33 KiB
Chapter 16: The Chain
Leon's door smelled like tallow and regret.
Not his regret — the chandler's. The shop below had been rendering a new batch, and the fumes had climbed the stairwell with the determined patience of a smell that knew it was unwelcome and didn't care. I knocked twice, waited, and heard something being moved off something else — books off a chair, probably.
He opened the door chewing. Monday morning, already eating like a man who'd forgotten meals were supposed to have schedules.
"No," he said.
"I haven't asked anything yet."
"Your face asked." He leaned against the doorframe, arms crossed. "You've got something. Something you can almost see but can't quite reach. And you need someone to hold the other end of the thread while you pull."
(Three layers. Stabilizer architecture mapped from memory but the interaction model between Layer 2 and the anchor — the feedback timing is wrong. Not wrong. Incomplete. The monitoring pattern has a rhythm I can feel but can't quantify without someone who thinks in volume where I think in precision—)
"Something like that."
"And you want to do this at the guild."
I'd told him the location in advance. A courtesy, because Leon's relationship with institutional spaces was the relationship most people had with dentistry: acknowledged as occasionally necessary, endured with visible discomfort, and avoided whenever plausible alternatives existed.
"The planning room at 14 Greystone Lane. This afternoon, second bell."
"I know how you feel about guilds," I said, before he could. "I also know how you feel about impossible problems. I've got a guild case — man named Ned Floundry, cursed, dying. The Compact classified it as unbreakable in eleven days. Two licensed curse-breakers failed. It's not one working — it's three nested layers, each designed to protect the others, built by someone who understood exactly how the Compact assesses, exactly how certified curse-breakers operate, and exactly which assumptions they'd never think to question."
Leon was quiet for three seconds. With Leon, three seconds was a geological age.
"Three nested layers. And the Compact classified it in eleven days." He'd caught the same thing I had — standard assessment ran four to six weeks. "That's not a classification. That's a coverup."
"That's my read."
He looked at me — past the surface read, into the part where the math lived. "You need me for the stabilizer — the middle layer. The forge-and-redirect, the death ward approach. But a nested stabilizer is monitoring two other systems."
"And I need someone who thinks in brute force where I think in precision."
The corner of his mouth moved. Not a smile — an acknowledgment that the puzzle had already started pulling. "You're asking me to come to the guild."
"I'm asking you to come to a room that happens to be inside the guild. Table, chalk, no institutional obligations. Nobody signs anything. Nobody takes twenty percent."
He considered this. "Fine. Second bell. If the chairs are as uncomfortable as you've described, I'm blaming you personally."
"That's fair."
I turned to leave. Behind me, Leon said: "Phelan." I stopped. "The fact that you're asking means this is the one you can't do alone."
It wasn't a question. Leon's observations never were.
The noise had been building since Godsday night — the mine observations feeding into the curse architecture feeding into the exploit mapping, all of it running in parallel tracks that kept almost converging and then splitting apart. Layer 3 sat in the middle like a locked door I kept walking past, testing the handle each time.
(Anchor keyed to life-force signature. Drift visible — the curse is changing Ned's signature as it degrades him. The lock is losing its grip but not fast enough. What accelerates drift without accelerating degradation? Separate the variables. How do you separate variables that share a host—)
I walked toward Greystone Lane with the noise running hot and the afternoon still hours away.
The guild planning room at 14 Greystone Lane hadn't changed since the mine expedition briefing two days ago. Same long table, same chalk supply, same institutional lighting that managed to illuminate everything without making anything look warm.
Mere and Devod were already there.
Mere sat at the far end with the collection kit, notes spread in three neat columns. The dampened cloth wrappings were open — she'd been checking the moss. Alive. Stable. Essential.
Devod was standing, because sitting implied commitment. He had a list on the back of a delivery manifest.
Between them: the evidence. Greenvale Provisions receipt. Patrol logs. Curse-breaker reports — Practitioner Vellen's from week three, Practitioner Dorath's from week five. Payment notes.
"He arrived forty minutes early and has been rearranging the chalk," Mere said, without looking up. "I arrived at the scheduled time."
