Creep experiments with a solution of polystyrene (M-w= 2.6 MDa, 16 vol.%, 25 degrees C) in diethyl phthalate are reported for stresses between 100 and 2,500 Pa (approximate to 3G(N) (0)/4). The aim was to look for a flow transition as reported for strongly entangled poly( isobutylene) solutions. The experiments with the polystyrene solution were repeated for cone angles of 2, 4, and 6 ( radius 15 mm) and showed no dependence on cone angle. The Cox-Merz rule was not fulfilled for stresses beyond about 800 Pa. The tangential observation with a CCD camera showed that the edge took a concave shape because of the second normal stress difference. Beyond 1,000 Pa, the concave edge develops into a crevice, thus substantially reducing the effective cross-section. This leads to runaway in a constant torque experiment. At p(21)= 800 Pa, head-on particle tracking confirms that the originally linear velocity profile takes a gooseneck shape, thus revealing shear banding. When the creep stress is stepped down to 100 Pa, this velocity profile evolves back to a linear one. The conclusion from this work is that even if nonlinear creep experiments are reproducible and a steady state is reached, this does not mean that the flow field is homogeneous.