Fairness is essential to ensure long-term satisfaction and engagement of employees, users, and clients—particularly in systems that rely on automated decisions such as scheduling. In this position paper, we argue that fairness in scheduling must be understood as a property over time: treating each scheduling decision in isolation—without considering how often agents have been favored or disadvantaged in the past—can lead to persistent unfairness. We propose a framework for fairness over time in repeated scheduling scenarios that captures individual utilities across a history of decisions and can be adapted to multiple applications and fairness concepts. We hope this framework will encourage further research and will serve as a foundation for both theoretical exploration and the development of practical scheduling systems that promote fairness over time.