The invention of sound recording led to the foundation of sound archives. And the possibility of sound recording was essential for the 'new methods' in comparative musicology, such as transcribing the music (which could now be done far away from the place of recording, with researchers able to listen as often as necessary to the same piece of music), measuring (e.g. pitch, duration, tempo, interpretation), and compiling a pool of examples for comparison from which 'maps' of music could be created. In the light of these premises, an understanding of the goals of the Phonogrammarchiv is important in any discussion of the development of comparative musicology as carried out in our institution and in cooperation with it. Following its founding motion, the Phonogrammarchiv was established as the world's first sound archive (a repository), but at the same time was to hold the role of a research institute. Archival practice was geared to research purposes: in order to ensure that sound recordings represent scientifically valuable sources, providing technical and content-related documentation was essential. The Vienna archive was founded from a theoretical point of view with a scientific 'vision', but without any stock of recordings. So the collection strategy was important concerning sound recordings as prerequisites for comparative musicology. As a case study Robert Lach's work on 'Russian prisoner-of-war songs' will be the main example to be presented. The following questions arose out of Lach's study and approach and will be discussed: 1. Can armchair ethnology and field research be seen as opposite methods in our discipline? 2. What is the impact of sound recordings which, per definition, are the prerequisite for ethnomusicology? 3. And what is the role of archives run for more than 100 years, what is the value of recordings stored, documented and preserved (for the future)? Concluding remarks reflect that in the course of time the research changed from a text-oriented point of view to exploration of music in context, a situation which causes the uncertain use of recordings made by others. The archive's project of publishing the complete historical collections possibly will offer a corpus for new studies. At the moment we (only) are dealing with 'case studies' which give certain insights, but for revisiting our discipline it might be fruitful to use the huge stock of digitised audio recordings and original documentation as well as elaborated comments. In the light of the discussions about the value of the recordings from today's point of view a corpus digitally available might offer new possibilities for overviews and comparison.