In this scenario the ``small'' 
is constrained
by the solar neutrino data, assuming matter-enhanced oscillations
[56]. Then the "large" can be used to explain either the
LSND results [57] or the atmospheric neutrino deficit [58], but not
both (unless the Kamiokande atmospheric data in the multi-GeV range
are dropped [59]).
If one takes the first option (Solar+LSND) and a mass
spectrum with two light neutrinos 
, plus a heavy neutrino
in the eV range, then the combined fit [56, 57] prefers a strong
mixing of 
with 
, with being essentially
a mixture of 
. High-sensitivity neutrino oscillation
searches in the 
channel are able to probe a
fraction of the LSND signal [57].
The second option (Solar+Atmospheric) leads one to a larger variety of mass-mixing solutions [56, 58] which are in the range of interest for long-baseline experiments.