Problem with non-radioactive hybridization to GC-rich DNA--Repeat

[Steven Goldberg]

I'll try again--

I have been using Amersham's ECL Direct kit in hybridizations
using a 580 bp fragment from an organism with a high GC con-
tent (ca. 72%).  There is no problem with Southern blots to
chromosomal digests of the genome--I see a single band in each
instance.  However, when I use this fragment against a cosmid
library of the organism, the fragment binds strongly to vector
sequences, giving me a uniform background.  I have verified
this by showing that even under stringent washing conditions,
the fragment binds to the cosmid (which is derived from a 
Streptomyces plasmid, also very high GC).  I thought about
increasing the stringency of the initial hybridization by
reducing the salt concentration from the recommended 0.5 M, 
but Amersham said this wouldn't help much; you can't increase
the temperature of the hybridization since the enzymes in the
buffer will be inactivated.  Has anyone had a similar experi-
ence and solved it or are there any other non-radioactive kits
which may avoid the phenomonon I have been seeing?