been scored repeatedly, through the coating to bare metal and then into the metal itself.
I stepped inside far enough to see clearly.
OPEN.
The word had been scratched into the shelf dozens of times, one inscription laid over another until the earlier letters became texture beneath the later ones. The marks were small and uneven. They looked as though they had been made with a fingernail over a long period.
I stood looking at them.
Peter Wainwright had locked himself inside. It was the only explanation that fitted. He had entered the cupboard, turned the deadbolt from within, wedged himself behind the batteries and stayed there. At some point he had begun scratching that word into the shelf above his head and had continued until he could not.
OPEN.
It might have meant the lower-deck hatch. It might have referred to some means by which the figures entered the station. There was no way to tell.
The things on the beach had been connected to this place before my shift began, before I logged on for duty at 22:47, before Graham arrived smelling of whisky. Before anything that had happened that night.
Peter had known. He had locked himself in the one room with a deadbolt operable from inside, then written OPEN into the metal until his fingers gave out.
I looked above him again and saw the notebook.
It was thin, perhaps sixty pages, spiral-bound with a black cardboard cover. The sort of cheap two-pound reporter's notebook sold by any stationer's. It lay flat on the shelf with its spine out, as though he had placed it there carefully rather than thrown or dropped it.
I reached past the scratched word and picked it up.
PETER WAINWRIGHT - FIELD NOTES was written across the cover in marker. Neat handwriting. The work of somebody who labelled things.
I opened it.
The first pages contained ordinary station records: equipment checks, calibration notes and buoy signal-quality assessments. The entries began in November 2025, continued through December and into 2026, all in the same pen and the same careful hand.
I nearly skipped ahead. Then I saw the notes in the margins.
They began in January, written sideways beside a routine calibration entry as though added afterwards when there was no proper space left. The handwriting remained neat but had become smaller and more deliberate.
"9 January 2026, 02:14. Buoy OD-17 camera capture. Four subjects prone approximately 30 metres east of pylon base. Barnacle growth from chin to hairline. Did not surface. Retreated before 03:00."
That was all. No disbelief. No question about what he had seen. Merely a log entry.
Four subjects. Barnacle growth. Retreated before 03:00.
The way somebody else might record that the wind had shifted north-east or pressure had fallen by two hectopascals.
I kept reading.
February contained six entries. March contained nine. The format did not vary: full date, time, buoy designation, number of subjects, behaviour, duration and outcome.
He had been documenting them since January and fitting the observations into the margins of his legitimate records. Perhaps he had nowhere else to put them. Perhaps he did not want them in the official log until he had authorisation.
Authorisation. That word again.
I turned the pages faster. His observations became more detailed as the months passed. He recorded their positions, whether they were prone or standing, how they moved and in which direction they retreated. He noted which buoy cameras gave the clearest angle.
He cross-referenced multiple buoys when more than one recorded a sighting on the same night. By April he was logging responses. What made them stop? What made them move?
That was where the frequency table began.
It occupied a page of its own, copied by hand on to graph paper that had been folded into the binding. At some point he had unfolded it and taped it flat. The tape had yellowed. He had been consulting it for some time.
The table had three columns: frequency in hertz, duration of exposure, observed response.
Whether he had performed the tests or merely documented chance exposures, I could not initially tell. Some entries were marked INCIDENTAL, meaning the response had been captured on camera without his causing it.
A ship's horn at roughly 200 Hz. A fog signal in the low 300s. A burst of marine radio interference measured at 340 Hz.
The response column read: Partial agitation. No response. Mild agitation, brief.
Then, roughly two-thirds down the page, there was a line in darker ink. He had traced over it more than once.
"360 Hz sustained. Secomak Type 447. Full convulsion and retreat. All subjects, immediate onset. Repeated - same result. Repeated again - same result."
I read it three times.
A sustained 360 Hz tone. The Secomak Type 447.
