e:
dB
ations
lter
ne.
n't
.ive
tup
While the standard form of a twin-T filter isn't a band-pass, it's a band-reject, it can have a very high Q. I'm (almost) sure some clever person can find the right combination of terminals to yield a bandpass. Now if you limit your design to ladder networks, the proof might be possible. I have a vague recollection of a proof that you can't have coincident poles in a passive RC ladder network (but I could be wrong about that, it's been a long time...).