Historical investigations of mineral flotation have highlighted the many factors which affect flotation. Some of these are related to the mineral, its degree of liberation (however defined), solution chemistry and flotation reagents. The complexity of the interactions between these factors often leads to uncertainty over whether the final grade/recovery results obtained in either laboratory or pilot scale tests are in fact the maximum recovery at a specific grade. The reason for this is that there has not previously been a general technique for establishing the floatability of an ore. This study describes the further development of the 'tree' flotation technique for minerals other than coal, and in particular its application to iron ores. Oxides have been chosen as a system of intermediate complexity between the simple coal system and sulphide minerals to illustrate the principles and develop the concepts. The 'tree' flotation procedure consists of the progressive flotation to completion of successive froth products, followed by reflotation to remove entrained material. It therefore represents an 'equilibrium' approach to floatability rather than a kinetic approach. By ordering the various products and tails in order of increasing grade, a cumulative curve of grade and recovery can be constructed against which an isolated batch or plant result on the same ore can be compared. In this investigation, a Jaspilitic iron ore was tested at various grind sizes to determine the 'best' flotation grade-recovery that could be obtained with a given set of reagent conditions. The separation characteristics of a process are best measured on a system which contains particles which behave as 'middlings' with respect to the process in question. For this reason a natural ore was chosen in preference to model systems in order to relate more readily to plant situations.