A chronological width term gives no clear prediction gain in 296 selected CHIME events

We test whether later fitted components are wider after adjustment for peak frequency in CHIME/FRB Catalog 2, release 20251231. In 296 selected events from 170 source labels, a chronological offset reduces source-excluded prediction error by 2.23 percent. A 95 percent source-resample interval is -2.48 to +8.43 percent. Resampling repeats model training and excludes every copy of a test source. The full-sample offset is 0.299 in log width ratio, with interval 0.064 to 0.540. Requiring both widths to exceed one time sample gives offset 0.038, with interval -0.227 to 0.323. This cut also changes source composition. Repeaters and apparent one-offs have opposite offset signs. Every tested prediction-gain interval includes zero. The result gives no clear predictive support for a chronological width term under the tested model. Shared fit errors, missing peak errors, resolution, selection and source mixture limit physical interpretation. No independent telescope confirms this offset. No new source, planet or emission mechanism is confirmed. Code, frozen rows, provenance, validation, tables and a figure are included.

astronomy · CHIME · fast radio bursts · component width · chronological asymmetry · source exclusion · null result