While the theoretical case for observing polarization is strong, it is a
difficult experimental task to observe signals of the low level of several
K and below. Nonetheless, polarization experiments
have one potential advantage over temperature anisotropy experiments.
They can reduce atmospheric emission effects by differencing the polarization
states on the same patch of sky instead of physically chopping between
different angles on the sky since atmospheric emission is thought
to be nearly unpolarized (see §5.2).
However, to be successful an experiment must
overcome a number of systematic effects, many of which are
discussed in [Keating et al.] (1997). It must at least
balance the sensitivity of the instrument to the
orthogonal polarization
channels (including the far side lobes) to nearly
8 orders of magnitude.
Multiple levels of switching and a very careful design are minimum
requirements.
To date the experimental upper limits on polarization of the CMB have been
at least an order of magnitude larger than the theoretical expectations.
The original polarization limits go back to Penzias & Wilson (1965) who
set a limit of 100n the polarization of the CMB.
There have been several subsequent upper limits which have now reached the
level of 
K (see Tab. 1), about a factor of 5-10 above the
predicted levels for popular models (see Tab. 2).
