Chapter 67: Upward First

Eli had returned from the depot after the previous watch with rain dried white at the seams of his coat. He had slept for forty minutes in a clinic chair, woken when B-6 began losing ground, and remained only because Mara had declared the recovery test would happen before the scheduled depot comparison.

Now Mara held the timer where the clinic could see it.

B-6 dragged air through the mask. Condensation clouded the plastic with each exhalation, cleared, then returned. His chest drew inward beneath the straps despite the attendants’ adjustments.

At forty-eight seconds, his saturation fell.

Movement around the three uncovered beds stopped. Mara watched the display and timer together. She did not interrupt the count or soften what everyone could already see.

At sixty seconds, B-6 was fighting just as hard for less air.

“Threshold failed.”

Mara lowered the timer and opened the locked tray. Three sealed bypass kits lay inside. She removed one, read its number aloud, and showed the intact wrapper to the clinic witness before breaking the seal.

The snap carried across the room.

She fitted the valve, primed the line, and checked both connections. The kit number and opening time went first onto the bedside chart, then into the public resource ledger. Eli checked the entries from beyond the treatment line. They matched.

B-6’s first assisted breath caught. The next went deeper. Air rushed through the bypass valve, and the strain along his ribs began to ease. By the fourth breath, his shoulders had lowered against the cot.

Mara kept one hand on the line until the saturation reading recovered. Only then did she point from the tray to the uncovered beds.

“Two unopened kits. Three beds without one.” Her eyes settled on Eli. “The depot consumes no additional clinic resource. If that procedure stops being passive, it ends.”

“Understood.”

B-6 had stabilized. The shortage had worsened.

Eli entered both facts in his notebook. Then he left the clinic for the South Branch Flood Depot, arriving twenty-three minutes before the scheduled comparison.

Rain hammered the depot roof hard enough to blur the sounds below it. Water ran down the custody cage, crossed the concrete in shallow threads, and collected around the yellow handling line.

Eli stopped outside that line.

His suspension permitted observation, questions, and a signed statement of limitations. It permitted no handling, certification, custody, breach authority, or creation of test data. He placed his closed notebook on the observer’s shelf and kept both hands clear.

Bell checked his position before beginning.

“You remain an observer.”

“Yes.”

“You proposed the comparison scope. You do not execute or certify it.”

“Understood.”

Inside the controlled area, Bell secured passive contact pickups to CR-4A and CR-4B. The preserved cover puck remained in its witnessed position. Ortiz read each pickup serial number into the coalition record. Lena Voss independently entered the same numbers into the Flood District record.

The handhole seal remained intact. The cage stayed locked. Ortiz retained the coalition key. Voss retained the Flood District key. Neither surrendered independent stop authority.

A designated Flood District handler entered after both authorities verified his authorization. He carried the calibrated striker and nothing else.

Bell shielded the analysis display. Ortiz prepared three control labels by random draw under coalition authority. Voss recorded the assignments under Flood District authority and sealed a duplicate mapping. Both therefore knew the assignments. Bell did not.

“Only Bell is blinded,” Eli said from outside the line.

Ortiz looked over. “Already in the coalition procedure.”

“And in the district procedure,” Voss said. “Our later reviews will be informed reviews.”

Bell waited until both entries had been read aloud before continuing.

The handler received the first instruction card. He applied three measured single taps to the designated control point, allowing the trace to settle after each impact. Ortiz recovered the card before the next was issued. The procedure was repeated at the second and third points.

One set came from CR-4A, one from CR-4B, and one from the exterior of the locked custody cage. Bell, Eli, and the handler did not know which coded sample belonged to which source.

Bell averaged each set without seeing the mapping.

The three references separated cleanly. One produced a longer structural delay and a broad, low-frequency tail. Another reached the cover sooner and carried a narrow secondary metallic peak. The third began with the cage’s split resonance but lacked the collar-to-cover transfer shape.

Before Bell opened the preserved record, the handler surrendered the striker. Voss placed it in a Flood District evidence bag. Ortiz verified the striker serial and seal number in the coalition record, then countersigned across the bag seal without assuming district custody.

“No further control impacts,” Ortiz said.

“Flood District concurs,” Voss said.

Bell retrieved the preserved raw file of the seventh knock. Its hash was read against the Flood District copy, the coalition copy, and the original capture log. All three matched.

The seventh knock had registered physically without a corresponding electrical event. Its collar channels had not supplied a usable lead during the original event, leaving the cover puck’s waveform as the critical preserved record.

Bell loaded that waveform beside the three coded references.

Eli could see the overlays but not the identities. No one was permitted to begin with CR-4B and reason backward toward it.

Bell aligned the traces by first physical arrival, then compared the initial rise, the delay to the cover response, the decay after the main strike, and the position of the narrow metallic resonance.

Coded Reference Two matched.

Its leading-edge shape lay over the seventh knock within the logger’s sampling tolerance. The interval between the primary impulse and the narrow cage-coupled peak also matched. Reference One arrived through a different transfer delay and decayed too broadly. Reference Three shared part of the metallic peak but not the collar-to-cover propagation sequence.

Bell enlarged the tail.

A weaker return appeared behind the narrow peak, above the bounded noise floor but far below the primary impact. The same return occupied the corresponding position in Reference Two. It was too weak to establish a second location, but too repeatable to dismiss as rain or recorder noise.

Bell wrote the finding, signed it, and turned the page so both records could capture the exact language.

“Blinded finding: the preserved seventh knock matches Reference Two in leading-edge form, propagation timing, primary decay, and cage-coupled resonance. A weaker upstream return is detectable in both records but is not localized.”

No one else marked agreement.

Ortiz opened the coalition mapping envelope immediately.

“Reference Two: CR-4B.”

Voss opened the Flood District duplicate. “District mapping confirms Reference Two is CR-4B.”

Only after both mappings agreed did Ortiz approach the display for an informed review. He compared the now-identified CR-4B control against the preserved waveform, checked the transfer interval against the coalition logger copy, and recorded his conclusion.

“Coalition post-unblinding review corroborates Bell’s result. CR-4B carries the matching physical signature. No localization beyond the collar.”