Devod held up the chalk tray. "They were sorted by color but not by length."
(Two people who haven't been in the same room voluntarily for twelve years. The door that opened in the mine cart hasn't closed. Devod still tracks Mere when she's not looking. Mere still speaks about him in the third person when he's three feet away. But they're here.)
The door opened. Leon walked in, stopped, and surveyed the room with the expression of a man confirming his worst expectations.
"They labeled the chalk," he said.
"That was me," Devod said.
Leon pulled out a chair, sat, and immediately shifted his weight. "The chairs are worse than you described." He looked at Mere, then Devod, then the evidence on the table. The complaints had been performative — Leon's version of stretching before a run. His eyes were already moving across the documents.
"Leon D'Nardis," I said. "Mere Fields. Devod Fields."
"The bookshop," Leon said to Mere. "Thresholds. Pre-Compact theory section. You rearrange the binding-theory subsection when you're thinking."
Mere looked at him with the flat attention she gave things that had just become interesting. "You've been to my shop."
"Twice. Your pre-Compact notation collection is the best in Drenwick."
Leon studied her for half a second longer than politeness required — the particular stillness of someone filing a data point that didn't fit the expected category. Then he looked at Devod. "You sorted the chalk."
"Someone had to. It was chaos!"
"I like you already." The smile was genuine — not the performative version he deployed for social navigation. Leon recognised the restless energy because he lived three doors down from it in his own head.
"Three layers," I said. "Two solved. One not. Ned Floundry has days, not weeks."
The room focused.
I drew the three-layer architecture on the table with chalk — nested circles, the degradation on the outside, the stabilizer in the middle, the anchor at the core.
"Layer 1: degradation curse. What the curse-breakers saw. Real, killing Ned, but protected by Layer 2 — a stabilizer that repairs faster than any dissolution can damage. Solution: ghostveil moss." I pointed to Mere's collection kit. "Dampening agent. Thirty to sixty minutes where the degradation is weakened enough to crack. Mere's preparation."
"Layer 2: stabilizer. Solution: forge-and-redirect." I tapped the middle circle. "Same technique family as the Greymarch death ward. Forge data matching the stabilizer's monitoring signature, feed it back, turn the repair function against itself."
Leon leaned forward. The chair complaints had been forgotten — the puzzle had eaten them whole. "Show me the monitoring architecture."
I drew it — the channels watching Layer 1 and Layer 3, the repair pathways, the feedback loops.
"Layer 3." I drew the innermost circle. "The anchor. Dead man's switch. Break Layers 1 and 2 without addressing this, and a kill mechanism fires directly into Ned's life-force. Keyed to his specific signature — bonded like a compass needle."
"And you can't forge that," Leon said. "Life-force signatures are unique."
"So Layer 3 is—"
"Unsolved."
The room was quiet. Then Leon pulled the chalk from my hand.
"Show me the stabilizer again. The monitoring channels — how does it watch Layer 1?"
This was where it started.
Leon drew over my diagram — his lines thicker, less precise, but spatially intuitive in a way mine weren't. He thought in systems the way I thought in seams.
"The stabilizer reads the degradation curse's energy signature every two seconds. Any deviation triggers a repair cycle — four to six seconds to complete."
"So you have a two-second window to inject forged data before the next read."
"The bracelet helps." The words were out before the calculation about whether to say them had finished running. The case demanded it.
I pulled back my sleeve. The bracelet sat against my wrist — dark mineral, oval stone shifted from amber-red to something warmer, almost copper. The stone didn't gleam. It absorbed.
Leon's stillness changed quality. Not puzzle-processing — reassessment. A conversation over wine and mismatched cups, six days ago, suddenly recontextualized.
"The Barrows," he said. "Second floor."
"Behind the death ward. Pre-Compact. Focusing matrix that refines magical perception — it's what gives me the precision for two-second injection windows. Without it, I can see the monitoring pattern but I can't forge convincing data at that speed."
"You shared the technique but not the tool." Not an accusation. An observation.
