When Arbortext Editor writes an XML document with no DTD (free-form XML), it lists its defined elements as processing instructions (PI) at the beginning of the file. Each PI specifies the name of the element, its attribute list, and its style category.

For example, a document containing the <title> and <author> tags may have corresponding PIs similar to:

The PI also specifies whether the element is empty and if it is a graphic. When the instance is reloaded in Arbortext Editor, the PIs are parsed and the attribute list and category for each element are retained from the previous session.