Chapter 3: Seven Points and a Sidewalk

## Chapter 3: Plans Are Not Remediation

At 4:19 a.m., Dev set the control probe on dry pavement three meters west of public cleanout E-1.

Mara watched from the laundromat doorway with the maintenance binder braced against her ribs. Gray predawn light lay across the sidewalk, turning old gum and salt stains into a municipal topographic survey. The cleanout cap occupied a square of patched pavement near the curb. Nothing about it looked capable of swallowing laundry, violating geometry, or improving anyone’s week.

“Control first,” Mara said.

“I am placing the control first.”

“You were narrating the cleanout.”

“I can narrate one thing while placing another.”

“That has not been established.”

Dev aimed his phone from a compact tripod so the control point, cleanout, instrument display, and a radio-synchronized clock application remained visible. He photographed the calibration labels, read the serial numbers aloud, and opened a raw log at 4:19:38. Only then did he place the second probe near E-1 without touching the cap or crossing the chalk boundary Mara had marked around its seam.

The first readings were ordinary enough to be useful. Pavement temperature shifted with the damp air. The control followed it. E-1 did not.

By 4:24, the cleanout remained 7.7 degrees colder than the control.

Dev changed probe positions, repeated the calibration check, and replaced the E-1 probe at the same marked distance from the cap. The separation returned: 7.7 degrees.

Mara inspected his first photograph through the open voice channel. “Rejected.”

“The display is readable.”

“The seam is in shadow.”

“It is physically present in the image.”

“So is half the sidewalk. Try documentation.”

Dev moved the reflector rather than the evidence. His second image showed the display, time reference, cap seam, and pavement marks in one frame. Mara saved the uncompressed original and logged its filename.

She then crossed to the exterior threshold with her handheld thermometer. Cold had stiffened her burned palm beneath its bandage, and the required angle forced her strained right wrist backward. Her first reading drifted. On the second, the sensor lifted when her grip loosened.

Dev reached toward the instrument, then apparently remembered that he valued his fingers.

Mara reset the reference point and tried again. Her results varied more than his calibrated probe, but every usable reading showed the same cold separation within the handheld unit’s tolerance. She documented the limitation instead of improving the sentence.

Independent handheld comparison supports sustained anomaly. Handheld readings less stable due to operator grip and positioning. Calibrated station preferred.

At 4:27, Dev’s contact sensor registered a vibration.

The display produced a narrow peak at 18.6 Hz, held for three seconds, then flattened. No truck passed. The laundromat equipment had no power, and the control sensor on the pavement showed no matching event.

“Repeat contact,” Mara said.

Dev removed the sensor, photographed the contact surface, and replaced it on the same marked point. The peak returned at 18.6 Hz. A third placement produced 18.5 Hz.

He moved the sensor to the control point. Nothing rose above background.

Mara took the handheld contact meter from its case. Her wrist objected as she pressed the pickup against the pavement, but two separate placements gave dominant peaks between 18.4 and 18.8 Hz. Each lasted just over three seconds. Neither appeared at the control point.

“Reproducible vibration,” Dev said. “Source unknown.”

“Good. Keep both halves.”

He entered exactly that.

A hair-thin black line occupied one section of the cap seam. It vanished when the reflector angle changed and reappeared when the light ran nearly parallel to the metal. Beside it lay a straight blue fiber, too narrow to identify and too rigid-looking to resemble the curled lint Mara swept from under the folding tables every night.

Dev magnified the image. “Visually consistent with the fibers behind Washer Fourteen.”

“Not identified.”

“Visually consistent, not identified.”

“Put it in that order. People get excited and stop reading after nouns.”

No one opened the cap. No one sampled the line or touched the fiber. Proximity established location, not origin, transport, or composition.

At 4:31:12, E-1 dropped another 2.1 degrees below its established cold baseline.

The sharper pulse lasted eighteen seconds. The control probe moved by less than a tenth of a degree. Dev marked the event without naming it, then checked both probes and left the station untouched.

Waiting became part of the procedure.

Mara used the interval to compare the three maintenance revisions inside. Each showed the laundromat drain crossing the front wall and reaching E-1. None showed a branch east of the cleanout. The oldest copy had faded nearly to illegibility, but its line continued along the mapped main. The later revisions changed symbols and lettering, not direction.

Dev traced the pavement in measured increments. A cold band extended from E-1 toward the east, angled away from the documented route. He repeated the line from the opposite direction and obtained the same boundary within the width of one pavement joint.

“Surface condition remains a possible explanation,” he said.

“Record the angle before the explanation eats it.”

“It cannot eat—”

“Dev.”

He read the measurements aloud. Mara entered them separately. Their records matched.

The second sharp pulse arrived at 4:43:04.

It repeated the same 2.1-degree drop within instrument tolerance and lasted nineteen seconds. The interval between pulse onsets was eleven minutes, fifty-two seconds.

Mara opened the earlier incident log. The unexplained machine events associated with the missing loads had clustered at approximately twelve-minute cycle intervals. The resemblance was close enough to require monitoring and nowhere near sufficient to establish a mechanism.

Dev looked from the graph to the incident log. For once, he shortened the sentence himself.

“Timing correlation. Causation unproved.”