The siren was fixed to the north catwalk railing. I had logged it at the beginning of the shift. Dead motor drive. Manual crank only. I had flagged it for maintenance and moved on.
I stood in the cupboard holding the notebook. There was nobody to speak to. Graham was no longer answering his radio. Peter Wainwright sat behind the battery rack with ruptured eardrums. There was only the smell of salt and copper, the hum of the backup batteries and me.
The siren's motor was dead. The crank was the only way to drive it. I had written that in the log and moved on.
I continued reading.
The final ten pages were difficult, although the handwriting never changed. It remained neat and careful to the end. What changed was the substance.
Peter had documented everything. He had identified the frequency. He had established a countermeasure. He had eleven months of evidence that it worked every time, without exception.
The final dated entries concerned authorisation.
"15 May 2026. Submitted incident report to station supervisor. Awaiting response."
"3 June 2026. Followed up report of 15 May 2026. No response on record. Resubmitted."
"29 June 2026. Escalated to regional office. Advised to route through standard maintenance and anomaly-reporting channels."
July, August and September contained further submissions, follow-ups and referrals to different offices. He had contacted the HM Coastguard Operations Centre. He had tried the regional anomaly desk and three separate supervisors.
Every entry used the same factual, patient, procedural tone. There was no frustration in the writing, only the date, the action taken and the response received.
The final dated entry read:
"14 October 2026. No authorisation received to date. Continuing to document. Will not act without authorisation."
That was the last dated record.
Two pages of shorter, undated notes followed. The handwriting was still his, but the sentences had become abrupt, as though he had been writing quickly or with cold hands.
He recorded two further encounters. He noted that figures had appeared on the station's east-facing camera, not only on buoy feeds.
"Interior proximity," he had written.
He recorded locking himself inside the store cupboard. He recorded that the Secomak Type 447 manual crank was operational and accessible from the north catwalk.
The last line on the final page read:
"Awaiting authorisation to proceed."
I closed the notebook.
Peter Wainwright had possessed the answer for eleven months. He had documented it patiently and completely. He had tried to give it to people, a great many people, and nobody had taken it.
Then he had locked himself inside this cupboard and waited because he would not act without authorisation.
The authorisation had never come.
I looked down at him. He did not look peaceful. He looked like a man who had been behind a battery rack for a long time.
The siren. The north catwalk. The manual crank I had noted during my opening checks as a minor inconvenience, a maintenance flag, something for the next technician to address.
360 Hz sustained. Full convulsion and retreat. All subjects, immediate onset.
I felt that I knew why the motor drive had failed, although the notebook did not say. Perhaps too many sustained cycles had burnt it out. Perhaps something had reached it before Peter could use it again.
It did not matter. The crank was there. The crank worked. I had put my hand on it during the inspection and confirmed it myself.
Peter had known what to do, but he had waited for somebody to tell him he was permitted to do it. He had filed reports. He had followed up. He had escalated through the proper channels in the proper order. Then he had written "Awaiting authorisation to proceed" on the last page of his notebook and died behind a battery rack with OPEN scratched into the shelf above his head.
I thought of Martin Shaw at the Coastguard Operations Centre and the incident reference he had given me.
Do not open the lower-deck hatch.
The line cutting off. Two return calls reaching only an out-of-hours recording.
I thought of Graham not answering his radio.
I stood in that cupboard with the notebook in my hand, the batteries humming and Peter on the floor, and considered what authorisation meant after 02:00 at an isolated station on Spurn Peninsula, with twenty-seven figures on the beach and, since shortly after 01:00, intermittent pressure sounding from beneath the lower-deck grating near the tidal-services hatch.
Peter had believed in the chain of command. I understood that. I was not mocking him for it.
I had made my own choice earlier when I decided not to report Graham for arriving late and smelling of whisky. But that had been me protecting a colleague. This was different.
