In this paper, I show that, by gathering evidence from the past literature and by presenting new evidence, subjects can undergo scrambling in Japanese, contrary to Saito’s (1985) Ban on Scrambling of Subject (BOSS), which has been a classic and wide-spread claim. In so doing, I argue that Japanese scrambling in general is subject to a version of minimality/superiority effect, a Feature-based Minimality Condition (FMC). I also discuss that apparent difference between the applicability of FMC in Japanese and the inapplicability of FMC in English is due to Feature-Splitting Parameter.