The paper specifies a novel coder/decoder that compresses sequences of 3D texture data to support the haptic rendering of animated volumes. Lossy compression gives 25% of the original size. Lossy compression and clipping gives 3% but requires the renderer to query for new data when the point of contact moves. We evaluated the codec by means of rendering volumes on a 3-DOF Phantom 1.5A that simulated viscosity for each texture element. In this setting, we studied the impact of information loss in controlled experiments. We validated that our codec preserves haptic perception. An area of application is the development of network-distributed virtual environments that generate visual and haptic feedback.