I discuss the possibility that a significant fraction of the extremely comm
on faint submm sources found in recent surveys are not in fact high-redshif
t galaxies, but actually local objects emitting only in the submm, with a t
emperature around 7 K. The majority of faint SCUBA sources clearly really a
re distant galaxies. However, if even a quarter or a third of the SCUBA sou
rces are actually local objects, the cosmological implications are signific
ant, as this would selectively remove the objects believed to be at z > 3.
Two hypotheses - very cold brown dwarfs and outer Solar system bodies - are
easily rejected. A third hypothesis - cold dark dusty gas clouds - is not
so easily dismissed. I show that the observational constraints on such a po
pulation - dynamical Limits on local missing matter, the FIR-mm background
and the absence of gross high-latitude extinction features - constrain the
mass of such objects to be in the mass range 0.1 to 10 Jupiter masses. On t
he assumption of virial equilibrium, their sizes are in the range 1-100 au,
with angular sizes around 0.1 arcsec. They would be completely opaque at v
isible and IR wavelengths. The characteristics deduced are closely similar
to those of the objects proposed by Walker & Wardle to explain 'extreme sca
ttering events' in quasar radio light curves, and which they propose fill t
he Galactic halo and explain halo dark matter. Indeed, at around 1 Jupiter
mass, the local population density would be similar to that in dark halo mo
dels. However, such objects, if they explain a large fraction of the SCUBA
submm sources, cannot extend through the halo without greatly exceeding the
FIR-mm background. Instead, I deduce the characteristic distance of the SC
UBA sources to be around 100 pc, consistent with being drawn from a disc po
pulation with a scaleheight of a few hundred pc. Possibly a 'Population II'
dustless version of such objects could exist in the halo. Regardless of th
e dark matter problem, the possible existence of such compact substellar bu
t non-degenerate objects is intriguing. Such objects should collapse on a v
ery short time-scale, but at such a low temperature that it is possible tha
t cosmic ray heating can maintain them in equilibrium. The main theoretical
objection is that such an equilibrium may be unstable on a thermal time-sc
ale. If however, such objects do exist, they may be seen as 'failed stars',
representing an alternative end-point to stars and brown dwarfs. It is pos
sible that they greatly outnumber both stars and brown dwarfs. The nearest
such object could be a fraction of a pc away. Several relatively simple obs
ervations could critically test this hypothesis.