matkat - you're right that this is a massive grey area, which is why the proper response to a conflict between two documents must be to refer back to the aircraft type certificate holder. Data acceptable to the administrator is not the same as data approved by the administrator. The AMM may be acceptable but it is not approved. On the other hand, modifications that alter the type certificate must be approved.
In general the AMM would be correct but once in service, modifications to aircraft result in changes to the maintenance documents to reflect those changes. The document for accomplishing the change necessarily precedes the maintenance manual revision and generally has direct or delegated approval from the regulator, so the SB should carry more weight. This is why Boeing told us on several occasions that the AMM is "provided for the information and advice of maintenance personnel". The Boeing IPC is not even published by Boeing - they delegate that task to a sub-contracted technical publications company.
Boeing advise that the original FAA approved drawings from which the AMM is drafted [maintenance tasks are drafted in full text as approved 'drawings'] are the only certification authority. Unfortunately, as they are produced and checked by human beings, these 'drawings' sometimes contain mistakes that were not detected during the approval process. The test schedule 'drawings' are revised when an SB is published but they are not easily available to operators - In fifteen years we have only been supplied with copies on two occasions, both in such circumstances as we are discussing here. While the AMM must remain the main document relied upon by maintenance personnel it should be used with caution - it is not as authoritative as many people give it credit.
So, where conflict such as that described in airbirduk's original post exists, Maintenance should refer any conflict between the AMM and other controlled data to Engineering for clarification by the type certificate holder, so they in turn can verify their approved data and give an authoritative reply. How to proceed while the problem is sorted out is a matter for Engineering, Quality Assurance or even the operator's local regulator, not maintenance personnel. A grey area indeed - there never is a final definitive answer, but that's what makes the back-room boys' job satisfying and worthwhile, is it not?
Yoda.