In the section on read workloads the author states:

"External file: leave the data as a log/heap and create a separate hash or tree index into it."

I am confused by this because a heap with a separate hash or tree index would be a "clustered index" and not a heap at all correct? Or am I misunderstanding?