The history of musical performance, of styles and performing practices, is a tale of plurality, yet this is not well represented in most historical performance research. Research methodologies, specialist instrument pedagogy and professional practice have tended to privilege sources that fulfil at least one of two criteria: proximity to the composer, and the ‘great performer’ status of their originator. These criteria are inherently problematic, though they have obvious benefits in terms of narrowing the field of information and providing the semblance of external validation. Crucially, they can also lead to valuable evidence being overlooked or discounted, particularly if that evidence challenges prevailing musical tastes. In this presentation I will focus on the recordings of a number of German clarinetists active between 1900 and 1940 who, though respected in their day, have no strong links with great composers and little posthumous reputation. Though undoubtedly accomplished, their playing as documented by the recordings also features elements that are challenging both to the modern practitioner and to listeners. Choosing to disregard these ‘problems’, I have used evidence drawn from the recordings in combination with composer-proximate and more ‘dubious’ written sources as a stimulus for technical and creative experimentation with Brahms’s clarinet music on historical instruments. The initial results of this ongoing project include unexpected insights into the relationship between instrumental technique and performing style, and they hint at the possible benefits of a more pluralist and less hierarchical approach to historical evidence in practice-based performance research.