In providing the information outlined in the previous section, an archive's cataloguers are only fulfilling a part of their function. If uncatalogued material is effectively useless (since an archive and its users may be unaware of its content in any but the most elementary sense, and may indeed be virtually unaware of its existence) then material on which information is solely available in an unindexed catalogue is only slightly more open to use. Knowledge of items relevant to particular needs will only be gained from such a catalogue by prolonged searching, by accidental discovery, or by a fortunate feat of memory by a member of the cataloguing staff. None of these constitutes a reliable basis for an archive's service.

Indexes and classification systems are provided to open up the archive by providing access points and channels to the information available in the catalogue and -by extension -in the collection itself. Additional access points are provided by sorting the information in the catalogue, or extracts from it, in alternative orders: thus a catalogue of classical music, catalogued by title, would usefully be supplemented by indexes of composers, performers and so on. Channels to information are provided by the establishment of a logical structure of indexing vocabulary and styles, so that cataloguers and researchers can be instructed to consider the collection in similar terms. For example, the same archive of classical music might wish to provide an index by chronological period. In such a case, the information would be directed into the channels by the decision either to name periods by actual year or century designations, or by epoch titles such as 'High Renaissance' and 'Baroque'. The nature of the decision taken is arguably of less importance than the fact that the decision is taken and consistently applied. It does not matter too much whether a period is called 'Nineteenth Century' or 'Victorian'; it matters a great deal if, on different occasions, it is called both (and a number of other names besides) so that a user looking up only one term will fail to find all relevant information.

The number and nature of the indexes provided by an archive will depend on many things, including the resources of the archive and the expectations of its users. The principal determining factor is the nature of the material collected. Archives of commercial recordings may concentrate on indexes of titles (which will require indexing, to restrict the potential confusion of catalogue titles in different languages or formats: 'Die Dreigroschenoper' and 'The Threepenny Opera'; 'Ninth Symphony' and 'Symphony No.9') and of authors, composers, performers, directors, etc. These concerns will be shared by collections of other types of music, such as ethnomusicology, but such collections will also share indexing requirements with other archives of field recordings -circumstances of recording, name of collector, and so on. Different types of field recording will have different priorities -place, for example, which is essential to collectors of dialect, linguistic and wildlife recordings, for example, is generally considered to be of only peripheral interest in oral history. Oral history and other essentially narrative recordings, however, require analysis by subject content -an area where vocabulary control and the structuring of an index are at the same time most important and most complex.

The style of index provided will also vary. In some archives and some circumstances, the preferred procedure is to reproduce effectively the bulk of the catalogue description for the relevant item at every entry in the index; in others, the index term is merely followed by a list of accession numbers for appropriate recordings. In some cases, attempts have been made to use extant bibliographic classification systems -such as Dewey or the Universal Decimal Classification -as a ready-made system of vocabulary control for subject indexing, while other archives use their own system of 'natural language' index terms. In a general chapter such as this one, there can be no real attempt to specify that any given solution is correct, or better than another: the choice between numeric classification codes and natural language, for example, is the choice between a system which comes ready-developed, but may prove difficult to use (because of the need to translate index statements and enquiries into number-strings), and a system whose development must be carefully planned and monitored but which may seem more accessible and 'friendly' to staff and users.

A final factor in the question of indexes to supplement or make accessible the archive's cataloguing effort is the provision of something like the indexing function through mechanisation or computerisation. At its most highly developed, technology may contradict the statements made in the first paragraph of this section: catalogue information may be held in a data-base which can be interrogated 'on line' via a computer terminal, so that physical indexes are not apparently necessary. A later section of this chapter will look at the topic of computer systems.