From H.G. Wells' The Time Machine to the recent Hollywood blockbuster Arrival, the notion of time-travel is a firmly established narrative trope. Yet tales of travel back and forth through time are essentially absent before the mid-1800s. This invites the question: How do people make sense of time-travel, and how does it build on the more basic building-blocks of our conception of time itself? Here, we investigate lay conceptions of time-travel using a gesture elicitation paradigm. Participants watched brief videos of time-travel stories and then recounted the plots. Combining qualitative analysis and machine learning extraction of co-speech gesture trajectories, we describe how participants' construals of time-travel cobble together more basic spatial construals time (e.g., length-duration; past-left vs. future-right), combined to create layered, ad-hoc, flexible representations of time. We discuss implications for how spatial metaphor can offer a foundation for more complex, elaborated forms of reasoning and understanding.