There is a difference between released memory and unused physical memory.
The application is releasing memory when it ends. However, Windows does not
mark all the released memory as unused physical memory, because it knows the
contents (at least for some of it). Windows tries to track the contents of
as much memory as possible, because this improves performance. You can see
this if you start your application, close it, and start it again. The
second startup is faster. Although the program released its memory, Windows
kept track of the contents and was therefore able to re-load the application
faster.

Applications that fail to release their memory are a problem, but it is a
very rare problem. You need a much more sophisticated monitor than the one
you are using to find it. You don't want to force Windows to return the
released memory to unused physical memory because that will prevent it from
performing the optimisations (when possible) while not providing any
performance improvement.

The example you have quoted is not a valid comparison. Instead, compare the
physical memory and swap file usage after loading the large file for two
cases: one where you have done the prior properties enquiry, and the other
without doing it.
--
Jeff Richards
MS MVP (Windows - Shell/User)
"Jon Danniken" wrote in message
...
Okay, I'll try to spell this out more clearly for you guys.

When I boot into this machine, I have 30Mb of unused physical memory.

When I look at the "Properties" of a 20Mb .wav file, the unused physical
memory drops to 15Mb.

When I close the "Properties" windows, the unused physical memory stays at
15Mb.

If I open 20Mb file (for instance, in SndRec32), the unused physical
memory
drops to 0, and a 5Mb page file is used.

xxxxxxxxxxxxxx

Hence my original question: How can I get Win98 to release memory after an
application (or the Properties window) is closed?

Note: As I stated as the *first thing* in my original post, adding more
RAM
is not an option at the current time; please do not waste time by
belaboring
this point.

Thanks,

Jon