The corridor led downward. Ymir didn't need to ask to understand that this wasn't a standard evaluation. Since the age of ten, he had known the routes, the rooms, the procedures. Nothing about this path matched a routine session.
They descended without speaking. The walls became smoother, access more restricted. At this level, there were no students.
The door opened onto a closed analysis room. Not a training hall. Not an open testing area. A space reserved for full examinations.
Ymir stepped in, scanned the installations briefly, then sat down without waiting for instructions. Fixed stations, heavy scanners, real-time analysis consoles. Nothing unnecessary. Nothing instructional.
A technician approached.
— We're going to reassess your profile.
Ymir didn't react, but the choice of words didn't escape him.
Reassess… so they already found a discrepancy.
Sensors were attached quickly: temples, neck, spine, chest. Finer ones along his wrists and jaw. An additional module was fixed at the back of his head.
Deep neural imaging.
Not used for stable low-level profiles.
So this isn't just an adjustment.
The main screen activated. No immediate instructions.
— Structural scan.
Ymir remained still. This part didn't depend on him. The imaging swept through the entire brain: cortex, hippocampus, deep connections, neuronal density, circuit organization.
Data streamed.
— Activity consistent.
— Connectivity within lower norms.
— No additional structure.
A second operator zoomed into a section.
— Nothing on extended motor network.
No Controller structure. Logical. If he had one, it would have been detected years ago.
The scan continued for a few more seconds, then shifted automatically into dynamic analysis.
— Running sequence.
A recording appeared. Partial angle, unstable image. The corridor from the previous night. Three figures. A brief interaction.
Then—
A movement difficult to read.
An incoherent trajectory.
One of the students thrown to the ground.
Pause.
— Reproduce it.
Not a request. A verification.
Ymir looked at the sequence a fraction too long. Not to understand—but to measure what they actually saw.
Incomplete image. Bad angle. They don't have the cause.
He executed.
A simple mental link. One target. Minimal influence. No excess.
The sensors recorded.
— Brain activity stable.
— Ixys fluctuation low.
— Response consistent with Level 1.
Silence.
A technician pulled up yesterday's data next to the current one.
— The spike doesn't match.
Another added:
— Neither in intensity, nor in structure.
Ymir remained still.
Of course. Yesterday wasn't controlled.
The test didn't stop.
— Continue.
Several targets appeared. Not visual. Simulated cognitive signatures.
— Establish a link. Two targets.
Ymir complied. The second connection trembled slightly.
— Three.
He tried. The third collapsed almost immediately.
— Stop.
The data was recorded without unnecessary commentary.
— Multi-target capacity limited.
— Rapid degradation.
— Early cognitive saturation.
Standard profile.
But they were no longer looking for the profile.
They were looking for the deviation.
Another module activated. This time, no simulation.
— Answer the questions.
The tone shifted slightly. Slower.
— Last night, interaction with three students. Conflict initiated by them?
— Yes.
Response time: normal. Pupils stable. Heart rate slightly elevated, but consistent with context.
— Did you use a capability?
— Minimal defense.
Pause.
The screens displayed multiple curves: electrical brain activity, micro-variations in heart rhythm, facial muscle tension, fluctuations in ambient Ixys around him.
— Defense… how?
Ymir let a fraction of a second pass.
— Instinctive.
Silence.
Analysis in progress.
— Instinctive how?
— I reacted.
Same tone. Same rhythm.
Data kept accumulating.
One technician spoke plainly:
— No mentalist pattern matches the observed projection.
— The footage is incomplete, another replied.
— But the force applied isn't.
Ymir kept his gaze steady.
So they have the result. Not the mechanism.
— Did you lose control?
Direct question.
Ymir took a light breath, as if thinking.
— Maybe.
Minimal variation in heart rate. Nothing exploitable alone.
The curves stayed within acceptable margins.
Not confirmation.
Not a clear lie.
Just…
insufficient.
The silence lasted longer this time.
Then the sensors were removed.
— That will be all.
No conclusion was given.
None was needed.
They already had what they wanted:
a measured deviation.
Unexplained.
Ymir stood.
— Continue your standard training.
Same tone as before.
But no one was pretending anymore.
He left the room without rushing. The corridor felt narrower than before—not physically, but in what it implied.
They weren't sure.
But they weren't ignorant anymore.
And from this point on—
they wouldn't be watching his results.
They would be watching his mistakes.
Ymir walked back up at the same pace. His rhythm remained steady, but his awareness had shifted.
Once could pass as an accident.
Twice became a pattern.
And here, patterns were always isolated in the end.