This paper presents and discusses a series of hybridization experiments carried out by Nils Herman Nilsson-Ehle between 1900 and 1907 at a plant breeding station in Svalöf, Sweden. Since the late 1880s, the Svalöf station had been renowned for its 'scientific' breeding methods, which basically consisted of an elaborate system of record-keeping through which the offspring of individual plants were traced over generations while being meticulously described. This record system corresponded to a certain breeding technique (the so-called 'pedigree method') and certain theoretical convictions (mutational rise of new varieties, and non-productivity of selection). Inspired by Tschermack's translation of Mendel's Pisum-paper, Nilsson-Ehle began his experiments in 1900 and published a first, major synthesis of his findings in 1908. If one compares these experiments as documented in the breeding records, with their representations in print, one encounters discrepancies in terms of procedure and presentation of data. This can be explained by the fact that Nilsson-Ehle was obliged to follow the recording and breeding procedures institutionalised at Svalöf, and these procedures, grounded in a taxonomic discourse, left little room for Mendelian hybridisation experiments. The twists and turns that this story takes are analysed in terms of Bachelardian philosophy of science, where the 'epistemological obstacle' functions as a central, analytic category. In contrast to Bachelard, however, I will characterise these obstacles as being of an institutional, rather than mental, nature. Thus characterized, moreover, they turn out to have been prerequisites as much as barriers to scientific progress.