Data copyright © Dr Graeme Earl unless otherwise stated
This work is licensed under the ADS Terms of Use and Access.
Dr
Graeme
Earl
Department of Archaeology
University of Southampton
Avenue Campus
Highfield
Southampton
SO17 1BJ
England
The collection is essentially made up of three sets of related data:
It is possible that one collection could be made up of material from more than one site. It is also possible for one site to yield more than one collection; the same site could be re-excavated and produce a second collection. Similarly a collection could be published in more than one publication and a publication could cover more than one collection. This results in there being many-to-many relationships between collections and sites and between collections and publications. However, although a collection could be included in multiple bibliographic references the published details of that collection must be constant - the condition of the pottery or the associated non-ceramic items are not dependant upon the reference but are intrinsic qualities of the collection. There is therefore an optional one-to-one relationship between the collection and the published details of that collection plus an optional many-to-many relationship between the collection and the bibliographic reference.
Within the main categories of data collection, site, and published details there are also multiple relationships between the category and some of the attributes. For example a collection may have been amassed as a result of a number of different methods of recovery, casual pick-up, systematic field walking and excavation perhaps. Thus there is a one-to-many relationship between collections and their recovery method. The various one-to-many relationships within the database are directly discernible from the data collection forms in which multiple boxes are present for some of the entries.
Within the data there are therefore a number of different relationships between the elements:
The relationships between the data items describing a collection can be formalised as a data model showing the relationships and their types between all of the data items.
The database is implemented as a number of tables based on the data model. The main part of the data is contained in four major tables - Collection, Sites, Published Details and Bibliography. Two link tables deal with the many-to-many relationships between collections and sites and collections and references. Nine tables hold the various one-to-many relationships. One table holds expansions of the surveyor´s initials along with their survey areas. Fifteen tables hold the explanations of the codes for the coded database fields.
All tables are keyed, that is there is a single field with a unique value or a pair of fields which combine to form a unique value defined as unique and used by the database manager to index the records. Where possible the key is constructed from primary database data. The four major tables, however, have an arbitrary numeric field as a primary key because there is no guarantee that any combination of fields in the tables will be either present or unique. The keys form links between the tables thereby establishing the relationships between the records in the various tables.
The structure of the database is illustrated in the Entity Relationship Diagram below - which shows the table and attribute names and also the nature of the links between the tables:
Two types of cascaded updates are used between the tables. Links from look-up tables are configured to cascade updates only. All other one-to-many links are configured to cascade both updates and deletes.