Abstract/Summary

This thesis examines the process of ‘fitting’, or determining the preferred spacing of the letters and other forms that comprise a typeface. Successful fitting is important to the readability and aesthetics of type, and is traditionally performed as a manual process. The objective is to determine to what extent this manual process can be modelled and expressed in an algorithm, to increase the theoretical understanding of fitting and suggest practical strategies. The research incorporates methodologies from several disciplines, including historical studies, algorithmic analysis of procedures and strategies employed in fitting, development of computational software, and empirical testing. The manual process of fitting was analysed from historical sources and contemporary practice. From the study, an axiomatic model was developed expressing the first principles of fitting Latin text for continuous reading and interdependencies between those principles. Prior work was evaluated in relation to the model and a new method was developed to fit typeforms with open counters, a class of forms historically reported to be difficult to fit. A composite algorithm was developed that traverses the typeforms in a typeface, fitting each form with the simplest technique applicable, until the fitted set is complete. The composite algorithm was used to fit a set of typefaces, which were tested in an online reader study. Readers were shown a series of text samples, utilising original and refitted fonts, and asked to mark letter sequences they felt exhibited poor spacing. The composite algorithm achieved lower rates of reported poor spacing for multiple letterform pairings than the alternative conditions. This supports a view that the axiomatic model capably represents the fundamental fitting process, that the novel method for fitting open counters can improve on prior techniques, and that the composite approach to algorithmic fitting, combining multiple discrete principles, holds benefits for the fitting of typefaces.