Peter had protected a process. He had trusted that if he did everything correctly, somebody above him would eventually examine eleven months of evidence and say, "Yes. Proceed. Use the siren."
Nobody had said yes.
I opened the notebook again at the frequency table and found the darkened line.
"360 Hz sustained. Secomak Type 447. Full convulsion and retreat. All subjects, immediate onset."
For another minute I remained where I was. The backup batteries gave off their familiar low, steady hum. Peter sat on the floor, and I held the answer in my hands, written more heavily than everything around it.
The siren with the dead motor. The one I had marked MANUAL CRANK ONLY at 22:47 and then stopped thinking about.
I closed the notebook, tucked it beneath my arm and stepped back into the operations room. I pulled the store-cupboard door closed behind me, but I did not lock it. It felt wrong to lock Peter inside again. I know that makes no sense, but that was where my mind had reached.
The camera feed remained on the monitor. All twenty-seven figures were still on the beach, arranged in the same rows, facing the station with the same slack faces turned towards the pylons.
Their positions seemed fractionally different from the last time I had looked. That might have been compression artefacts from the ageing camera. I could not be certain. But I counted carefully.
Twenty-seven.
I raised the radio and called Graham on the standard channel.
"Graham, come in. Mercer here."
Nothing.
Static, and the hiss of the wind off the North Sea.
I tried again. "Graham, radio check. Come in."
Still nothing. I set the radio down on the console and looked at the clock. It was just past 01:00. The sound from the hatch began perhaps ten minutes later.
It came from the lower deck. Wet and sustained, like something worrying at a seal from the other side. Not a knock or a bang. More like pressure, applied slowly, rhythmically, patiently.
The tidal-services access hatch, which I had never had any reason to think about before that night, was apparently directly beneath the main floor grating, because at first I felt it more than heard it. A vibration through the soles of my boots. Then it grew louder.
I tried Graham again.
Nothing.
I brought up the lower-deck camera, but the exterior floodlights were still out and the interior angle was poor. I could see the base of the antenna housing, part of the equipment rack and a strip of floor near the tidal-services hatch. That was all.
The hatch appeared to be closed.
Whatever was working at it was doing so from underneath.
I kept the radio in my hand and tried every few minutes.
"Graham, come in."
"Pritchard, radio check. Graham, if you can hear this, key twice."
Nothing.
At some point I realised I was standing on the floor grating. I took one step to the left and stood on solid decking instead.
That felt slightly better.
Then I heard him.
His voice was coming up through the grating, muffled by the deck but recognisable. That flat, faintly slurred cadence I had been listening to for four years. He was saying something, one word over and over, but I could not make it out.
I pressed the sole of my boot against the edge of the grating and leant closer. I still could not understand it. I could hear only the shape of the word, one syllable, perhaps two, repeated at the same interval and growing quieter each time, as though he were moving away, or as though whatever was producing the sound were running down.
I tried the radio again.
He did not answer.
I stood there and listened as the word I could not hear became smaller, until I could no longer hear it at all. The station went quiet except for the creaking from the tidal-services hatch, the wind and the hum of the batteries.
I sat on the floor with my back against the console and Peter Wainwright's notebook in my lap, and I waited. I do not know what I was waiting for.
Perhaps I thought Graham would come back on the radio. Perhaps I thought I would hear him on the stairs. I am not sure I was thinking in a straight line by then.
I read the frequency-table entry again.
Then I read the three entries before it. Different dates, same outcome.
360 Hz. Sustained tone. Full retreat.
Peter had written IMMEDIATE ONSET and underlined it once.
The creaking from below stopped at about 02:00.
The station was very quiet after that. Just the wind, the instruments and the low hum of everything electrical doing what it was supposed to do.
I checked the beach camera. The 27 figures had not moved.
The same rows. The same identical faces.
I looked at the clock.
02:14.
I heard the stairwell door.
