How do we separate code from data, without asking a programmer?
We do all understand what a tree is: nodes and branches: it suits us to call leaves a type of node, and we also summarize the roots as one root node.
One branch is called <head>; another <body>. Identifiers seldom have more than a passing relation to their English meaning.
The <head> branch contains the document title and other information that is not considered content; <body> is therefore where we start.
The document is considered a box; a rectangle. Each branch is a rectangle within the one below it.
The document tree, then, is data.