In routine dosimetry we assume separability of the collimator (S(c)) and phantom (S(p)) scatter components that together comprise the total scatter factor (S(c,p)). In practice, the addition of blocking also affects the photon fluence attributable to the treatment head and flattening filter in a complicated way. The reduced aperture blocks out some of the head scatter contribution, while the block and tray add back secondary scatter. In the following we present techniques for directly measuring the aperture effect on S(c) in air or in a full-scatter phantom. The change in S(c) is found to be a scaleable quantity that can be modelled as a simple linear fit to the ratio of projected open-to-blocked equivalent square fields. Measurements have been made for 6, 18 and 24 MV photon beams on one Varian 2500 and two Varian 2100c accelerators. Results indicate a progressive loss of collimator scatter contribution with increased field blocking that is amplified with increasing energy. Block and tray scatter only contribute significantly to S(c) for large fields and treatment distances of 80 cm or less. Application of these corrections in monitor unit calculations is presented.