A deliberate desk reset On a clear, level desk, place the paired tread flats crown-up, roll the rim about one palm length, release it completely, wait roughly two seconds, and view one face. Roll again to reset. Keep hands, hair, cords, desk edges, pets, and electronics clear; physical rolling and settling remain unproved.

One shell, one free rotor The front and rear treads are digitally phase-locked into one shell; the captive inner rotor is the only intended free rotary degree of freedom. A fresh replay cleared its nominal 360° sweep at 72 samples while rigid-body checks blocked shell-relative rotation, rear withdrawal, and axial rotor escape.

A reveal made from angles At the modeled home orientation, five apexes register and the authored optical audit reports a 0.8531 visible fraction on both faces and four affine oblique proxies. Five apexes also register at 2° and 4°, then drop to one at 6°, so the honest description is angle-and-settle, not stationary-only sensing.

Three parts, no hardware The baseline design contains a front drive ring, rear lock ring, and gravity-lag reveal rotor—no screws, pins, bearings, magnets, adhesive, batteries, or purchased components. All three sealed STLs pass the replayed digital bed, mesh, wall, and bounded support-face gates; no successful slice or print is claimed.

Measured, then bounded Digital evidence records 0.60 mm nominal journal diametral clearance, 1.20 mm rotor axial clearance, 0.20 mm nominal shell endplay, supported four-finger snap geometry, and a 61.43 g solid-volume estimate. Snap force, printed fit, rolling feel, wear, durability, safety, recognition, and delight remain unknown.