"The technique was ours. The tool was—"
"Yours. Until it wasn't." He looked at the bracelet for three seconds, then back at the diagram. "Show me the monitoring pattern. Does the two-second interval vary under load?"
The personal implication processed, catalogued, respected, and set aside in favor of the problem. Leon's trust wasn't binary. It was a gradient, and he'd just watched the needle move.
(He would have done the same. The Mallory crystal buyer, unnamed and unquestioned. Compartmentalization isn't betrayal. It's architecture.)
"Consistent under normal conditions," I said. "But when the degradation shows damage, the stabilizer accelerates. Monitoring interval drops to under a second. Repair cycle doubles in speed."
"It panics," Leon said.
"It prioritizes."
"Same thing, different framing." He was drawing now — thick lines showing energy flow, repair pathways. "You did this to one system in the Barrows. The stabilizer is the same architecture, just nested."
"But it's monitoring two layers, not processing external input. It cares about the health of the entire curse."
Leon grinned. The sharp one — the one that meant his brain had caught something. "That's not a problem. That's a feature."
I waited.
"Systems that monitor other systems," he said, "can be made to accelerate the very failures they're trying to prevent."
(Monitor reads degradation. If monitor receives forged data showing catastrophic collapse — the stabilizer doesn't just repair. It overcompensates. Dumps energy into reinforcement. The repair energy has nowhere productive to go. It overloads the channels—)
"You feed it a crisis that doesn't exist," I said. "The stabilizer pours repair energy into the degradation curse, but the degradation isn't actually collapsing. The repair function, running at emergency speed with nowhere to dump its output—"
"Turns into the weapon," Leon finished. "The stabilizer doesn't just fail. It actively tears Layer 1 apart from the inside. You're not dismantling it. You're weaponizing it."
"What if it just stops?" Devod said. "The monitoring thing — can't it switch off? Like a horse that baulks?"
"Monitoring systems don't have that option," Leon said. "Either it overloads and causes damage, or it locks up. Both outcomes work for us."
"Right. Just thinking." Devod made a note. "What about cooling it down? Redirect the excess energy somewhere safe?"
"The overload is the weapon," I said. "We want it."
"What about using water? The mine had magical water—"
"No," Mere said, without looking up.
(Three bad ideas. But Devod doesn't generate ideas to be right — he generates them like Leon generates volume. The ratio holds because the search space is wide enough to occasionally contain something nobody else was looking for.)
Leon looked up from the diagram. "That's Layer 2 solved. The stabilizer becomes the primary attack vector against the degradation. Two layers, one exploit."
"What about fire?" Devod said. "Could you set fire to the monitoring channels?"
"No," Leon and I said simultaneously.
"Exploring options," Mere finished.
The pause settled. Two problems solved. A third sitting in the center of the table.
Mere had been reading — not the moss notes, but the curse-breaker reports, spread beside the patrol logs and the Greenvale Provisions receipt.
"Both curse-breakers followed the same procedure," she said.
Leon stopped drawing. Devod looked up.
"Practitioner Vellen, week three. Standard dissolution framework, single-layer assessment. Practitioner Dorath, week five. Same framework. Different practitioner, identical methodology." She tapped the reports. "The procedure doesn't account for nested workings. Neither assessment includes a protocol for detecting concealed secondary or tertiary structures."
She looked up. Her eyes met mine.
"Who wrote the procedure?"
(If the procedure doesn't test for nested workings, then nested workings are invisible to anyone following it. And if someone knew the curse had three layers when they wrote a procedure that could only see one—)
Leon went stone-still. He understood institutional corruption structurally, the clarity born of a career spent operating outside institutions because he'd seen what happened inside them.
Devod's face changed. Something rawer than anger. He didn't understand the magical mechanics. He understood the sequence. Reports filed. Nothing happened. Curse hit.
Mere looked puzzled by the silence. "The data is straightforward. The procedure omits detection protocols for nested structures. Given the eleven-day classification, the Compact's control of ghostveil moss growth sites, and the shell companies monitoring remaining wild sources — the probability of coincidental omission approaches zero."
