Vigil
Nadia monitored variable stars. Not the dramatic ones — novae, supernovae, the brief catastrophes that make careers. The quiet ones. Stars that brightened and dimmed over days, and her job was not to explain the pulsing but to record it.
The telescope did the looking. The photometer did the counting. She did what neither instrument could: she decided when the data stopped being data.
Every clear night she tracked V Cassiopeiae through its 14.2-day cycle. The star swelled and contracted — a breath visible only in numbers. Three years of watching. The period had not changed. The amplitude had not changed. What changed was everything around it: atmospheric seeing, cloud cover, moonlight, the Earth's rotation swinging Cassiopeia from zenith to horizon and back. The signal was constant. The noise was the world.
Dawn was the problem. Not because it arrived — it always arrived — but because it arrived gradually. The sky didn't switch. It brightened at exactly the rate that matched the precision she needed. By the time the photometer was useless, it had been unreliable for twenty minutes. There was no boundary. Just a region where the star's light and the sky's light became the same thing.
She wrote the same line in her log every morning. "Observation terminated. Signal-to-noise below threshold." The star was still there. The photometer was still counting. What had ended was the difference between them.
Kauffman had suggested automating the cutoff — software to track the ratio and close the shutter when it dropped below three sigma. She tried it. The algorithm couldn't distinguish the variable dimming from the sky brightening. Both looked like signal loss. One was the thing she was trying to detect. The other was the thing ending the detection. The numbers were the same. The situation wasn't.
She couldn't put "situation" in a methods section. What she wrote instead was: "manual termination based on observer judgment." The reviewers didn't question it. They assumed judgment meant criteria. It didn't. It meant her hands knew before her eyes confirmed — a recognition that the same photons arriving at the same rate had stopped being information.
Tomorrow the star would return. Same brightness, same period, same slow pulse. The sky would darken and the signal would separate from the noise and she would open the shutter and it would look like she had never stopped. But she had stopped. Every morning she closed the shutter, and every evening she chose again. The continuity was not in the looking. It was in the returning.