Micromixing has been studied in a semi-batch reactor agitated by a Rushton turbine using two different reaction schemes. With one scheme, the selectivity was similar at the same mean specific energy dissipation rate and feed location at the 3.6 L scale to that found earlier at the 25 L scale. Also, at the 3.6 L scale, a study of backmixing at different feed locations showed that it was greatest when feeding into the trailing vortex, which is also the best feed location to improve micromixing. At that location, all pipes gave a similar selectivity at the same exit velocity. Comparison with the limited literature showed that one of the two criteria for predicting backmixing fitted the present results quite well but the other was out by two orders of magnitude. The other reaction scheme was used to compare micromixing in a 3 L vessel under baffled conditions with that without them. Such a comparison has only been made once before at a larger 25 L scale and with a different reaction scheme. The earlier surprising result was confirmed, namely that the mean specific energy dissipation rate required for equivalent selectivity without baffles was less than that required with them.