She said all of this in the tone she'd use to describe a shelf reorganization.
"They were set up," Leon said quietly. "The curse-breakers. Given a map to the wrong room."
"They were given the only map the Compact distributes," Mere said. "Which is the same thing."
(Two tracks now. Three-layer curse, three-layer conspiracy. Ned wasn't silenced by the curse alone. He was silenced by the procedure that ensured nobody would hear what the curse was saying.)
"Could we trace the Greenvale receipt?" Devod said, voice tighter. "Follow the chain?"
"Ned doesn't have time for an investigation," I said. "He has days."
"After," Devod said. The word cost him something. "After we fix him."
"After."
Devod looked at his list. Crossed something out. Wrote something else. "Right. So — what about the anchor? The third layer. The kill switch. If you can't fake Ned's — what did you call it — life-force signature, and you can't just remove the anchor without triggering it, what if—" He squinted at his own handwriting. "What if you built a second anchor? A decoy? Something that looks like Ned's signature but isn't, and you switch them?"
"You can't fabricate a life-force signature," I said. "It's not a magical construct. It's the sum total of a person's biological and arcane presence. It's as unique as a face."
"What about masking it? Like — camouflage?"
"The anchor isn't looking for Ned. It's bonded to him. Camouflage only works on systems that search. This one already found what it's attached to."
"Right." Devod frowned. Made another note. "What if we removed the anchor's power source? Cut off its energy?"
"The anchor draws from the curse's own energy loop. Cutting the power means dismantling the curse, which triggers the anchor. It's circular by design."
"Circular," Devod repeated. "Like a dog chasing its—"
"More or less."
"What if—"
"Devod," Mere said. Not sharply. Just… precisely. The way she said things when she'd finished listening and was about to speak.
Devod looked at her.
"The binding salts," I said.
It came out before the conscious thought formed — the connection firing in the noise and arriving at my mouth by way of a shortcut my social filters hadn't had time to intercept.
Mere looked at me.
"What you did with Sniff," I said. "The dog curse — three weeks of binding salts in the water bowl, creating a localized dampening field. You built a suppression zone from intuition and patience. What we're doing with the ghostveil moss is the same principle. Your approach from the bookshop. Formalized, scaled up, applied to a three-layer lethal working instead of a single-layer fear curse. The binding salts were the sketch. This is the finished drawing."
Mere's expression didn't change. Her eyes held mine for a moment — not searching, not evaluating. Recognizing.
"The principle scales," she said. "That's what principles do."
No false modesty. No deflection. Just the flat statement of someone who had already known the connection and didn't require external validation to believe it. The room noticed — Leon with a fractional tilt of his head, Devod with something moving behind his eyes that he didn't have the vocabulary for and wouldn't have used if he did.
(Rare. I don't credit people aloud. I note competence, file it, factor it into my calculations. Naming it out loud — saying the thing instead of thinking it — that's Mere's territory, not mine. She says what she sees. I see what I don't say. But the dog and the moss and the principle that connects them — that needed to be said. Because the room needed to know that the woman monitoring substrate integrity at the end of the table had built the conceptual foundation for the cure before any of us knew there was a cure to build.)
"Right," Devod said, "so the binding salts worked on the dog. The moss works on the curse. But the anchor—"
And then it came.
Not announced. Not prepared. Buried in the seventh idea after six that had made at least one person in the room physically uncomfortable, delivered in the voice of a cargo driver who'd spent his life getting heavy things from one place to another.
"If you can't unlock the lock," Devod said, "move what it's locked to."
He said it the way he said most things — casually, practically, with the easy confidence of a hundred solved problems in a context nobody in this room would have thought to reference. He was looking at his delivery manifest, not at us. The note he'd been writing when he said it was about something else entirely.
"Like a boat," he continued, already moving past it. "Cargo chained to a dock pylon — you don't cut the chain, that takes too long and wrecks the chain. You move the boat until the chain goes slack. Then it slips off. The lock's still locked. The chain's still whole. But the thing it was attached to isn't where it was anymore, and—"
He stopped. Because the room had gone silent.
