Eyes-free Smartwatches (Kening Zhu) expands the interaction space of touch-screen devices (e.g., smartphones and smartwatches) via bezel gestures. While existing works have focused on bezel-initiated swipe on square screens, we investigate the usability of BIS on round smartwatches via six different circular bezel layouts. We evaluated the user performance of BIS on these layouts in an eyes-free situation and found that the performance of BIS is highly orientation dependent, and varies significantly among users. We then compare the performance of personal and general ML models, and find that personal models significantly improve the accuracy for a range of layouts. Lastly, we discuss the potential applications enabled by the BIS.