A number of problems have been identified with existing colour specification systems and their physical exemplifications: colorimetric conditions (illuminant, observer, spectrophotometer geometry, specular component inclusion/exclusion, etc.), restrictions of the notational system adopted, and limitations of the colour atlases themselves (sample size, sample error, sample range, etc.). Developments in computer science, and in particular CRT displays and colour printing devices, now provide increasingly more affordable alternatives to traditional colour ranges and atlases. Techniques have been developed to increase the accuracy of printed colours relative to their CRT originals, and devices are manufactured to calibrate colour monitors. However, colour selection software developed to date is still primarily, based on existing colour ranges and colour specification systems, or on a more general mechanism specific to all individual application. It is argued that basing such software on an existing model of colour specification inherits some of the problems of that system: notational methodology, means of representation, and atlas limitations, Ir is proposed that the restrictions imposed by such software be relaxed by providing a flexible method of constructing application-specific colour specification systems with conversion to a standardized notation for accurate colour communication. Furthermore, it is proposed that such a tool would be invaluable for the evaluation of human colour perception.