Dataset details: CRDA/37/DS/1

67 CD-ROMs were transferred to NDAD by the British Railways Board between 23 October 2000 and 23 November 2000. A single CD-ROM containing the Microsoft Access database was received on 23 October 2000 and a further 66 CD-ROMs containing TIFF image files of documents from the BRB privatisation archive were transferred on 16 and 23 November 2000. It should be noted that the TIFF files were produced by Rank Xerox as part of their contract with the BRB to establish the imaging archive. The purpose of the files was to assist the integration of the images into the Optika FilePower system. They do not represent data that has been exported from the BRB's databases.

The Access database preserved here is derived from five databases created to simplify management of the British Rail Privatisation Archive. Each Document record should contain a brief description of the material to which the record relates, the department or subsidiary company that generated the material, correspondence reference, originator and time frame.

The database was designed to facilitate searches for documents based on several key attributes, most notably by originating British Railways Board department; by subsidiary company; and by document category.

For background information on the British Railways Board privatisation archive and the BREAD system see the Series Catalogue.

The Access database converted and preserved here is 'breadtab.mdb', one of three Access database files supplied on CD by BRB. 'Privarc.mdb', on the same CD, contained the same tables, structure, objects and data as 'breadtab.mdb', and is likely to reflect a backup or personal copy of the main database file. The other Access file on the CD was 'bread.mdb' a database with the same table structure, void of data (the Documents table was empty; the remaining tables were links to their namesakes in 'breadtab.mdb'), but with a number of forms, queries, macros and modules defined. This reflects the common practice of storing data separately from other database objects such as queries and forms. In addition, 'bread.mdb' includes a link to table Import1 in external database file 'Import1.mdb'. This database file was not provided, but examination of queries and macros in 'bread.mdb' suggests that this was a database file containing data ported between 'breadtab.mdb' and the Oracle/Optika system (see the Series Catalogue).

The object definitions of 'breadtab.mdb' and 'bread.mdb' have been exported using the Database Documenter tool supplied by Microsoft with Access 97, and are included among the Dataset Documentation (see the Dataset Documentation Catalogue, references CRDA/37/DD/4/2/1 and CRDA/37/DD/4/2/2).

The tables in 'bread.mdb' and 'breadtab.mdb' were exported from the database file in comma-separated format,using the standard conversion tools supplied by Microsoft with Access 97.

The three remaining tables have been created from data on the 64 CDs containing the scanned document images from the Oracle/Optika system. 'Images' effects the link between a document record in the Documents table and the file(s) containing its scanned image. In addition to the document images and related files, each CD also contained one file called MULTSECT.CSV and one called SEPSECT.CSV: these have been concatenated to create the derived tables 'MultipleSection' and 'SeparateSection'. These appear to implement relationships between document images comparable with the use of the ParentID field in the Documents table.

The data from the Microsoft Access database is contained in the first five tables listed below. Tables 6, 7 and 8 are derived tables. For a more detailed explanation of the three derived tables see Digital Processing and Conversion. The Documents table contains the bulk of the data; the remaining tables from the Access database are lookup tables with expanded details of some of the acronyms and codes in the Documents records. The Documents table also has a logical relationship with itself, implemented in the field ParentID: a document record may be deemed to be one of many parts of a 'parent' document, whose UniqueID is entered in the child record's ParentID field.

The dataset comprises the following table(s):

Although the Access database was clearly of great value to its users as a way of locating documents, the data suffers from a lack of on-entry validation of field values, and enforcement of referential integrity among its logical relationships. The following points were noted:

Unmatched codes

Many records in the Documents table contain either null values or unmatched codes in fields which should join to a corresponding lookup table: the table below summarises this. Some of the missing codes have been established and are recorded in the Field Descriptions.

Row numbers and Record IDs

The UniqueID field in the Documents table is a row number which was automatically generated by Access as each new record was created. However, the highest value in UniqueID (91270) does not match the number of records in the Documents table (91259), because there are no records with the UniqueID 90331 through 90341 (inclusive). This is the only break in the UniqueID sequence.

The last record in the Documents table (UniqueID = 91270) should be considered blank: all fields contain only the default values specified in the Access table definition. This is common in Access tables where the only field specified as mandatory is an automatically-generated primary key.

In three records in the Documents table where ParentID has been specified, miskeying by the person entering the data has resulted in clearly erroneous references to non-existent Document records. For these records, NDAD has identified with reasonable certainty what the real ParentID should be:

Missing Values

The following is a summary for each field in the Documents table of the number of rows with blank, null or spaces entered in that field (zeros are not, for the purpose of this analysis, interpreted as missing values):

Derived tables

Of the records in the Documents table, 20075 have related entries in the Images table. The Image records with UniqueID 90331 through 90342 inclusive (11 in all) have no corresponding record in the Documents table. These records also occur in the MultiSection table.

The checks described in Content Validation above were run against the original Access database file, using Access's built-in data-query tools, and against the converted data using NDAD's online data-query tools: consistent results were obtained in both methods.

The derived tables have been checked against the files from which they were derived to confirm that they have the same number of rows, and a random sample of rows has been compared to confirm the transformation process.