In the humanities, histories of technology generally examine objects through representations, such as patents, illustrations, fictions, and newspapers. With objects flattened on the page, this approach rarely attends to how technologies worked in context and over time. In response, we prototyped a series of early technologies, which cannot be immediately accessed in museums or archives; they have been lost, destroyed, or broken, yet they deeply inform present day devices. In this talk, we present three of our prototypes: an early wearable technology, an early wire recorder, and an early reading machine. The early wearable prototype features a skull cravat-pin (designed by Gustave Trouvé and Auguste-Germain Cadet Picard in 1867), which was said to snap its jaw and roll its eyes. We replicated the mechanism of the skull using electromagnetic technologies available in the nineteenth century. We found that the size of the skull, as described in historical records, could not have contained the electromagnet necessary to propel the skull’s jaw. This suggests that textual representations of the stick-pin were distorted by Victorian hyperbole. Elsewhere, our wire recorder prototype recreates a precursor to data storage: Valdemar Poulsen’s 1898 experiments with sound, whereby people yelled into a transmitter while running a trolley along a wire. Despite its claims for high fidelity, our findings suggest this process was highly contingent on the bodies and movements of users. Finally, the reading machine prototype explores the maintenance of Fournier d’Albe’s 1913 optophone, which translated images into sound, by one of its blind users, Mary Jameson. Early results of this research stress that Jameson contributed significantly to the optophone’s development, even if she is never mentioned in patents and rarely in other textual accounts. Together, these prototypes highlight why trial-and-error experimentation (as a method for humanities research) matters for historical inquiry into technology off the page.