Mobile gaze interaction is challenged by inherent motor noise. We examined the gaze tracking accuracy and precision of twelve subjects wearing a gaze tracker on their wrist while standing and walking. Results suggest that it will be possible to detect whether people are glancing the watch, but not where on the screen they are looking. To counter the motor noise we present a word-by-word textual UI that shows temporary command options to be executed by gaze-strokes. Twenty-seven participants conducted a simulated smartwatch task and were able to reliably perform commands that would adjust the speed of word presentation or make regressions. We discuss future design and usage options for a textual smartwatch gaze interface.