The process of desilverising by fractional crystallisation with reflux was investigated through laboratory batch tests and continuous laboratory bench scale tests at CSIRO Clayton, to a 300 kg/hr pilot plant at Port Pine. The pilot plant built at BHAS comprised a vertical cylindrical refining column with internal diameter 250 mm and height 3.2 m, attached at the top to an off-set cylindrical stirred crystalliser tank. Crystals were made in the crystalliser by intermittent immersion of steel cold fingers, which were water cooled between immersions. Gas heat was used for melting the crystals at the base of the column and for side compensation heating. A computerised in strumentation system was employed for data logging and for controlling some critical plant operations. A mechanical device was first employed to promote counter-current plug flow of crystals down the column against the rising stream of reflux liquid, but this was found to be unsatisfactory owing to excess noise, frequent mechanical failure, and excessive accumulation of dross. Plug flow was finally promoted by pulsing the column. Low silver levels, less than 1 gitonne, were achieved with both methods of crystal transport under total reflux conditions. The pulsed column gave results for silver and bismuth which agreed with the theoretical predictions of a computer model, but there was some selective drossing of antimony. In the mechanically agitated column, excessive drossing promoted selective removal of most of the antimony and some of the bismuth, and the results after extended operation were well removed from the theoretical predictions. Plant failures and operating problems prevented the achievement of satisfactory product quality with the necessary reliability under production conditions. Other priorities necessitated the termination of the project before the process could be optimised.