Apache is built around a parent process which accepts connections from clients, and one or more child processes which do the actual work in processing the request and returning the accessed page. The number of child processes adjusts itself dynamically to the number of connections to the web server, within the bounds given here.
The web server will start up with this many child processes.
The maximum number of child processes. Each client connection uses a child process, so this value indirectly controls how many clients can simultaneously connect to the web server. If this value is to low, clients will be locked out, so be liberal. This is intended mostly as a break to keep a runaway web server from bringing the entire system down with it if things should go awry.
A child process will exit after serving this number of accesses. It is intended as a failsafe against possible memory leaks (not that any are known).
Apache makes sure that there always are a few idle child processes, to be able to handle transient load spikes. If fewer than this number of child processes are idle, more are started.
Apache also makes sure that there never is more than this number of idle child processes. If there are, some of them are killed.