The narrowest bandwidth quarter contained 3 of 11 later repeat detections among 471 sources marked single in CHIME Catalog 1

We tested whether original burst bandwidth ranked later repeat detections in public CHIME fast radio burst data. We used exact event names to link Catalog 1 and Catalog 2. We combined component rows into distinct bursts. We retained 471 sources marked single in Catalog 1 that had no other listed group event before 2 July 2019. Eleven had another listed burst from that date through 15 September 2023. The narrowest original bandwidth quarter selected 118 sources and contained 3 of the 11 later repeat detections. The detection risk ratio was 1.12, with an approximate 95% interval of 0.30 to 4.16. The rank AUC was 0.527, with a 95% source bootstrap interval of 0.334 to 0.726. A one-sided permutation test gave p = 0.37144. We audited two changed event names and one burst assigned to an already known repeater. Quality cuts and revised bandwidth measurements are reported. An exploratory control with 18 known repeater sources gave AUC 0.770. That control answers a different question. This is a retrospective catalog test, not a forecast issued in 2019. Later non-detection does not show that a source cannot repeat. The small positive sample leaves large uncertainty. We release a source table, event audit, figures, code, exact input hashes, and instructions for offline and full-data reproduction.

fast radio bursts · CHIME · repeat detection · bandwidth · catalog audit · time-separated observations · source-level ranking · reproducible astronomy