The term "buffer pool" implies -- to me -- a pool of buffers in the same sense as an object pool design pattern. But, as far as I can tell, each buffer pool is a single buffer that can be shared among table spaces. Am I missing something?

I don't know if you are missing something, but you are jumping to conclusions. All of these terms were invented in the early 1980's well before object-oriented nomenclature came into wide use.

A bufferpool is memory allocated for buffering table and index pages. This is known in today's terms as memory cache.

It is a "pool" of individual bits and bytes of memory that are associated together into a single bufferpool. One can have multiple bufferpools in a database.

You might want to read this manual first, and then post here if you have any specific questions. I don't know what you are reading now, but it doesn't appear to do a good job of explaining things:
ftp://ftp.software.ibm.com/ps/produc.../db2dae951.pdf

Here are all the manuals:
IBM - DB2 Version 9.5 for Linux, UNIX and Windows English manuals

Unless you need the z/OS manuals:
IBM - DB2 for z/OS - Technical Resources