entering the space from March until October. For this reason, I question the validity of the south-facing results. Similarly, horizontal blinds could also be applied to the other façades for conditions where the authors assume the blinds to be closed. 3) The authors have assumed that dimming ballasts consume 15% power when they maintain the electric light level at 15%. Typical modern dimming ballasts require 23 /25% input power at a minimum of 5% light output. The authors analysis overestimates the energy savings for the automatic dimming system since 15% light output would actually result in 35% input power. 4) The authors assume that 100% of the electric light striking the task in the first and second cubicles is provided by the row of luminaires that is directly overhead. With a direct lighting system, this may be a reasonable approximation, but for indirect lighting, there will be more light provided by neighbouring luminaires that remain uncontrolled. This spill light would affect the on/off set points as well as the dimming levels that should be applied to individual rows of luminaires. 5) The application of estimated overcast sky daylight factors to clear sky conditions will lead to errors in work plane daylight levels since the ground-reflected sunlight and window-incident sunlight are not considered in the authors’ approach. More elaborate computer models could be applied to address the overcast and clear sky conditions more accurately, although these models are certainly much more time consuming.