GJ 3323 extraction differences retain 4.93 m/s weighted scatter after 13 large differences are removed

We test whether a few exposures cause the large difference between two public HARPS-RVBank version 02 velocity columns for GJ 3323. We use 121 zero-flag exposures in two intensive intervals before the fiber change. The difference between RV_mlc and RV_mlc_j has weighted RMS scatter of 5.76 m/s. Removal of the 13 largest absolute centered differences leaves 4.93 m/s. The largest 13 weighted squared contributions account for 37.5 percent of the total. Thus, a small number of exposures does not account for this product difference under these rules. Fixed two-period fits give difference amplitudes of 1.25 m/s at 5.3636 days and 1.91 m/s at 40.54 days. Block-sign tests do not resolve either wave at a two-test 5 percent threshold. The products share spectra. We do not identify a physical cause, confirm a planet, or set a planet exclusion limit. The package contains the exact input, source hash, code, all controls, a figure, and offline steps.

astronomy · GJ 3323 · HARPS · radial velocity · extraction sensitivity · paired differences · null result