For the fundamental setting up of realistic recommendations for the quality and quantity of public lighting the knowledge of requirements is necessary. These requirements can be in principle derived directly from the correlation between accidents and the quality of lighting, and indirectly from the correlation between the desired visibility and the quality of lighting. Today, however, there is no method giving results which are sufficiently reliable, sufficiently accurate and sufficiently capable of generalization for the setting up of recommendations. Therefore new recommendations are based on common sense, practice and subjective judgements. Against this are major objections. Firstly it is difficult to perfect new technological developmen...