The effects on bubble growth dynamics of physically altering the substrate surface on which gas bubbles are nucleated following depressurisation of a saturated solution have been studied. Seven different modified surfaces have been investigated, with modification methods ranging from ageing (by repeated washing) to the production of both random and ordered mechanical abrasions in one, two or four orientations. The main effect of treatment was found to be on the size of the detaching bubble (d(max)). rather than either nucleation or growth kinetics. The abrasions form boundaries which impose a barrier to the normal lateral expansion of the neck across the substrate surface. This leads to a significant decrease in the detaching bubble size, with the sample treated with mechanical abrasions in four orientations producing a mean detached bubble up to 50% smaller than that on an unmodified surface. It was concluded that ordered mechanical abrasions could be used to produce smaller bubbles in accordance with the relationship: d(c) = (d(max) - d*(max))(2) / xi for 1.26 mm less than or equal to D-max less than or equal to 3.58 mm. where d(max)* represents the smallest detaching bubble size attainable in the system and xi the coefficient of proportionality of the d(2) : d(c) correlation. Both d(max)* and xi are constant for the system investigated regardless of the extent of physical surface modification of the substrate surface, xi approximating to the mean bubble size. A further rudimentary taste test revealed a preference for soda water containing bubbles of smaller size.