This chapter examines a well-known detection process in radar signal processing known as the log-t detector. This was introduced into the signal processing literature by G. B. Goldstein in 1973. The innovation of Goldstein's analysis is the realisation that the detector based upon the statistic can be extended to produce a constant false alarm rate (CFAR) decision rule for a large class of underlying clutter models. The chapter re-examines Goldstein's log-t detector, demonstrating why it is also CFAR when applied to Pareto Type I distributed clutter. A new version of it, based upon order statistics was presented, which was shown to also possess the CFAR property with respect to a large class of clutter models. The advantage of the log-t OS detector was that it could manage interfering targets better than a conventional log-t detector. In homogeneous clutter, and during clutter transitions, the log-t OS detector tended to perform no better than the OS-CFAR.