The paper deals with practical points arising during the course of the Cape-Radcliffe photometric programme. The principal conclusions are : ( a ) The errors introduced by the Schilt photometer and the process of reducing the photometer readings are considerably smaller than the total mean error. ( b ) Tests with artificial star images show there is no detectable difference in results whether the plates are developed by brushing, rocking, or in stagnant developer. Also there seems no advantage in using backed plates for this work. ( c ) There are considerable variations in colour equation between plate and plate, of which only a part can be accounted for by purely accidental errors. Some of the variation is correlated with exposure time. The colour equation also varies with the size of the diaphragm used in the modified Schilt photometer, an effect most probably due to the chromatic aberrations of the camera lenses. In some series of plates there is a significant change of colour equation with apparent magnitude (“photographic Purkinje effect”), and in one series this effect has probably been exaggerated by the use of plates with too long a range of spectral sensitivity for the colour correction of the lens concerned. The variation of colour equation with focal setting is small. ( d ) The P g magnitudes do not use any appreciable amount of light of shorter wave-length than 3800 A. ( e ) The brightness of the night sky at Pretoria is about 12 m .0 per square minute of arc for both P g and P v measures; at the Cape the values are 12 m .0 and 11 m .6 respectively. Sky fog appears to have negligible effect on the current programmes. It is, however, unsafe to attempt to set up an accurate magnitude scale by in-focus photometry when within about 5 m of the limiting magnitude of the telescope, unless corrections can be introduced for the sky fog. ( f ) Three methods of setting up an accurate magnitude scale by in-focus photography have been tried. Objective diaphragms proved unsatisfactory, but promising results were obtained with rotating sectors and also with a variable exposure-time method. Both procedures deserve further investigation. ( g ) Although the Cape photometric lenses were designed for a field of diameter 15°, it is inadvisable to use them more than 3° off axis, owing to rapidly increasing distance corrections. ( h ) For general magnitude work by any method the Earth's atmosphere sets a limit (possibly about 0 m .01) to the accuracy it is worth aiming at. There is a natural limit, also of about the same order, set by the inherent vagueness of the definition of a stellar magnitude.