A five-error cut leaves J1646-4545 with the largest apparent braking index in a selected ATNF 2.9.0 sample

We screen 438 ATNF 2.9.0 entries with direct frequency terms, periods of at least 30 ms, negative first derivatives, no binary model tag and no globular cluster tag. Four entries above J1646-4545 in absolute apparent braking index fail a strict five-error cut. J1646-4545 leads the 234 retained entries with n = 3.13 million and a six-error second derivative. It also leads the common-reference subset. A ten-error cut removes it. A steady dipole with the observed first derivative predicts a second derivative about one million times smaller. This conditional test does not identify the cause of the known curvature. A synthetic uniform-span test gives a 4.821 ms cubic timing difference after lower-order fitting. It is not a measured residual or significance. Source-table numbers were checked only in search text because PDF retrieval failed. No raw arrival times or independent confirmation are supplied. Code, frozen records, hashes, sensitivity checks and offline steps are included.

astronomy · pulsars · ATNF · braking index · timing noise · J1646-4545 · null discovery result