The Adaptec solution was to turn off the mirroring and look at both drives individually. Both had the same corrupted data, as might be expected - so at least the RAID was working!
A bit more of a look round and a trawl through the event log showed that a checkdisk had run and 'fixed' a lot of indexing errors. The timing is consistent with this triggering the RAID resync, so now I think that RAID is a red herring and the real problem is the checkdisk, or a corruption that triggered it. I have sent the checkdisk log to Microsoft for suggestions to reverse the process and restore the data.
Some of the JPEG files have blocks of random pixels in the middle, which suggests that the file allocation table (or NTFS equivalent) is the problem, therefore, hopefully the file data fragments are still on the disk. Thanks to the mirroring it looks like I will have 2 chances - my feeling is to find a software package to try myself, and if that fails to find a service I can send the other disk to.
I would welcome recommendations in case Microsoft are unable to offer a solution - I am in the UK and the disk is NTFS format on Windows XP PRo SP2.
Thanks for the replies so far.