FM>> Yep, exactly same symptom (but I haven't tried your workaround).
 FM>> To my mind, *my* workaround (not zipping the indexes into the QWK)
 FM>> strongly suggests PKT2QWK as the culprit, 'cos MESSAGES is OK but
 FM>> the indexes aren't.

 RS> Yes, but it doesnt explain why processing the individual archive files of
 RS> PKTs each in their own run does not. THATs the bit of evidence which made
 RS> me think that PKTJOIN was the more likely culprit. Atleast in my case.

Was your case just after you became a point?  My system was in dire
straits at that time, and I was sending out invalid packets.  
PKTJOIN would have produced invalid packets too, via the GIGO logic
I include in all my programs.

 FM>> Sure, but I think the problem is in PKT2QWK.

 RS>> Without much evidence tho.

 FM>> The indexes are fucked but MESSAGES are OK.

 RS> That doesnt prove a damned thing about it being a bug in PKT2QWK.

 FM>> The indexes are created by PKT2QWK.

 RS> Using info which it gets elsewhere. So whatever affects that info
 RS> it uses, can be the cause of the dud NDX too.

Using info that PKT2QWK read.  The packets themselves don't have
ANY index data in there.  That means PKTJOIN is let off the hook
completely.

 FM>> Therefore PKT2QWK is fucked

 RS> Nope.

 FM>> (either internally or because it's grabbing some stuff I've left
 FM>> lying around). That's my evidence.

 RS> Its useless.
 
Frank's analysis is perfectly correct. 

BFN.  Paul.
@EOT:

--- Mksmsg