The incident form began with a blank labeled PREVENTABLE CAUSE.
Marco stared at it while a nurse in ADR's surface trauma clinic put the seventh stitch through his calf.
The examination tablet rested on a rolling tray beside his knee. Below the blank waited twelve checkboxes: equipment failure, procedural deviation, unreported condition, worker impairment, creature migration, environmental change, third-party interference, and other.
He could select more than one.
He could not select the mountain did several things at once while people made decisions with incomplete information.
“Hold still,” the nurse said.
“I am.”
“Your foot disagrees.”
Marco relaxed his ankle. The stitch tightened.
His shoulder scan showed a moderate strain without a tear. The medical order prohibited lifting above fifteen pounds, overhead movement, underground duty, and bow use until review. Workers' compensation covered treatment because the injury occurred on shift within assignment. Paid recovery required him to attend every appointment and accept restricted work if Ridgeway offered it.
His Foundation had barely moved since the incident. Pain alone had taught the System nothing useful.
The workers' compensation nurse separated medical capability from employment authority on three different forms.
Current mobility: limited by wound and pain.
Current lifting: below occupational baseline due to shoulder restriction.
System Rank: G.
Delver license: RESTRICTED — MEDICAL.
“The Rank doesn't change because I got hurt?” Marco asked.
“Because it reflects capability you've demonstrated over time, not what an injured leg can do this morning,” she said. “The System is not your doctor. I can't change your Rank. I can restrict your license until the tissue heals.”
Ray, listening from across the curtain, said, “Three forms beat one letter again.”
The nurse handed Marco a fourth form for agreeing.
Across the clinic, Ray waited for transport to surgery with his right leg stabilized in a vacuum splint. The fracture had not broken skin. That was the best version of a bad result.
“You fill the magic box?” Ray asked.
“Not yet.”
“Good. Means you haven't lied quickly.”
“What did you choose?”
“They took mine away after I wrote all of the above except impairment.”
The nurse pointed at Marco's calf. “He also tried to walk on that leg.”
“Professional opinion,” Ray said.
“Rank F does not include orthopedic diagnosis.”
“Government oversight.”
The ambulance team arrived for him. Before they moved the stretcher, Ray looked at Marco.
“Write what happened before anybody tells you what would be convenient to remember.”
Marco selected equipment failure, unreported condition, creature migration, and environmental change. Under procedural deviation, he selected unknown pending review.
He left preventable cause blank.
The first interview lasted three hours.
ADR provided the room, coffee, and two investigators. The county provided a safety observer. Ridgeway provided Dana and a lawyer whose main contribution was asking everyone to finish questions before answering them.
Ridgeway's issued bow, work jacket, helmet, alarm, route cards, and remaining company arrows sat in evidence bags along one wall. The bent rest had been photographed from four angles. Dust from Marco's boots had its own chain-of-custody label.
Investigator Cho began at the previous morning.
“Why did you call Ray Mercer to the track?”
“It exceeded the profile of the animals assigned to our route.”
“Did you identify a blindjaw?”
“No.”
“Did Mr. Mercer?”
“No. He said it wasn't a crawler that belonged on our side of Zone Two.”
“The report says possible large unknown creature.”
“Possible.”
“Did you request a full Zone 1 shutdown?”
“Ray requested the whole North Spur. Shaw closed North Seven and the service branch. Operations kept Five and Six under amber precautions.”
“Did you believe that was unsafe?”
Marco considered the question. “I believed something large had crossed Seven. The sensors showed no active threat. The patrol found no creature. I didn't have the authority or evidence to grade the whole spur.”
“That wasn't my question.”
“Then yes. I felt unsafe.”
The lawyer made no sound. Dana watched him without expression.
Cho asked about the absent beetles, the pulsing seams, the barrier indicator, the order to hold at North Three, the collapse, and the route Marco chose with Wes. Every fact became smaller when separated from the others.
“You engaged a crawler while carrying an injured worker.”
“We set him down behind the pipe curb before I drew.”
“Why engage instead of waiting?”
“The crawler blocked the shelter approach. Something heavier was moving behind us. We used a cart to change its line and fought only after it committed.”
“You did not kill it.”
