Both for research and for training in psychoanalysis psychoanalytic dialogues must be made publicly accessible. With traditional procedures (supervision, case seminars, publications) condensations are inevitable, for which, however, no binding rules exist. Empirical research results are quoted to show that these condensations are highly selective. We cannot see any immediately plausible external criteria to decide which kinds of selective condensations are justified and which are not. Heimann's conceptualisations of the three "positions" within which psychoanalysts function as therapists is introduced to explain why analysts especially fail to report their own interpretations. If one tries to evade the devil of selective condensation by audio-recording, one falls into the deep blue sea of the "systematic acoustic deficit". Of the "running commentary" within the analyst only the small part which is uttered as therapeutic intervention can be registered acoustically. From a discussion of the problems analysts and patients experience with audiorecordings it is argued that the analyst/patient dyads consenting to this procedure are a systematic sample and thus not representative for the total of analysts and their patients. Finally a research program is presented which tries to partly fill the systematic acoustic deficit by gathering "proning retrospectives". The latter are spoken by the analyst after each "proning" (= session) on the same cassette which recorded the therapy. In a free part the analyst free associates. In a second, the "intervention explication" part the analyst identifies roughly 3 "important" interventions and tries to formulate for each its source, its purpose and the point in time when he had conceived his formulation. An evaluation design for thus gathered information is described.