(
) Just many sharp humps as in 
example, data can be fitted with >1 value of parameter.
(
) If several parameters, can be Gaussian in
some and flat in others if depends weakly (or not at all!) on them.
(
) Gaussians maybe distorted by effects due to
end of range.
), (
), or (
) don't apply and
distribution of 
isn't Gaussian, then one can only be
unhappy!