of the images before they actually begin. Again, we can speak of a technique “signature” that needs to be preserved by displaying this major isolated defect, wich was also not correctable. It is the only one of that magnitude that has remained present in this restored version.

As a sound object, SCUM Manifesto was difficult to work with. The mono sound was likely to have originated from a single mi-crophone (the one which on the table). Quite certainly omni-di-rectional, it thus captured all the sounds on a unique track — that is Delphine Seyrig’s voice, the clattering of the typing machine, the humming noise of the cathode (screen) monitor as well as the acoustics and the natural reverberation of the room. All these sound elements are embedded within the mass; it was thus impossible to isolate them from one another. The sound restoration aimed at rendering the speech more audible without altering the “ama-teur” dimension of the sound recording. As the audio level of the reference copy was very low, we first had to increase it. In the four sequences where one can hear the television, the sound is not lo-cated within the same frequency range. Each frequency required particular treatment. In the rest of the video, the continuous, pow-er line (frequency) noise produced by the television screen was re-duced. Then, the audio levels were homogenised in order to re-duce very the significant audio level variations that were aggressive to the ear. Nevertheless, there is one exception: the ending voice-over commentary, recorded by Carole Roussopoulos, who was too close to the microphone, leading to unpleasant saturation. In or-der to avoid muffling the voices, which would run the risk of ren-dering the speech unintelligible, the decision was eventually made to maintain the volume difference that was present in the original video. Finally, the brilliance of the voice and the percussion on the typing machine were highlighted to regain some brightness.