Not quiet — silent. The kind of silence that happens when multiple brains simultaneously arrive at the same conclusion from different directions and the collective intake of breath leaves no air for sound.
(The anchor is keyed to Ned's life-force signature. Bonded to it. Locked onto it like a compass needle pointing north. It has to be — a kill mechanism without signature verification is a kill mechanism that fires into whoever's standing closest, and whoever built this curse was precise enough to nest three layers. They weren't sloppy. The anchor validates before it delivers. Confirms the signature matches, confirms the target is the target, then fires. Same reason a sending-stone won't discharge into the wrong recipient — the keying is the safety and the constraint.)
(But constraints cut both ways. If the signature drifts far enough from what the anchor expects, the validation fails. Not broken — failed. The mechanism fires and finds nothing it recognises. The curse is already causing that drift — the degradation is changing Ned's signature as it damages him. The anchor is losing its grip. Slowly. Too slowly. The drift is natural, involuntary, a side effect of the curse's own damage. But if you accelerate the drift — if you move the boat — the anchor's validation window closes. The lock is still locked. The mechanism is still intact. But the thing it's locked to doesn't match anymore, and the lock is holding nothing. The kill trigger fires. It fires into empty space. The chain goes slack and slips off and Ned is—)
(Ghostveil moss. Standard preparation: dampening agent, suppresses magical energy, creates a thirty-to-sixty-minute window for Layer 1 dissolution. But at a different concentration — altered preparation, different ratio — the moss doesn't just dampen the curse. It dampens the connection. Suppresses the resonance between the anchor and Ned's signature. The curse is already causing drift. The moss, at altered concentration, accelerates it. Not by damaging Ned — by widening the gap between what the anchor expects and what it finds. The lock is still locked. The chain is still whole. But the boat has moved.)
(One ingredient. Two applications. Standard preparation for Layer 1 dampening. Altered concentration for Layer 3 drift acceleration. Mere's botanical expertise for both — she's the only person in this room who can prepare ghostveil moss at all, let alone calibrate two different concentrations for two different mechanisms from the same source material.)
Leon was the first to speak. His expression had undergone a journey — surprise, assessment, the rapid cascade of technical implications, and finally something that lived in the neighborhood of respect with a brief stopover in disbelief.
"Did he just—" Leon looked at me. Looked at Devod. Looked at the diagram on the table. "Did he just solve Layer 3 with a boat analogy?"
Devod looked up from his manifest. "What?"
"The anchor," Leon said, and his voice had the quality it got when his brain was running ahead of his mouth, which was rare because Leon's mouth was very fast. "The anchor is keyed to Ned's life-force signature. The curse is already causing drift — the signature is changing. If you accelerate the drift faster than the anchor can recalibrate, the lock loses its grip. The trigger fires, but the target has moved. It fires into nothing." He was standing now. When had he stood? "The ghostveil moss — you can use it twice. Different preparation, different concentration. One batch dampens the curse for Layer 1. The other batch accelerates the signature drift for Layer 3. Same ingredient. Two applications. That's—" He stopped. Laughed. Actually laughed — the real one, not the transactional version. "That's the most elegant piece of lateral thinking I've heard in five years, and it came from a cargo analogy."
"It's how chains work," Devod said, still looking confused about why the room was staring at him. "You don't cut things if you can move things. Cutting is expensive. Moving is—"
"Moving is what you just did." Leon sat back down. Stood up again. Sat down. The chairs were forgotten. The guild was forgotten. Everything was forgotten except the shape of the solution forming on the table. "Phelan. Tell me I'm wrong."
"You're not wrong."
"The drift acceleration — the moss at altered concentration — it doesn't damage Ned?"
"The moss suppresses magical energy. At standard concentration, it creates a dampening field — Layer 1 application. At altered concentration, it suppresses the resonance between the anchor and the signature it's bonded to. The drift is already happening naturally — the curse is changing Ned as it damages him. We're not causing new damage. We're accelerating a process the curse itself initiated."
"That works," Mere said.
