The author presents evidence that very small hydraulic gradients should be used in laboratory permeability tests to minimize errors arising from seepage-induced consolidation. These errors will be carried forward to practice in significant ways. The gradients must be such that the seepage-induced effective stresses are substantially less than the maximum past effective stress. In cases of normally consolidated soils, there will always be induced errors: the softer the clay, the greater the errors. These errors seem to be compounded by conventional constant and falling head permeability tests, because of the high gradients involved, and can be readily minimized by the flow pump method. In addition, this test can be performed in substantially less time than either a constant or a falling head test. In using conventional tests one is faced with a paradox. The use of high gradients reduces the time of testing but introduces substantial errors in the test results. However, the use of low gradients extends the testing time to unacceptable limits. The flow pump test solves both these problems. If the clay is very soft and normally consolidated, even the lowest gradients can cause significant consolidation. It may well be that indirect measurements, using inversion techniques, will be required for these cases. The laboratory data presented are based on gross changes in sample thickness and thus on average void ratios. For a complete analysis consideration of the time-dependent void ratio distribution within the sample during the test is required. (Author/TRRL)