It was not loud. In fact, it was very quiet. The door at the foot of the interior stairs had a particular sound: a soft catch, followed by a creak on the second step which I had been meaning to log for a month.
I heard the catch.
I heard the creak.
I stood up.
Graham came into view at the foot of the stairwell.
He was upright, his weight evenly distributed, both feet on the floor. He looked like Graham. The same jacket, the same build, the same thinning hair. He was looking towards me.
His jaw was slack. Not relaxed. Slack, as though the hinge no longer worked properly.
Starting at his chin and spreading upwards towards his cheekbones was a texture I recognised from the camera feed. It was still in its early stages, still spreading, but it was there.
I said nothing. I do not know what I could have said.
He began walking towards the stairs. Slowly. Not unsteadily, just slowly.
His mouth was moving.
I could see it from where I stood. I watched as intently as I could, but I could not read what it was forming. Perhaps it was a word. Perhaps it was nothing. Perhaps it was the same word I had heard through the grating, still running down.
I do not know.
I was still watching his mouth when the fire alarm sounded.
It came from the lower deck. The station used a Gent electronic fire-alarm sounder, and it was loud in a way that occupied every available space inside your head the instant it began.
One moment I was watching Graham's mouth. The next there was nothing but the alarm and the red pulse of the indicator light on the panel behind me.
His mouth was still moving. I could not read it. I could not hear anything. The alarm was a wall.
I looked at him for one more second, searching for something - a flicker, some recognition in the eyes directed at me from the foot of the stairs.
I did not find it. Or I could not tell.
That is the honest answer.
I could not tell.
The notebook was in my hand. The Secomak Type 447 warning siren was on the north catwalk. Nobody was going to ring back and tell me I was authorised to use it.
I turned and went for the catwalk hatch.
It opened outwards, and the wind struck me at once. Cold and heavy with salt, driving in from the North Sea. November on Spurn Peninsula has a particular kind of cold. It gets inside your collar and stays there.
I noted it in the way you note anything when your mind is running on procedure. Filed it. Moved on.
The Secomak Type 447 stood where it always had, mounted beside the north railing on a bracket bolted through the catwalk grating. Its horn pointed along the exposed seaward beach, an odd position in relation to the railing but the correct one for carrying a warning down the shore. The manual crank was fitted on the right-hand side.
I had walked past it hundreds of times during shift inspections. I had never touched the crank. The motor drive was the point of the system. You operated a switch and the siren ran.
The motor was dead. That had been the problem since the opening checklist.
As it turned out, it was also the solution.
I took hold of the mounting bracket with my left hand, because the crank was on the right and I needed something to brace myself against.
The bracket was cold painted steel, slightly rough where the paint had chipped along the edges. I secured my grip, reached across with my right hand and found the crank handle.
It was smaller than I had expected, about the size of a door handle. There was nothing dramatic about it.
I began turning.
The first rotation was stiff. The mechanism had not been operated manually for years, perhaps longer. It ground, caught, then released into a smoother movement.
The second rotation was easier.
By the third, the tone had begun to build: a low, sustained note rising from the horn and carrying past the railing, across the beach and over the surf.
360 Hz.
I know what that sounds like now. I knew then because Peter's notebook contained a page on which the frequency had been written out and underlined twice, and I had spent the better part of an hour staring at it.
It is not a pleasant sound.
It is not meant to be. It is a warning tone designed to carry over distance, to cut through wind, surf and the general indifference of the atmosphere.
It did.
I looked down at the beach.
All 27 figures convulsed at exactly the same moment.
Not one and then another. Every one of them simultaneously, as though something had passed through the entire group at once. A current, a signal, or whatever word applies when 27 bodies respond to the same input in the same instant.
They went rigid, folded in ways they should not have been able to fold, then straightened again. Those in the front rows stumbled backwards into those behind them.
It did not look exactly like pain.
I do not know what it looked like. It was involuntary. Something they were not doing by choice.
