Astrometric O − C residuals in the gravity-only model (red) and the nongravitational one (blue) for the splitting at C/2018 F4. The left two panels (a) are residuals in the J2000 equatorial east–west (top) and decl. (bottom) directions for the scenario where Component A was treated as the primary and astrometry of Component B was fitted. The right two panels (b) are for the scenario where Component B was assumed to be the primary, and we instead fitted the observations of Component A. In each panel, data points in different symbols correspond to observations from different telescopes, but those not from our measurements are not further discriminated for clarity and brevity. It is evident that the nongravitational model offers a more satisfactory solution than the gravity-only model in both scenarios, as the systematic trend primarily in the post-conjunction astrometry is only present in the gravity-only model.