Based on a large sample of press data extracted from the British National Corpus (BNC), the book undertakes a detailed investigation of present-day English proper names, an important but under-researched area in English linguistics. Employing the statistical technique of binary logistic regression, this book presents a new method of analysing non-discrete categories in linguistics with reference to the grammatical notion of gradience and the principle of parsimony. The focus is particularly on the grammatical factors influencing the choice between use and non-use of the definite article – a well-known issue of uncertainty in modern English. The study also concentrates on multi-word organisation names, which have been little studied, although they occur frequently in newspaper language and have special characteristics of their own. By making precise predictive statements about the conditions under which the definite article is preferred or dispreferred, the book is also able to shed light on the theory of linguistic performance.