Then they began moving back towards the water.
Slowly at first, then faster. They did not run. Their movement was not coordinated enough for that, but it had direction. All of them were orientated towards the surf line. The rows collapsed into a single mass moving away from the station pylons, away from the siren and towards the North Sea.
I kept cranking.
My right arm did the work. My left hand held the bracket and kept me braced against the railing while the crank pulled at my shoulder and the wind pressed from the other side.
For the first minute, the grating beneath my feet remained solid.
Then I felt the first bolt give.
It did not snap. It pulled loose. The bracket mounting at the north-east corner shifted by about a centimetre, and the grating flexed beneath my right foot.
I looked down. The bolt remained in place, but the metal around it had deformed. The grating was pulling away from the mounting on one side.
I shifted my weight to the left and kept turning.
The second bolt gave about 30 seconds later. The same side, a different corner. The catwalk shifted by roughly 7 degrees, then caught against another bracket further along the railing.
It held, but it was no longer level.
I was standing on an angled metal surface, with the wind driving in from the sea and 27 pale, barnacle-covered figures retreating across the beach below me, and I was still turning the crank.
I want to be clear that I was not calm.
I was functioning. That is a different thing.
My hands were doing what they needed to do. The rest of me was somewhere else, operating on something that was not quite thought.
The figures were perhaps halfway to the surf when my shoulder went.
It was shortly after the third minute. I know because I had been counting rotations without intending to. My mind had selected a number and was keeping hold of it.
I had reached roughly 170 rotations when the joint dislocated.
I felt it before I heard it, which is an interesting order for those events. First came the sensation of wrongness, of something moving in a direction it had not been built to move. Then came a sound I am not going to describe.
Then my left arm stopped working.
It did not hurt immediately.
That is the thing about dislocations. There is a window of a few seconds while the body processes what has happened and the pain signal remains in transit.
I used that window.
I released the bracket with my left hand whilst keeping my right on the crank. Then I hooked my right forearm around the bracket, bracing the inside of my elbow over the painted steel and leaning my body weight into it.
I kept turning.
The crank was on the right. My right hand could still operate it. My left arm hung at the wrong angle, and I did not look at it.
The pain arrived.
I noted it and kept turning.
The grating shifted again beneath my feet, a longer movement this time. I widened my stance and leant harder into the bracket.
The tone continued to carry across the beach, 360 Hz, steady and sustained.
The figures kept moving.
I could see Graham.
He was near the back of the group, making 28 of them now. He remained upright and moved with the same peculiar gait he had used at the foot of the stairs.
That was not how Graham Pritchard had walked when he was Graham Pritchard.
He moved with the others towards the surf, his arms still making the same reaching gesture. Even as he walked away from the station, he continued reaching back towards it.
I watched him reach the water.
The North Sea took the front of the group first. They did not stop at the edge. They walked in. The surf rose to their knees and then their waists, and they continued. Those behind followed.
Graham was among them, still reaching, still moving. The water climbed to his chest, then his shoulders.
Then he was gone.
I kept cranking until the last of the 28 figures went under.
Then I continued for a little longer, because I did not trust the last figure merely because it had disappeared beneath the water.
When I finally stopped, the beach was empty.
There was only wet sand, surf, dark water and wind.
I stood on the tilted grating with my right forearm still hooked around the bracket and my left arm hanging incorrectly at my side. The siren wound down, its tone falling away as the mechanism lost momentum.
I looked at the beach for a long time.
The fire alarm was still sounding below me. Even with the siren dying, I could feel its low vibration transmitted through the station structure. Somewhere beyond the catwalk hatch, the red panel light continued to pulse.
The beach was empty.
I checked again, sector by sector, as I would check an instrument panel. North to south. Surf line to pylons.
Nothing.
Only sand.
They were gone.
Getting back through the hatch was harder than I expected. My right arm still worked. That was not the problem. I had been bracing against the bracket for so long that my right hand had locked around it.
