This quantitative study, based on a computerized corpus of texts written by five men in early 16th-century Nuremberg, analyzes the way linguistic, social and stylistic factors work individually and in interaction to influence variation observed in the texts. The study provides evidence that consonantal variation in early modern written texts is not random. To a surprising degree, it is possible to account for the structured heterogeneity in the writings studied by using methodologies established for spoken language in modern day communities. Like spoken languages, variation precedes change in the written language, and again like spoken language, not all variation is followed by change. That is, while variation cannot always be demonstrated to be structured, much of it is clearly and reliably attributable to the same complex of linguistic, social and stylistic factors which shape the structured heterogeneity of spoken languages of our own time. Of particular importance is the quantification of an individual's relationship to an emerging ideology of language standardization, and the way that relationship interacts with written language variation.