“No.”
“Your incident note says successful redirection.”
“It left the approach. The purpose was getting Wes to medical.”
Cho moved to the choke.
“Who proposed rupturing the blindjaw's second scent structure?”
“I described the anatomy and uncertainty. The patrol lead authorized the shot.”
“Did Salvage identify a weak point?”
“It drew my attention to damaged recoverable structure. I knew the organ from surface post-processing work. Salvage did not tell me what a live matriarch would do.”
“But your shot redirected it.”
“The shot, bait, maintenance branch, carts, patrol shield, Ray's choke plan, and the animal's existing injury redirected it. The response delver killed it.”
The investigator paused over her notes.
There were heroic versions of the incident available. Marco could have made himself the only person who noticed anything, the archer who found the impossible target, the low-rank worker who saved the shelter.
Those versions would be easier to repeat and less true each time.
The interview ended after dark.
Before releasing him, the county observer asked Marco to mark confidence beside each statement.
Track measurement: high.
Beetle absence: high, with no complete population count.
Seam pulse spacing: high for the measured section.
Pulse cause: unknown.
Blindjaw response to scent damage: moderate, single event with multiple simultaneous stimuli.
Marco appreciated the column. It let facts remain useful without forcing certainty onto them.
His cleanest action—the casualty carry—required the least theory. His most interesting observation—the repeated seams—required the most restraint.
Sarah waited in the lobby with Emma. Lucy had been left with a neighbor because the clinic had already given her enough images to invent.
Sarah stood when Marco entered.
She hit his uninjured arm once with the flat of her hand.
Then she held him carefully around the shoulders.
“You were supposed to come home after the amber light,” she said into his jacket.
“That was the plan.”
“Bad plan.”
“The plan was fine. The barrier disagreed.”
She let go and looked at the dressing visible below his shorts. “Ray?”
“Surgery tonight. They expect him to keep the leg.”
Sarah closed her eyes.
Emma carried his evidence-release folder. “Did you level?”
“Not then.”
She looked ashamed of the question.
Marco touched the top of her head. “I checked too.”
The second interview reconstructed the barrier.
Maintenance logs showed brief sub-alert fluctuations for eleven days. Each remained inside green tolerance. A sensor calibration test had been scheduled for the afternoon after Marco found the track. The barrier field failed when a localized stone shear severed two buried conductors at once.
No single reading had required a shutdown under the old procedure.
Together, they described a condition the procedure did not name.
The state hazard model had rated North logistics for paired Rank G support under a licensed F-rank route lead, with Rank E patrol response inside fifteen minutes. Those assumptions held until the barrier and main route failed together. Afterward, the same workers faced an evacuation problem above the published band.
Investigators did not rewrite the original hazard grade to E merely because an E-grade matriarch appeared. They marked the affected route UNRATED — SUSPENDED while engineers rebuilt the clear model. Hazard grade described expected repeatable risk, not the worst thing hindsight could find.
That precision mattered to Marco. If every rare breach permanently raised a site one letter, authorities would eventually produce labels too conservative to distinguish anything. If rare breaches never changed assumptions, the labels became advertising.
ADR's operations controller had kept North Five and Six active because the service branch was sealed, patrol coverage had doubled, sensors remained green, and the economic loss of a full closure was estimated at $186,000 per shift. The decision complied with policy.
The county observer wrote that down.
So did the fact that policy had not accounted for prey displacement or repeated sub-alert pulses.
Marco did not need the controller to be evil for the choice to hurt people. The company had priced evidence. The price had been wrong.
On the fourth day, Nia asked him to visit ADR's sealed evidence laboratory on the surface campus.
“They want target confirmation,” she said. “You don't touch anything.”
The blindjaw matriarch occupied three stainless tables pushed together. Its head had been separated behind the response strike. Clear covers enclosed the scent structures.
An ADR anatomist showed Marco the arrow channel through the right fan.
“This one?”
“Yes.”
“And the left was already ruptured?”
“Yes.”
The anatomist rotated a scan. Both organs divided into repeating chambers around a pale central lattice. Each chamber was nearly identical in volume. Natural organs repeated approximate structures. These looked measured.
