It has long been common to conceive of observation mechanically—as a process of data registration that is marginal and automatic. This paper explores the ways in which this idea has been facilitated by conceptions of instrumental technology as hindering individual variation in the practice of scientific observation. Understanding instruments in this way parallels attitudes towards technology in industrial manufacturing, where individual skill and technique have also been considered sources of uncertainty to be extinguished by mechanization, and if possible, automation. My paper claims that both approaches commit a fundamental error: that of taking an element of an activity as definitive of an entire practice. I will argue that this strategy represents a rhetorical device which functions to obscure rather than reveal the practice of scientific observation, precisely by failing to account for the inherently variable nature of skilled practice.