Bubbles produced by a volume of 500 cc of water falling through a distance of 1.07 m into a water-filled basin were allowed to rise into an adjacent water-filled tube whose top was sealed at a level of 1.7 m above the level of water in the basin. The rise of these bubbles was recorded on video at a height of 1.5 m above the level of the splash: larger bubbles were recorded first. A model has been devised to describe the rise of such bubbles. The rise speed of the bubbles at the level of the video camera decreased with time after a splash, becoming nearly constant after a few minutes. The model used this long term rise speed to estimate the nitrogen saturation in the water. Oxygen saturation is measured by an electrode. Given the saturation it was then possible to use the model to calculate the initial spectrum of bubbles rising up the tube from the splash (i.e. the spectrum of such bubbles a second or two after the splash, when bubble fractionation or coalescence has ceased). The smallest bubble that could be seen depends on the gas saturation, but was typically of initial radius 20 μm, corresponding to a radius of 50-70 μm at the level of the video. Such spectra were found at different saturations, distances from the splash and salinities. At gas saturations of 105%-120%, a peak appears in the spectrum at a radius of about 20 μm. The time of admission of bubbles into the tube after a splash could also be restricted. For unrestricted sampling times, dN/dr varied as r−1.5, when expressed as a power law. The spectrum above the peak value became steeper at later sampling times. At salinities below about 10 ppt, the number of bubbles of calculated initial radius < 600 μm is reduced. While no attempt was made to produce a realistic breaking wave, these results are relevant to attempts to define a source function of bubbles at sea, and to comparisons between fresh and salt water experiments.