I believe that you could probably get a fairly effective
pattern-warping by using a non-tuned reflector which is significantly
longer than your collinear radiator.

Passive non-resonant elements have very little effect.
That's what allows a lot of multi-band beams to work.

As a point of reference, the radiation patterns of sidemounted FM broadcast
transmit arrays are affected by a non-resonant mounting structure (the
tower) -- especially for vertical polarization. Measured patterns from the
manufacturer's test ranges demonstrate this, as do the NEC-2 studies in
several of the papers on http://rfry.org .

Check... that's just the sort of situation I was thinking of.

The effect on the pattern isn't all that strong, though... only a few
dB. As one example, the old ARRL VHF book gives the plans for a
four-bay stacked dipole antenna system. With the dipoles placed on
four sides of the mast, the antenna system has a gain of about 6 dB
(they don't say dBi or dBd but I assume it must be the latter). With
all four dipoles on the same side of the mast, the pattern is said to
be a cardioid of about 9 dB gain.

They don't say how deep the back-side null is. The cardioid pattern
might have a high enough F/B ratio to help with the original poster's
situation (mountain-side reflection), or he might need to use a tuned
reflector of the 5%-longer-than-resonant variety to get a more
directional pattern.

--
Dave Platt AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
I do _not_ wish to receive unsolicited commercial email, and I will
boycott any company which has the gall to send me such ads!