[BioC] limma block design

[Mayra Eduardoff]

Hi there,
I have another question regarding limma and doing a block design on
paired samples.
this is how I understand this design to work, maybe someone can tell
me If I got it right:

the function duplicateCorrelation calculates the mean correlation
between your blocked samples (in the variable $consensus) which can be
used as input for the correlation parameter in the lmFit function.
This variabel is then used for variance estimation in the denominator
in the t-statistik formula. Is the nominator of the t-Statistik paired
though (meaning the mean of the differences) ?
and what do I do with a really low $consenus value like 0.03 ?

best regards,

Mayra