The pressure collar technique enables the induction of embolism in plant xylem. This artificial cavitation is based on air seeding processes which occur when specific pressure gradients between the air and water phase of the xylem are exceeded. Standard pressure collars and a new point injection technique, which builds up a local potential gradient, were used to study the time and spatial pattern of this process. On twigs of Norway spruce (Picea abies), the cross-sectional and axial pattern, and the time-course of embolism formation were analysed via conductivity and ultrasonic measurements as well as staining experiments. Furthermore, the release of air from the twig surface was studied by immersing twig sections in water. In cross-sections, embolized areas induced by the point injection technique were smaller compared with the standard collar and restricted to a circle sector. Embolism propagated from the pressure collar towards the nearest distal and proximal nodes but not further. A release of air was also observed predominantly at the internode attached to the pressure collar. Embolism rates increased within minutes and reached approximately 80% loss of conductivity after 10 min treatment with the standard collar. The size of air entry points and embolism rates correlated significantly. Embolism formation in wood therefore depends not only on vulnerability thresholds but also on the extent of air-water interfaces within the xylem and on the time of exposure to pressure gradients. These aspects and the propagation of pressure within samples are crucial for pressure collar experiments. In addition, wood architecture influences the extent and pattern of embolism caused by air seeding processes.