The two-hour reading on position thirteen came back lower than Gorrak’s notation book’s printed minimum.

For eleven seconds, Zul’jinn allowed himself to consider that the trial had already failed.

Then Gorrak said, "It is moving. Slowly, but the gauge moved between my first reading and my second, thirty seconds apart."

"A failing bond does not move in the direction this one moved."

Zul’jinn took the gauge himself and confirmed it. The reading was below threshold but climbing.

The rate of climb was shallow compared to the void-compound bonds at the same elapsed time, but it was present and consistent. Vorra had been right that the manganese binder cured slowly. She had not said it did not cure.

He sent the two-hour reading south with a single line: below threshold, climbing, consistent rate. Full six-hour reading to follow.

The six hours passed with the rest of cycle two proceeding around position thirteen as though nothing unusual were happening there.

Nothing about the rest of the arc depended on what position thirteen’s reading turned out to be. Positions fourteen through seventeen went in on standard compound, on schedule, and Gorrak’s verification found all four within the acceptable range without incident.

Arka’garr came by midmorning, on his way between two of his own patrol checkpoints, and stood at the edge of the arc for a few minutes without interrupting anything.

"How is it going," he said, when Zul’jinn noticed him.

"Slower than I would like, faster than the substrate allows." Zul’jinn did not stop what he was doing. "Position thirteen is running a substitute compound. If it holds, it changes how much of the arc I can finish with what I have left."

"And if it does not hold?"

"Then I redo it with what little margin I had, and I am exactly where I was yesterday, except a cycle behind." Zul’jinn glanced at him. "I will know by this evening."

Arka’garr nodded and did not press further. He had learned over the past weeks that Zul’jinn’s silences during a calculation were not invitations, and his answers, when they came, were complete enough not to need follow-up questions.

He went on to his next checkpoint.

At the six-hour mark, Gorrak took the final reading on position thirteen.

It was within the acceptable range. At the very bottom of it, lower than any reading Gorrak had recorded across the entire arc so far, including the redone position eleven from cycle one.

But within range, which meant the manganese binder produced a usable bond under field conditions, with a result that had no margin above the minimum at all.

"No margin," Zul’jinn repeated, when Gorrak gave him the number. "Which means every manganese application has to be technique-perfect."

"Position eleven’s redo happened because of a pace error at the socket’s curved interior, and that error still left the bond inside the acceptable range with room to spare. A manganese bond with the same pace error would not have room to spare."

"It would fail," Gorrak said.

"It would fail, and it would fail in a way that looks identical to a correctly-applied bond until the gauge is run on it."

"Which means the failure would not be caught until verification, which is too late to redo within the same cycle." Zul’jinn was already walking toward the materials shed. "I need to know which of the eighteen workers apply compound with technique consistent enough that I would trust them with no margin for error."

He and Gorrak worked through the roster together.

Of the eighteen installation workers across the three crews, four had technique Gorrak rated as consistently precise across every application he had verified so far. This included two from Mekka’s reallocated sixteen, who had been on the job for only a few days and had not yet produced a single deviation.

"Four people," Zul’jinn said. "Four manganese-eligible positions, if I assign all four to manganese work and keep them together so I am not splitting a precise team across two compound types in the same day."

He went back to his re-survey of the eleven remaining positions and walked the arc a second time with the criticality notations in hand.

This time he sorted not by criticality alone, but by which positions could absorb a no-margin bond if everything went correctly, and which positions needed the cushion standard compound provided even at the cost of substrate.

Of the eleven remaining, four were positions where the adjacent completed installations and the wall’s own mass gave enough structural redundancy that a bottom-of-range bond would not represent a meaningful reduction in the arc’s overall stability.

Position thirteen was already one of them. He marked three more: two near the eastern flanking tower, one at the arc’s midpoint where two earlier high-density positions sat on either side.

The remaining seven were positions where he was not willing to accept a no-margin bond under any circumstance, including the position closest to the northern gate’s own structure, where the load path ran through the gate mechanism itself.

"Four manganese, seven standard," he said, writing the final allocation. "Four positions on manganese substrate, applied only by the four precision-rated workers, working as a single dedicated team."

"Seven positions on standard compound, the remaining substrate, applied by the other crews as planned."

He calculated the substrate draw under the new allocation. Seven standard-density positions against nine positions’ worth of remaining substrate left two positions of margin.

It was not generous, but it was margin where five minutes ago there had been a deficit.

He wrote the revised consumption projection and sent it to Vorra with a note: this is what the shared pool needs from the Yohan side under the new allocation. If your corridor treatment’s draw is final, this should clear with two positions of substrate to spare on my end. Confirm what that leaves for the Arch, because I do not want to assume my margin is also your margin.

He sent it south and went to find the four precision-rated workers.

He told them what they were now responsible for in the same way he told anyone anything that mattered: plainly, without softening it and without making it sound larger than it was.

"Four positions. Manganese binder, not the standard compound. The bond it produces meets specification, but at the bottom of the range, with nothing above it."

He looked at each of them in turn. "Every application from this point has to match what Gorrak verified on position thirteen exactly."

"Not close. Exactly."

"If any one of you is unsure whether your hands can hold that line for four positions in a row, tell me now, and I put someone else in your place and think no less of you for it."

None of the four said anything.

"Good," Zul’jinn said. "Then let us go look at position fourteen, because that is where you start, and I want to walk you through it once before you do it yourselves."