Okay, positing that in fact your meter and shutter are correct (since
you shoot chromes with good results, they can't be terribly far out),
let's check whether you mean "underexposed", as you typed, or
"underdeveloped" -- if underexposed, then for some reason your film
isn't getting enough light, and nothing you do in development will
really correct the problem (yes, you can gain a little speed with the
right developer or extended processing -- the latter at the cost of
greatly increased contrast -- but not enough to make up anything like
two stops on that film); if underdeveloped, then you just need to
increase time or temperature to compensate.

Underexposure (with properly working meter and shutter when shooting
Velvia) could occur if your meter were working differently when set to
EI 320 (or whatever speed you're using) instead of Velvia's much lower
speed; this could happen with an electronic meter if there were a bad
contact in the speed setting wheel, or due to a bad battery (lower light
levels will show greater errors, generally), or due to a "tired"
selenium cell in a battery-free meter.  It could occur if a focal plane
shutter were "capping" at higher shutter speeds (trailing curtain
catching up to the leading one, though this will typically also cause
unevenness across the frame), or if your aperture were inaccurate at
smaller stops.  It could happen if a multi-range meter were in an
incorrect range (incident dome in place but meter still set for
reflective, or spot attachment installed with meter still in averaging
mode, or even a meter with multiple sensitivity ranges in daylight range
with indoor lighting); many meters become inaccurate near the lower edge
of a range.

Underdevelopment could occur due to exhausted developer (reusing stock
solution too many times or without appropriate increase in process time,
or just too long in the bottle), temperature lower than indicated for
the time used (which could come from a dial thermometer that's been
dropped, among other things), developer solution diluted 1:1 and using
stock solution times, wrong developer in the bottle (though I don't
offhand recall a common modern developer slower than D-76), using
prewash with dry fill times, using intermittent (i.e. inversion)
agitation with rotary times or just using less agitation than indicated
in the time/temp table.  The time you're using matches the Massive Dev
Chart time for 320TXP in 120 format, which *should* be for intermittent
agitation, something like ten seconds every minute by inversion and/or
twirling stick.

How to tell which is which?  If the shadows are very empty but
highlights only slightly thin, you're probably underexposed (normal
contrast, mid-tones as thin as shadows should be).  If the highlights
are very thin as well as all other tones, you're underdeveloped (very
low contrast, i.e. little difference between highlights and shadows).

Oh, if you're using sheet film (for which, BTW, the Massive Dev Chart
shows 6 minutes instead of 9, presumably for constant agitation as with
tray development), you could get just about what you're seeing if you
loaded the sheets with the base side toward the lens (well, actually, it
should be more like 3-4 stops under, but if it's sheets in trays, you're
also pushing it about one stop).  Notch code goes at the right side of
the top edge to put emulsion toward you in the dark (I say it that way
because I've seen older film that was notched in landscape orientation
instead of portrait like all the modern film I've seen -- my method is
correct for both, but "upper right" in portrait with film notched in
landscape will have you loading base toward the lens).

After reviewing this thread, to date, I have one suggestion, if it is
a possibilty.  

With 35mm and rollfilm there are light struck edge numbers and film
identification markings.  If these are fully developed and the images
weak, this would indicate underexposure.  If the edge markings are
weak, as wel as the image, this would indicate underdevelopment.

As I haven't used sheet film in decades, I can't say wheter or not
they have usable edge markings.  If they do, your on your way.  If
not, then you're going to have to answer the question some other way.

Regards.

Robert Vervoordt, MFA

  I just checked the revised times on the Kodak web site,
they are still the same for Tri-X sheet film. My experience
in the past is that the peculiar upswept curve of this film
(and the old Plus-X sheet film also) results in rather thin
looking negatives even though the contrast index is what is
expected; the negatives look under developed but print OK.
The idea of this stuff is the produce very bright
highlights. The contrast in the shadow areas can become very
low especially wehre a high flare lens is used.
  The advise about edge markings is good although it shows
only gross errors. If overall contrast is OK when printing
but shadows are either too dark or low in contrast it can be
somewhat remedied by increasing exposure by a stop or so.
However, because the contrast of the film increases
continuously with exposure, this will not ever push the
shadows into a straight line area of the curve, with higher
contrast, because there isn't any straight line area.
  Tri-X sheet film is good for specific types of images but
IMHO is not a good general purpose film. Its hard to know
what is still available now. My favorite was Agfa film but I
am not sure they still make it in sheets and Ilford is very
questionable at the moment. Fuji Neopan 400 might be a good
choice or T-Max 400. The latter is a short toe film with a
long straight line section. Contrary to it reputation it
does not produce bullet-proof highlights, it only requires
care in processing since it is more sensitive to errors in
time, temperature, or agitation than most other films.

Well, I tried distilled water tonight and still get the same results.

So far I've...

Compared my meter against two other cameras.
Tried two different lens/shutters.
Verified I'm not putting the film in backwards.
Tried two different developers and two different fixers.
Checked my thermometer.
Mixed new chemicals with distilled water.
Used all brand new chemicals and new film.
Double,triple, quadruple checked everything.
Used two timers, simultaneously.

About the only thing I haven't eliminated is the film. One thing is
pretty clear though is that the film is under-developed rather than
under-exposed. Or maybe both.

I really appreciate all the information you guys have provided.  You
guys are amazing!

I'll let you know when I figure it out.

Thanks!
Kevin

It turns out the film was fine. As I suspected, I was doing something
stupid. It's been a while since I've done serious photography and I
have forgotten some basic things. In this case, my test shots of the
gray card for determining film speed and development time was done
with the lens focus way beyond infinity, and I had forgotten that you
have to compensate the exposure when you do. Anyway, I redid the tests
with the focus set at about infinity. Now everything works like you'd
expect. The reason switching to tmax seemed to solve the problem was
that as the same time I also corrected my focusing issue. I certainly
feel stupid, but at least everything makes sense now. Now for Tri-X my
speed is about ISO 250 and development time in D-76 is about 7
minutes.  I think that sounds more like it.

Cheers,
Kevin