She said it before Leon finished processing. Before I finished explaining. Two words, flat, delivered with the certainty of someone whose pattern recognition had run the logic to its conclusion while the rest of the room was still catching up.
"The concentration differential is manageable," she continued. "Standard preparation targets the ambient magical energy in the curse's operational field — broad suppression, general dampening. The altered preparation would need to target the resonance frequency between the anchor and the life-force signature specifically. Narrower band. Higher concentration. Different preparation timeline." She looked at me. "I need the anchor's resonance characteristics. The frequency it's monitoring."
"I can get them. Bracelet-enhanced observation, one more session at Ned's bedside."
"Then I can prepare both concentrations from what we harvested." She paused. "I'll need twelve hours for the altered preparation. The moss has to be processed differently — not dried, not standard extraction. A living preparation. The root structures I preserved will be essential."
(She kept extra. In the mine, when she said "Not everything I wanted, but enough" — she harvested more than the minimum. Preserved root structures. Extra stock. Because Mere prepares for the problem she can see and keeps her options open for the ones she can't. She didn't know about the second use. But she'd prepared for it anyway.)
Devod was looking at Mere. His face was doing something he'd never have the words for. A father watching his daughter be extraordinary in a context he didn't fully understand, recognizing the competence without comprehending the technical specifics, proud in a way he wouldn't name because naming it implied a claim to the relationship, and the relationship was built on twelve years of distance and a lie he didn't know how to unmake.
Mere was looking at the moss. She didn't see her father's face. I did.
(Filed. Not mine to share. Not mine to fix. But filed.)
"So," Leon said, and his voice had the measured quality of someone organizing a detonation sequence. "Layer 3 first. Anchor loosened via accelerated drift — ghostveil moss at altered concentration, Mere's preparation. Then Layer 2 — stabilizer confused via forge-and-redirect, bracelet precision, the weaponized repair mechanism. Then Layer 1 last — degradation cracked during the dampening window from standard-preparation moss. Thirty to sixty minutes."
"That's the order," I said.
"Layer 3 first because the anchor has to lose its grip before you disrupt the stabilizer. Otherwise the kill trigger fires the moment Layers 1 and 2 go down."
"Correct."
"And Layer 1 last because the dampening window from the standard moss is time-limited. You need Layers 3 and 2 cleared before you spend that window on Layer 1."
"Correct."
"How long for Layer 3?"
"I need to observe the anchor again," I said. "One more bedside session. The bracelet can map the resonance characteristics — recalibration speed, grip tolerance, the threshold where the bond becomes too weak to maintain targeting. Then Mere can calculate the concentration."
"How long at the bedside?"
"An hour. Maybe two."
"And after that — how long for the actual drift acceleration during the cure?"
The noise wasn't answering a question. The noise was building a structure. And the structure was pulling me in.
(Execution sequence. Layer 3: altered moss applied, drift acceleration begins. Estimate forty-five minutes to critical threshold — the real number needs the resonance data. Layer 2: forge-and-redirect once the anchor's grip drops below functional threshold. Bracelet precision for two-second injection windows. The stabilizer reads false crisis data and overloads its own repair mechanism. Fifteen minutes once forging begins. But the bracelet's reservoir — partially depleted from the wellspring discharge in the mine — the focusing matrix draws from the reservoir during precision work. If the reservoir runs dry mid-forge—)
"Phelan?"
(—the precision degrades. Without the bracelet, injection precision drops from sub-second to approximate. The margin between convincing forgery and detected intrusion in a two-second window is three-tenths of a second. The bracelet buys me that margin. Without it — forty percent detection probability per injection cycle. The stabilizer learns. Each failed forgery makes the next one harder. So the bracelet is essential. Reservoir has to hold. Seventy-five percent, minus observation cost, minus sustained forging cost—)
"He's doing the thing," Leon said, from somewhere that wasn't quite the room.
