It's been a while since I looked of the ISO spec (which I still can't believe I had to buy to read) -- but you can certainly infer by looking at legal characters laid out by LC. In reality -- only a handful of unprintable characters are technically allowed in a MARC record -- but you have to remember that when MARC was created -- it was for block reading -- and generally, early (and current) readers stop on hard breaks.
> -----Original Message-----
> From: Code for Libraries [mailto:[log in to unmask]] On Behalf Of
> Jonathan Rochkind
> Sent: Thursday, May 19, 2011 11:49 AM
> To: [log in to unmask]
> Subject: Re: [CODE4LIB] is this valid marc ?
> On 5/19/2011 2:33 PM, Reese, Terry wrote:
> > Jonathan,
> > Karen is correct -- CR/LF are invalid characters within a MARC record. This
> has nothing to do if the character is valid in the set -- the format itself doesn't
> allow it.
> I'm curious where in the spec it says this -- of course, it's an intellectual
> exersize at this point, because even if the spec says one thing, it doesn't
> matter if everyone (including tool-writers) has always understood it
> differently. (This is a problem for me with lots of library 'standards' including
> MARC. "Oh yeah, it might APPEAR to say/allow/prohibit that, but don't
> believe it, 'everyone' has always understood it diffferently." Or two parts of a
> spec which contradict each other).
> In the glossary here:
> It does say "Consequently,/code points/less than 80 (hex) have the same
> meaning in both of the encodings used in MARC 21 and may be referred to as
> ASCII in either environment." Which could be interpreted to include control
> chars such as CR and LF. (Thanks Dan Scott). Of course, the glossary section
> may not actually be an operative part of the standard, or it may not mean
> what it seems to mean, or everyone may have always acted as if it meant
> something different. Welcome to MARC.
> But I'm not succesfully finding anything else that says one way or another on
> the legality. Most of the ascii control chars do seem to be missing from Marc8
> (whether by design or accident), but that doesn't neccesarily mean they're
> illegal in a MARC record using some other (legal for MARC) encoding.
> But I believe Terry that it's not allowed (I believe Terry about just about
> everything). It's just really an intellectual exersize in the difficulty of finding
> answers in the MARC spec at the moment.