I have  come to  the following  conclusions by  examining the  damaged files:
 
1) No line has been added or deleted.
 
2) Actually, after  digging a  little deeper,  I saw that  moving 5  lines from
   line 175  in the raw image  file to line  169 completely cured the  file. In
   other words,  two blocks of 5  lines have been  SWAPPED in the file  with no
   loss of data.  I swear that LISTSERV  does not block DD card  input lines by
   any value, least of all 5!!
 
3) Idea: 5*80=  400. Size of a  default DMTNJI RSCS buffer?  400 bytes... Nodes
   affected? Everything down and including  SUVM. The question is: does anybody
   near SUVM install RSCS V2 recently (or suchlike?)
 
  Conclusion:  probably not  a LISTSERV  bug. Will  wait to  see if  it happens
again. Meanwhile  I'll re-issue  a PUT,  we have  three new  LISTSERVs: CUNYVM,
JPNSUT30 and LEHIIBM1.
 
  Re: my  reply to Niall: I'm  sorry I did  not resend the original  message. I
never use MAILBOOK,  it's too slow on  our system (paging -  40/sec today). The
problem was that  the STANFORD mailer keeps  sending MAIL files with  a lrecl >
80 and it causes an  EXECIO error 1 sending the file.  Version 1.5g will ignore
this error and proceed normally -- the mail standard is 80 chars/line.
 
  Eric