From SARA to Xara Xara is the working name for a new version of SARA, thèSGML aware retrieval application' originally developed for use with the British National Corpus (BNC) in 1994. The system has been completely rewritten as a general purpose tool for searching large XML corpora, with a particular focus on the needs of corpus linguists, with close attention to new XML-based encoding standards, and with the benefit of hindsight derived from a decade of feedback from hundreds of SARA-users world wide. The Xara system combines the following components: (1) an indexer, which creates inverted file style indexes to a large collection of discrete XML documents; (2) a server, which handles all interaction between the client programs and the data files; (3) a Windows client, which handles interaction between the server and the user. The modularity of this architecture has several advantages, permitting, for example, the development of multiple specialized client programs for different applications or styles of usage. In addition, an index building utility, called Indextools, is supplied with Xara, which simplifies the process of constructing a Xara database. Its chief function is to collect information about the corpus to be supplied additional to that present in any pre-existing corpus header, and to produce a validated and extended form of the corpus header. It can also be used to run the indexer and test its output.