A test was made of the previous unexpected observation that embolized vessels were refilled during active transpiration. The contents of individual vessels in petioles of sunflower plants were examined, after snap-freezing at 2-h intervals during a day's transpiration, in the cryo-scanning electron microscope, and assessed for the presence of liquid or gas (embolism) contents. Concurrent measurements were made of irradiance, leaf temperature, transpiration rate, and leaf water potential (by pressure chamber). Up to 40% of the vessels were already embolized by 0900 (transpiration rate ~5 µg·cm−−2·s−−1, water potential about –300 J/kg), and the proportion declined to a minimum (as low as 4%) at 1500. This was the time of highest transpiration rate (~25 µg·cm−−2·s−−1) and most negative water potential (–600 to –700 J/kg). Images of vessels with mixed gas and liquid contents showed water being extruded through pits in the walls of the vessels to refill them. The data indicate that: (1) the water columns are weak and break under quite small tensions; (2) embolisms are repaired by refilling the vessels with water on a short time scale (minutes) throughout the day; (3) the vigor of this refilling process is adjusted by the plant on a longer time scale (hours) to the intensity of the water stress; (4) the pressure chamber balance pressure (P) does not measure tension in the vessels; (5) P is also not a measure of water stress (as measured by vessel embolization); and (6) P is a measure of the plant's response to water stress, i.e., a measure of the vigor of the refilling process. The test confirms the previous observations and negates all the assumptions and evidences of the Cohesion Theory. The data are fully consistent with the Compensating Pressure Theory, which predicted the relations demonstrated in this experiment. Using the assumptions of that theory it is easy to outline a simple mechanism by which the refilling of vessels might be achieved by reverse osmosis, and the adjustment in (3) might be achieved by osmoregulation in the starch sheath.