In this paper, we propose, implement, and evaluate a motion-based communication system for field robots: robots which operate in dynamic, unstructured, outdoor environments. We perform two pilot studies in order to guide our development of the system, then evaluate it alongside an audio communication system, an LCD display, and a system of blinking LEDs. We compare the usage of these four systems with three different robots from up to five different viewpoints of interaction in a large study administered via Amazon Mechanical Turk. We contribute in two ways to the development of a more robust form of field human-robot interaction (HRI) wherein robots can select the most appropriate communication vector for a given situation and context. First, we contribute a motion-based communication system for field robots along with three baseline systems against which to test it. Second, we present results from our development of this motion system, showing that it is easier to learn than a baseline blinking LED system, viable for use underwater, aerial, and terrestrial field robots, and less negatively affected by adverse viewpoints than other communication methods.