When I finally let go, my fingers would not cooperate. I had to shake them loose one at a time, watching them as though they belonged to somebody else.
I held my left arm against my body. I had established quickly that moving it was a bad idea.
The catwalk finally rested about 12 degrees out of level. Three of the four bracket bolts had pulled loose from the railing mounting during the last few minutes of cranking. It had held, but I did not want to remain standing on it.
I turned carefully, caught the hatch frame with my right hand and lowered myself back into the station.
My left arm swung as I dropped down, and I made a sound I am not going to describe.
Then I was inside.
I sat on the floor beside the hatch. I did not make it as far as a chair.
The floor was a standard non-slip composite surface. It was cold, solid and exactly where I needed it to be.
The fire alarm had stopped at some point. I was not sure when. Probably during the last few minutes of cranking. I had stopped registering it.
The station was quiet now, except for the wind against the structure, the low background hum of the instrument array and, somewhere beneath me, water moving around the pylons.
Normal sounds. Station sounds.
I sat and listened for a while.
My left arm was useless. That was the clinical description. The shoulder had come apart shortly after the third minute of cranking. I had felt it go, shifted my grip and continued because there was no version of stopping that made sense.
Now the joint sat at an angle that did not correspond with how a joint should sit. The whole arm, from shoulder to fingertip, had become a dull, insistent wrongness. Not sharp. Constant.
I kept it pressed to my side and did not look at it.
The sky beyond the east-facing windows was changing. I noticed it gradually. The black had faded into deep grey-blue, and a thin line of paler grey lay above the North Sea.
It was the colour the sky takes perhaps 40 minutes before sunrise, when it is no longer dark but is not yet anything else.
My shift was nearly over. The night was nearly over.
That was happening whether I was ready or not.
I got up from the floor. It took a minute. Then I walked to the instrument console.
The Coastguard VHF uplink was still functioning. Its green status light was steady.
The antenna array had been Graham's last job. He had gone down to inspect it just before everything went wrong. Whatever he had done before the figures reached him had apparently been enough. Or perhaps the array had been sound all along and he had simply needed something to do.
Either way, the uplink was live.
I opened the template for the 06:00 maritime safety information transmission. It was the standard joint Met Office and MCA format: pressure, wind speed and direction, sea state, cloud base and active warnings.
I had completed it on every shift for four years. I knew the order by heart. I could have done it half asleep, which was useful because I was close to that.
I read the pressure from the display first.
1,012 hPa and rising steadily.
I entered it.
Then wind speed and direction. The anemometer on the upper mast showed 14 knots from the north-east.
I entered that as well.
I had to check the ceilometer twice because my eyes were no longer focusing properly, but the result was consistent: broken cloud near 670 metres, with another layer near 1,220 metres.
I recorded the layers in the summary field using the correct terminology.
Then I checked the sea-state data from the buoy network.
Monitoring buoy OD-17 was transmitting normally again. Different readings cycled through in sequence, exactly as they were supposed to, instead of the same frozen values repeating every 15 minutes.
I looked at that for a moment.
Then I moved on.
An inshore waters strong-wind warning remained valid until 18:00. I confirmed it without amendment. It was accurate.
At 06:00, I keyed the microphone and read the report aloud. The same categories I used on every shift, in the same format and the same order.
Pressure.
Wind speed and direction.
Cloud layers.
Sea state.
Active warnings.
I said every word correctly.
My voice remained level, which surprised me.
"Inshore waters strong-wind warning remains in force until 18:00. Mariners are advised to exercise caution over Outer Dowsing shoal and adjacent North Sea waters. This is the scheduled 06:00 maritime safety information transmission from Spurn Head Coastal Monitoring Station, station identifier Spurn Echo Four. Next scheduled transmission at 09:00."
I released the microphone key. The console light changed from red to green.
