The concept of a hierarchical arrangement for filesystems dates back to the dawn of modern computing history. And no, it wasn't invented by Microsoft in MS-DOS 2.00. :-)
The hierarchy, or layering, of the filesystem is accomplished by directories. By definition, a directory can contain zero or more subdirectories, and zero or more data elements. (We're not counting the two directories . and .. here.) On UNIX (and derived) systems, pathnames are constructed by having each component separated by a forward slash. (On other systems, like VAX/VMS for example, pathname components were placed in square brackets and separated by periods.) Each slash-separated component represents a directory, with the last component representing a directory or a data element. Only directories can contain further subdirectories and data elements.
Since a directory can contain subdirectories, and the subdirectories can contain further subsubdirectories, this “chain” can theoretically go on forever. Practically, however, there are limits to the length of this chain, because utilities need to be able to allocate buffers big enough to deal with the longest possible pathname. It's for this reason that the names of the individual components (be they directories or data elements) have a maximum length as well.