Geographic data for any particular area is stored in separate layers. For example, roads are stored in one layer, parcels in another, and buildings in a third. To enable the data in each layer to integrate when displayed and queried, each layer must reference locations on the earth's surface in a common way. Coordinate systems provide this framework. They also provide the framework needed for data in different regions to be referenced in different ways. Each layer in the geodatabase has a coordinate system that defines how its locations are georeferenced.
In the geodatabase, the coordinate system and other related spatial properties are defined as part of the spatial reference for each dataset. A spatial reference is the coordinate system used to store each feature class and raster dataset, as well as other coordinate properties such as the coordinate resolution for x,y coordinates and optional z- and measure(m)-coordinates. If required, you can define a vertical coordinate system for datasets with z-coordinates that represent surface elevation. For an introduction to these properties, see The properties of a spatial reference.
Prior to ArcGIS 9.2, spatial references were low precision. They allotted a limited space for coordinate storage, so it was not possible to store coordinates close together and have a large domain extent at the same time. The high-resolution spatial references created and maintained with ArcGIS 9.2 allocate more space for coordinate storage, allowing coordinates to be stored closer together while expanding the x-, y-, z-, m-domains of the dataset. Since the resolution and coordinate system define an extent for the dataset, there is no longer a need for a user-defined domain extent. To take advantage of these benefits, you can upgrade the spatial reference of any existing data to high precision. For more information, see Migrating to high precision. Low-precision spatial references continue to work in ArcGIS 9.2 as they always have. See Working with pre-9.2 spatial references for details.