All content in the repository is ultimately accessed through properties (that is, objects that implement the Property interface). The API does not distinguish between “real” content and meta-content.
Such a separation would only duplicate the entire API, since one would probably want to provide the same functionality for handling both meta-content and primary content. The distinction is in any case only meaningful at the level of the application, not the repository. Any particular application built on top of a compliant repository may, of course, choose which content is to be considered “meta”, and which primary.
However, the API does provide the concept of the primary child item. Any one of a node’s child items may be specified as its primary child item. This item can be directly accessed (without knowing its name) with the method Node.getPrimaryItem(). The primary item of a particular node (if it has one) is declared in its node type.