Centaur Longevity Revisited: Reclassification of the Bailey and Malhotra Sample with Modern Orbital Solutions and Astrometric Constraints
Bailey & Malhotra (2009; BM09) assigned dynamical classes to individual Centaurs using 2007 Minor Planet Center orbital elements. They classified 2005 TH173 as quasi-stable (Q) from a 17-day observational arc. We revisit this scheme using modern JPL Small-Body Database elements (2026-07-26), REBOUND WHFast integrations, and diagonal clone ensembles to separate orbital uncertainty from intrinsic chaotic divergence in dynamical lifetimes. For Centaurs with observational arcs >=30 d, the within-object spread in log10 lifetime is 0.54 dex (factor 3.5). This spread is independent of relative semimajor-axis uncertainty, sigma_a/a, across seven orders of magnitude when (a,e,i,q) is fixed (n=65 complete cases; n=96 with a Kaplan-Meier estimator). Semimajor axis and inclination explain the spread, while measurement quality does not. We find no dependence steeper than 0.034 dex per decade in sigma_a/a, limiting the observationally reducible component to 0.24 dex relative to the intrinsic floor. Of 970 catalogued Centaurs with covariance information, 560 fail q>5.2 AU. Of the remaining 410, 355 are already at the dynamical floor, suggesting further astrometry is more likely to secure reliable orbits than improve classifications. Applying the same method to BM09 confirms the classifications. Of 50 comparable designations, 98% of D-class assignments are reproduced; only 2005 TH173 changes from Q to D. None of the 30 multi-opposition 2007 orbits changes classification. Two short-arc objects leave the Centaur region: SN55 reaches q=35.5 AU, while 2002 FY36 exceeds a=30 AU. The rejection of Q for 2005 TH173, now Uranus-crossing (e~0.31) with a median lifetime of 2.99 Myr at 10 Myr integration, is the clearest example. Overall, intrinsic dynamical chaos, rather than astrometric uncertainty, sets the main limit on lifetime-based classification for well-observed Centaurs.