Salvage pressed at Marco's attention.
“The lattice continues into the jaw plate,” he said.
The anatomist zoomed farther. Fine silver-brown lines crossed from soft tissue into mineralized bone at regular intervals.
A geologist standing behind the display leaned closer.
“That's contamination from the collapse.”
“Inside both organs?” Nia asked.
The geologist requested a cut sample. The anatomist removed a section under the cover and placed it in a scanner.
The first result listed calcium, mana-bound silica, iron, and ordinary blindjaw tissue.
The second result added a line Marco recognized.
SYSTEM RESIDUE TRACE — UNCLASSIFIED.
“Background bucket,” the geologist said.
“Dry Creek used the same words,” Marco said.
Everyone looked at him.
He explained the loose gray fragment and the trace in the exhausted mineral seam. The geologist asked for dates, route numbers, and assay provider. Marco had all three in his notebook.
She pulled up the North Spur survey map. The stone shear that cut the barrier conductors ran along one of the repeated seams Marco had logged the day before the incident.
“That seam shouldn't be there,” she said.
“Because the survey missed it?” Marco asked.
“Because this cut was exhausted and resin-stabilized seven years ago. The base rock on the original scan is continuous.”
She ran the sample again.
The residue line remained.
The geologist searched ADR's internal catalog. Hundreds of entries used related unknown-residue labels. Most came from exhausted G- and F-grade sites, old barrier foundations, processed monster waste, and survey instruments operating near mature dungeon stone.
“So it's common,” Nia said.
“Common at trace scale,” the geologist replied. “And inconsistent. Different composition, different spectra, sometimes nothing but an interface tag. Research time follows hazard and value. Trace anomalies from low-output sites lose both competitions.”
Marco thought of the one gray flake in a spice jar.
“Has anyone collected them across sites?”
“Probably. Graduate students collect everything too.” She did not mean it as an insult. “No published aggregate has shown a stable use.”
One fragment was meaningless. A hundred might still be a hundred versions of instrument noise. Marco had no reason to make it larger except that the same broad label kept appearing near patterns other people had little economic reason to study.
He wrote down the catalog reference before access changed.
The fragment in his spice jar remained worth nothing.
A lab supervisor entered, read the display, and changed the sample status from incident evidence to restricted anomaly review. The clear cover locked with a soft click.
“Is that because it's dangerous?” Marco asked.
“Because it is unknown and connected to a reportable barrier failure,” the supervisor said. “Unknown is enough.”
The explanation was reasonable.
The sample still disappeared through a door his badge could not open.
Two weeks after the incident, evidence control released Ridgeway's equipment. The company sent the bent work bow for repair, returned the jacket to its gear cage after decontamination, and logged nine surviving work arrows. Marco's thirteen personal dungeon arrows had remained locked at home throughout the shift.
Ray's surgery succeeded. Wes went home with a concussion protocol and a fractured collarbone. The other injured workers returned to light duty.
The incident findings added six verified observations to Marco's field record: evacuation navigation, casualty extraction, team communication, controlled ranged support, threat reporting, and retreat discipline.
They did not change his Rank or lift his medical restriction. They would remain part of the record if he later sought broader professional work.
Each observation named the task, verifier, conditions, and limits. A violent hour had become useful evidence, not a shortcut around healing or judgment.
ADR issued a $750 safety commendation to every worker whose documented action materially supported the evacuation. The letter stated twice that accepting the payment did not waive claims.
Emma checked before depositing it.
The unassigned point still waited that evening after Marco finished the shorter debrief and listened to the recorded route call twice.
Will: 10 → 11. Unassigned Attribute Points: 0. Integration estimate: 24–72 hours.
He did not feel braver. His calf itched beneath the dressing, and his shoulder stopped six inches short of full reach. After integration, Will eleven might give him more room between panic and action. It could not choose correctly for him.
His System Rank remained G. The portal now displayed DELVER LICENSE: RESTRICTED — MEDICAL / VERIFIED FIELD RECORD UPDATED.
The mismatch had a form, requirements, and a board. The residue had no verified hazard, repeatable use, market, or procedure.
In the locked laboratory, the sample carried no rank at all.