This paper addresses the legendary total solar eclipse of 29 May 1919. Two British teams confirmed the light bending prediction by Albert Einstein: Charles R. Davidson and Andrew C. C. Crommelin in Sobral, Brazil and Arthur S. Eddington and Edwin T. Cottingham on the African island of Príncipe, then part of the Portuguese empire.By jointly analyzing the two astronomical expeditions supported by written and visual sources, I show how, despite extensive scholarship on this famous historical episode and the historiographical emphasis (...) on the plural dimensions of knowledge construction, many human and non-human actors have been kept in the shadow of the eclipse. I do so by focusing on what I call knowledge from the periphery together with knowledge from below, grounded literally on how localities (sites) affect choices and events, and growing outward to encompass a wide range of participants. I show how the geopolitical status of the two nations where the observational sites were located, and specifically Portugal's condition of colonial power, affected main decisions and events, while highlighting the active role of participants, ranging from experts from the peripheries and those involved in the travels to local elites and anonymous peoples, some of whom contributed to the observation of totality. (shrink)