Structure
Records in automated systems may be of fixed length or variable length. Fixed length records in a file must each contain the same fields and -any given field must always be of the same length from one record to another. If a field is repeatable, it must be repeated the same number of times in each record. Variable-length records may be variable because they do not contain the same number of fields (a given field may not occur in every record) or because a given field is not always of the same length. Bibliographic records do not fit naturally into the fixed - length pattern.
An example of a database whose records can be treated as fixed length is a mailing list for officials within organizations. In each record there can be fields for each of name, job title, province, state, postal code. Few of any of these data elements will always be the same length, but most of them will have a similar length. Each data element is allocated a number of characters and any space not used will be filled with blank spaces. The system designer has to allocate to each data element a suitable number of characters which will not be too small to accommodate the majority of records, but not too large also, to waste space by having records full of spaces. Most records in a mailing list will include all the fields mentioned above and there will not be too many fields in a mailing list database which are completely filled with spaces. In bibliographic records, the situation is different.
- Many fields occur only from time to time, for example cover title, ISSN, edition. The edition, appears in bibliographic records only when the document contain specific edition statement or when it is known to be other than a first edition.
- Many fields occur for a variable number of time in each record. For example: A document may have one author or it may have any number of joint author. The same document may be a member of more than one series. A record may be assigned more than one subject descriptor.
- A data element may vary in length between records. This is true in most systems which do not code their data but record them in ordinary language.