The present thesis discusses the history and meaning of the term and concept called grammaticalisation. Linguists usually ascribe the coinage of the term grammaticalisation to Antoine Meillet (1866-1936), who allegedly played a vital role in the history of grammaticalisation. It is also widely acknowledged that grammaticalisation was in some way 'revived' during the 1970s, and that Talmy Givón had an important role in this, as demonstrated by the popularity of the saying 'Today’s morphology is yesterday’s syntax' (taken from one of his articles). I show that Meillet wrote little about grammaticalisation and that he hardly ever used this word, and possibly did not mean for it to be viewed as a term / label. Moreover, the paper in question (Meillet, 1912) is basically a general introduction to a concept which he sees as a continuation of a notion with a long history. In addition, I prove that there are no clear links between Meillet and Givón’s work in the early 1970s. Despite the general acceptance that Meillet coined grammaticalisation, my thesis proves that it could have been coined more than once, and that it does not always mean the same thing to all users. I show that sometimes the term is accompanied by examples which others have used to illustrate lexicalisation, a term which some employ for a process that is seen as the opposite of grammaticalisation. I therefore advocate careful use of our definitions of terminology and concepts, and insist that we should define our notions, instead of letting examples do the work of illustration and definition. Finally, I question whether it is true that grammaticalisation is unidirectional. I research the history of the view that grammaticalisation is a unidirectional process. Grammatical relations can be expressed by different means – e.g. word order, content words becoming grammatical markers, or parts of words being given a function. I believe all these should be compared, in order to improve our knowledge of how languages change and why. I claim that they all represent sub-processes of a superordinate category which I have labelled supergrammaticalisation.