We report a new series of experiments to explore the phenomenon of low-temperature annealing of fission tracks in apatite that feature a number of improvements over previous work. Grain mounts were pre-irradiated Cf to increase confined track detection and allow briefer thermal neutron irradiation. We co-irradiated and etched four apatite varieties (Durango, Fish Canyon, Renfrew, Tioga) over five time steps equally spaced from 3.66 to 15 ln(s). A length standard was co-etched with all experiments to ensure that subtle differences are within detection limits. Finally, we used a standard etching protocol, allowing the data to be co-modeled with extensive high-temperature data sets and recent analyses of induced tracks that underwent ambient-temperature annealing over year-to-decade time scales. Ambient-temperature annealing occurs at two different rates, with faster annealing at early stages that decreases to a slower rate that converges with empirical fanning linear or curvilinear models. The nature of this decrease varies among the apatite species examined, but no patterns could be determined. The fitted models make geological time-scale predictions consistent with those based on high-temperature data only, and also make predictions consistent with reasonable inferred low-temperature histories for all four apatite varieties. The empirical fanning curvilinear equation encompasses low-temperature annealing at month-to-decade time scales, but low-temperature annealing at shorter time scales may occur by a distinct mechanism. We consider but rule out annealing by radiation from short-lived activated isotopes. We also reconsider the notion of the initial track length, and the appropriate length for normalizing confined track length measurements.