As seems always to be the case in the information field, the terminology relating to databases is loose and imprecise. Twenty years ago we used to talk about abstracting and indexing services, meaning lists of references to journal articles and other forms of publication, usually with abstracts of the articles, and with a variety of indexes. They were always printed services, published anything from fortnightly to annually. There were also data compilations, such as the Annual Abstract of Statistics, the Handbook of Chemistry and Physics, better known perhaps as the Rubber Bible, Wisden and Bradshaw. There wasn't any particular connection in the information officer's mind between an abstracts journal and the other things I've mentioned, except of course that they were all sources of information. Fifteen years ago the abstracts journals were being printed using computer support, which meant that machine‐readable versions existed, and were available for searching. Searches were done in batch mode, you may remember. Not long afterwards, information people began to acquire the ability to search these computer‐held versions for themselves, and the term ‘database’ began to be heard, meaning a machine‐searchable file of bibliographic references. Then various organisations began making collections of numerical data available, and for a little while these were called ‘databanks’, to distinguish them from ‘databases’, although not everybody actually made the distinction. More recently, there have been files available which contain names and addresses, or other sorts of directory information, which while not being bibliographical are not numerical either. There are also files of connected text, and even files which consist of computer programs that are available for downloading and use by the customers.