(—twenty-two to twenty-eight percent reservoir cost for the full Layer 2 sequence. Starting at seventy-five, ending at approximately fifty. Sufficient. But Layer 3 observation also requires bracelet precision. Add eight to twelve percent. Sixty-three percent entering the cure. Layer 2 costs twenty-two to twenty-eight. Landing at thirty-five to forty-one. Layer 1: minimal bracelet assistance, five percent. Final: thirty to thirty-six percent.)
"He does this," Leon was saying. The words arrived as shapes, not sentences.
A chair scraped. Devod, shifting his weight. The sound registered as friction coefficient and distance — three feet to the left, irrelevant — and was discarded.
(Energy budget. Personal reserves: seventy to eighty percent after rest. The hyperfocus — this, right now — is costing five to eight percent per hour. Two hours: ten to sixteen percent. Landing at fifty-four to seventy before the cure. Layer 2 forging costs personal reserves on top of bracelet reservoir. The death ward cost sixty-five percent. The stabilizer is more complex but the bracelet's buffer reduces drain — estimate forty-five percent personal cost. Starting at fifty-four to seventy, minus forty-five, landing at nine to twenty-five.)
Someone put a hand on the table. The vibration traveled through the wood and registered as data.
(Nine percent. Floor scenario. The Barrows — fifteen percent at the end, barely conscious. Nine is below the threshold for controlled magical work. If the floor hits: Mere maintains the dampening field. The moss does the work. Layer 1 dissolution occurs naturally — the degradation curse, deprived of stabilizer and anchor, has no reinforcement. Standalone working being dampened by ghostveil moss. The same principle as binding salts on the dog. Mere's principle. Scaled up.)
(Contingency sufficient. The chain holds even at the floor.)
(The chain.)
(Three methods for three problems. Ghostveil moss — Mere's principle, scaled from a dog to a man. Forge-and-redirect — my technique, born in the Barrows, modeled by Leon, weaponizing the system's own protection. Anchor drift — Devod's concept, delivery-driver logic, move the boat instead of cutting the chain. One ingredient used twice. One technique adapted from a dead ward to a living curse. One analogy that saw the problem the way none of us could because none of us thought in cargo chains and dock pylons.)
(Four people, three methods, one cure. The chain holds. The chain—)
"Phelan." Mere's voice. Closer than before. When had she moved?
(Layer 3 first. Altered moss, drift acceleration. Bracelet tracks resonance decay — nonlinear, slow initially, then accelerating past the anchor's recalibration ceiling. The anchor will over-correct. Chase the signal. Every over-correction widens the gap. T-zero: anchor loses grip. Begin Layer 2. False crisis data at two-second intervals. Sub-second precision. Every injection. Fifteen minutes. The bracelet will hold. The math works if everything goes to plan.)
(When does anything go to plan?)
"How long has he been—" Devod, quiet. Worried. The words arrived from very far away.
"Eight minutes." Leon. Flat. Counting.
(Contingencies. Forgery detection: abort, recalibrate, retry. Maximum three attempts before reservoir drops below threshold. Anchor re-acquisition: Mere's reserve preparation, higher concentration. Bracelet exhaustion: unassisted Flaw Sight, forty percent detection probability, pray the statistics cooperate. Personal reserve collapse: fall back on Mere's field. Trust the principle. Trust the team.)
(Trust the chain.)
A glass of water appeared on the table. I knew it was water because the noise catalogued the reflection pattern and the condensation forming on the outside and discarded it as irrelevant to the execution architecture.
"He needs to stop," Devod said. The concern in his voice was genuine and useless and arrived as texture, not meaning.
"He can't." Leon again. The voice of someone who'd watched this before. "Not until it resolves or he crashes. Let him run. Don't touch him."
(Tomorrow. Mere needs twelve hours for the altered preparation. I need one hour for resonance mapping. The math says yes if the variables cooperate. If the resonance decay is nonlinear in the direction I'm predicting. If the stabilizer's monitoring pattern holds steady. If my precision holds for fifteen sustained minutes. If—)
(If. If. If. The chain has seventeen variables and twelve unknowns and the solution space is a corridor I can almost see the end of, almost, the architecture is building itself in real time and the bracelet's focusing matrix is sharpening the image but the end of the corridor keeps—)