I stood with my right hand on the edge of the console and my left arm held against my side, and I thought about Peter Wainwright.
He had been in that store cupboard for I did not know how long. His cheap two-pound reporter's notebook covered eleven months, from the routine entries of November 2025 onwards. Eleven months of documented encounters. Eleven months of margin notes.
Eleven months of frequency tables, buoy-footage time stamps and careful, methodical records.
He had known about the siren. He had known exactly what it did. He had possessed the information all along.
He simply had not been authorised to act.
He had been waiting for somebody above him in the chain to say, "Go ahead."
That authorisation had never come.
He had ended up wedged behind a battery rack, both eardrums ruptured, with the word OPEN scratched into the metal shelf so many times that the letters had worn down to almost nothing.
Whether he had meant the lower-deck hatch or the figures' means of entry, I still could not tell.
I was not angry with him. I do not think I had enough energy for anger. It was more that I understood him. I understood the instinct to wait for somebody else to make the decision.
I had done it myself with Graham.
I had logged his arrival at 23:24, 39 minutes late, then decided not to escalate the smell of blended Scotch whisky on him. I had told myself that one bad night was not worth ending a 22-year career.
I had waited too.
It had not caused what came out of the sea. I knew that. But it was still the same habit: observing, recording and leaving the decision for somebody with more authority.
I had simply run out of things to wait for sooner than Peter had.
The difference was that I had reached morning.
Graham had not.
I kept returning to that in the same flat, factual way I would return to an instrument reading that made no sense.
Graham Pritchard, 54, with 22 years at the station and no previous adverse action, knew more about buoy telemetry than anybody I had worked with. He had stood in the surf with his jaw hanging slack and barnacle crust spreading from his chin, reaching towards the station pylons as the water rose around him.
Then he had gone beneath it.
I did not know what had happened between his radio falling silent and his appearance at the foot of the stairwell.
I did not know whether any part of him remained by then, or whether the thing that climbed the stairs had merely been wearing him.
I was probably never going to know.
That part would remain open. I supposed some things did.
The shift log was still open on the secondary monitor. The last entry was time-stamped 03:07.
FIRE-ALARM ACTIVATION, LOWER DECK. CAUSE UNDETERMINED.
That was the last thing I had written before going through the catwalk hatch.
I watched the cursor blinking in the entry field.
I did not know how to write what came next. I am not good at those entries even on an ordinary day. Incident reports, anomaly logs, anything requiring a description of events outside the normal operating parameters.
The format had no field for figures on the beach, a colleague lost to the sea or deployment of a warning siren without authorisation.
Unauthorised. Effective.
There was no drop-down option for any of it.
I left the log as it was. No further entries.
Somebody else would have to resolve the paperwork. That was acceptable. It was a problem for daylight.
Outside, the grey above the North Sea had grown lighter. The water remained dark, and the beach beneath the station was empty.
I had already checked it twice through the east window, sector by sector, north to south, surf line to pylons.
Sand.
Only sand.
The tide was coming in, covering the places where the figures had lain and stood. Within another hour or two, there would be no trace of them.
The sea was good at that.
I sat in the console chair, which was more comfortable than the floor. I placed my right hand flat on the desk and looked at the instrument panel.
Everything showed normal operation. Pressure rising steadily. Wind consistent. Uplink green. Buoy network nominal.
The station was functioning.
So was I.
That was what I had needed to prove, and the transmission had proved it. I was still there, still inside Spurn Head Coastal Monitoring Station, still capable of reading a pressure value from a display and speaking the correct words into a microphone.
Peter Wainwright had documented everything, waited for somebody to act and died waiting.
I had acted.
The cost was Graham, the use of my left arm and whatever came next: the report, the questions and the parts of that night I would have to say aloud to people who had not been there.
The sky continued to brighten.
The 06:00 transmission was complete. I kept my eyes on the instruments and waited for daylight relief and the inevitable official response.