This paper investigates the effect of sensor placement on the separability of convolutively mixed signals in an acoustic environment. This is an area that has received little attention in the literature, and is dealt with by considering the conditioning of the mixing system as a function of source-sensor geometry. It is shown, for simple acoustic geometries, that the mixing matrix has poor conditioning for a range of source-sensor placements in addition, and not simply restricted, to symmetric source-sensor geometries. A degradation in the performance of BSS algorithms can be expected if the sensors are placed at, or near, these `sensor singularities'. These results have important practical application to the diagnosis of the performance of BSS algorithms in acoustic environments. Also, see: http://www-sigproc.eng.cam.ac.uk/~jrh1008/
