A fixed phase term does not reduce later DM error for 103 selected CHIME bursts of FRB 20180916B

We test a stable phase relation in fitted dispersion measure (DM) for 103 selected events in CHIME/FRB Catalog 2. A time and phase model fits 84 events before 2022 and predicts 19 later events. With equal total weight per activity cycle, later RMS error is 0.439 pc cm^-3, against 0.418 for a time-only model. Prediction skill is -0.104, with a nominal cycle-bootstrap interval from -0.594 to 0.109. Measured burst controls and a harmonic model give no point improvement. A restricted bright sample has a small positive score. A conditional simulation recovers an added DM change of 0.678 pc cm^-3 across the sampled phase range in only 46 to 49 percent of draws. The null result does not exclude modest plasma gradients. We also show why linear time and phase fits give identical predictions for nearby periods while cycle labels stay fixed. No new physical signal is confirmed. All fitted observations come from one telescope. Code, aggregate results, figures, exact source identity and external retrieval steps are included. Source measurements remain external.

astronomy · FRB 20180916B · CHIME · dispersion measure · plasma screen · time validation · null result