Building on XPath

Originally its own notation In its original drafts, XPointer was built on TEI pointers, a creation of the Text Encoding Initiative. As XSLT developed, and XPath became a separate spec, the draft was rewritten to take advantage of XPath. (This means the W3C only supports two families of document selectors - XPath/XPointer and CSS selectors - instead of three.)
A few simple extensions XPointer has consistently supported the 'bare' fragment identifiers of HTML, mapping them to XML ID attributes. It also provides a notation for moving through the document tree using numbers.
Adding points Sometimes pointing to a particular node in a document isn't exactly what you need - you just need to reference a particular point in the document.
Adding ranges Hypertext, like most document-oriented applications, isn't always clean. Links often need to cross node boundaries, and sometimes you want to be able to say things like "link every occurence of the word glycerin to its glossary entry". XPointer allows you to specify ranges between points and ranges identified by their string content.