A new ‘transparent soil permeameter’ has been developed to study the mechanisms occurring during internal erosion in filter materials for embankment dams. Internal erosion or suffusion is the process where fine particles are removed from a matrix of coarse grains by seepage of water, and which ultimately leads to instabilities within the soil. The laboratory-based experiments in this thesis utilises a novel approach where up-scaled glass particles are used in place of soil particles, and optically matched oil is used in place of water. Rhodamine dye in the oil allows the fluid to fluoresce brightly when a sheet of laser light is shone through the sample, while the glass particles appear as dark shadows within the plane of the laser sheet. This technique is known as Planar Laser Induced Fluorescence (PLIF) and enables a two-dimensional "slice" or plane of particles and fluid to be viewed inside the permeameter, away from the permeameter walls. During a test, fluid is passed through the solid matrix in upward flow, with the flow rate (therefore hydraulic gradient) being increased in stages until internal erosion or bulk movement of the entire assembly develops and progresses. A high speed camera captures images of the two-dimensional plane over the duration of a test, which are then analysed using Image Pro and ImageJ processing software. Until now, the fundamental mechanisms that lead to internal erosion have been rather speculative, as there has been no way to physically observe the processes behind the initiation and continued movement of particles. This visualisation experiment allows internal erosion mechanisms to be studied away from permeameter walls where boundary effects do not occur. The technique was validated by confirming Darcy’s (1856) law of laminar flow, and Terzaghi’s (1925) theoretical critical hydraulic gradient for an upward flow through materials with no top stress. Results of replicated materials tested by Skempton and Brogan (1994) and Fannin and Moffat (2006) also confirm this methodology to be valid by way of material behaviour, permeability and the alpha factor (Skempton & Brogan 1994). An assessment to predict the stability of soils was carried out using the Kenney and Lau (1985), Kezdi (1979), Burenkova (1993), Wan and Fell (2008) and Istomina (1957) approaches, with the Kenny and Lau and Kezdi methods proving to be the most robust across the particle size distributions tested. In the tests, unstable materials showed a migration of fine grains under hydraulic gradients as low as ic = 0.25, while stable materials showed little movement of particles, and eventually failed by heave. Image processing using Image Pro and ImageJ were successful in producing quantitative results, however with further enhancements to the test equipment and methodology, these could be improved upon. The testing technique developed in this thesis has proven to be successful in the study of internal erosion of filter materials. The technique proves that optically matched glass and oil can behave similarly to soil and water materials as used in previous laboratory testing, and that the PLIF technique and image capturing has merit in understanding the mechanisms occurring during internal erosion processes.