The tornado east of Buffalo, Kansas was shrouded in rain when I was south of 
the town at the intersection of 75 and State Road 39 looking to my north 
northeast.  I thought I caught a glimpse and aimed my camcorder at an area 
that featured a long beaver's tail stretching to the southeast and a sudden,
strong inflow jet. It was apparent that the storm had strengthened dramatically 
and today I see from the pictures on Stormtrack and elsewhere what I missed.

I had made OUN around 18z and thought I would have to choose between south
central Oklahoma and southeast Kansas.  What I liked about southeast Kansas
was that low level shear looked more supportive of tornadoes when the LLJ
increased later.  850s closer to the Red River Valley had been progged at
nearly due west and weak, which made me nervous.  A cursory surface analysis
showed recovery ahead of the dryline after the morning showers and elevated
storms moved away. I expected decent instability in both potential targets,
but gave southeast Kansas the advantage for the improved helicity potential
and the possibility of some boundaries from earlier convection.  As it
stands I'm not sure our Kansas storms benefited from the latter.  Either
way, I wanted to be in Winfield by 21z.  

But some supercells got in the way.  I turned back south to intercept the
storm near Perry because I thought the Cimarron Turnpike's east and then
southeast curvature from I-35 down toward Morrison offered an exceptional
chance to flank the storm as it approached from the southwest.  Like Greg, I
too had imagined how I would negotiate storms moving forty-plus knots in
unfavorable terrain.  But reality was that as quickly as these mesos wrapped
up, you had to engage them more closely than ideal for "outbreak" style
navigational strategies.  The mesos on both storms seemed to move to the
forward flank more rapidly than normal.  Therefore each approach cost ten to
twelve minutes of "catch up" before I was in position to see anything again.
The net effect was that I never stepped out of my vehicle, never mounted
even so much as a window clamp tripod, and took only two still images and
thirty seconds of video all day.

On the Perry to Independence storm I witnessed a single, very elevated
funnel visible for one second less than it took to grab my camera before
another hill interrupted the view.  I was in the midst of repositioning
during the reported tornado near the Elk City Lake, KS.  Afterwards, on my
way north and west on SR 400 to meet the Elk County storm (which became
Wilson County / Buffalo tornado), I crossed paths with at least fifteen
emergency vehicles from various counties and townships.  

I assumed I would redeem myself on the Wilson County storm.  It was well to
the west and I'd learned my lesson.  But this storm produced several
interesting features, and seemed to make a slight right turn around Fall
River Lake which slowed its forward speed.  Of course this invited closer
scrutiny.  I observed a large elevated funnel that reminded me of the
beginning of the Attica, KS 5-12-04 tornado.  But the meso occluded so fast
that within seconds my view was obliterated.  I continued east on SR 39
until it merged with highway 75 and then I missed the continuation of 39 and
found myself southbound, reciting every obscenity I know.  In my rearview
mirror I saw the silhouette of a cone in the rain and turned into a church
parking lot to film in the direction just east of due north, or east of
Buffalo, Kansas by perhaps one mile.  Time was about 2354z.  It was clear
something was happening in there, though I never fully registered the
tornado with my eyes. 
I should be more reluctant to chase solo under those circumstances again.
It reminded me very much of May 8, 2003 in northeast Kansas.  Your chance to
see anything is already reduced by the hills and trees, further lowered by
the time spent catching up to the storm again and again, and even more by
requirements of driving and navigation.  It makes the odds quite long.