RFC 822
This document is obsolete. Please refer to RFC 2822.

     RFC #  822

     Obsoletes:  RFC #733  (NIC #41952)

                        STANDARD FOR THE FORMAT OF

                        ARPA INTERNET TEXT MESSAGES

                              August 13, 1982

                                Revised by

                             David H. Crocker

                      Dept. of Electrical Engineering
                 University of Delaware, Newark, DE  19711
                      Network:  DCrocker @ UDel-Relay

     Standard for ARPA Internet Text Messages

                             TABLE OF CONTENTS

     PREFACE ....................................................   ii

     1.  INTRODUCTION ...........................................    1

         1.1.  Scope ............................................    1
         1.2.  Communication Framework ..........................    2

     2.  NOTATIONAL CONVENTIONS .................................    3

     3.  LEXICAL ANALYSIS OF MESSAGES ...........................    5

         3.1.  General Description ..............................    5
         3.2.  Header Field Definitions .........................    9
         3.3.  Lexical Tokens ...................................   10
         3.4.  Clarifications ...................................   11

     4.  MESSAGE SPECIFICATION ..................................   17

         4.1.  Syntax ...........................................   17
         4.2.  Forwarding .......................................   19
         4.3.  Trace Fields .....................................   20
         4.4.  Originator Fields ................................   21
         4.5.  Receiver Fields ..................................   23
         4.6.  Reference Fields .................................   23
         4.7.  Other Fields .....................................   24

     5.  DATE AND TIME SPECIFICATION ............................   26

         5.1.  Syntax ...........................................   26
         5.2.  Semantics ........................................   26

     6.  ADDRESS SPECIFICATION ..................................   27

         6.1.  Syntax ...........................................   27
         6.2.  Semantics ........................................   27
         6.3.  Reserved Address .................................   33

     7.  BIBLIOGRAPHY ...........................................   34


     A.  EXAMPLES ...............................................   36
     B.  SIMPLE FIELD PARSING ...................................   40
     C.  DIFFERENCES FROM RFC #733 ..............................   41
     D.  ALPHABETICAL LISTING OF SYNTAX RULES ...................   44

     August 13, 1982               - i -                      RFC #822

     Standard for ARPA Internet Text Messages


          By 1977, the Arpanet employed several informal standards for
     the  text  messages (mail) sent among its host computers.  It was
     felt necessary to codify these practices and  provide  for  those
     features  that  seemed  imminent.   The result of that effort was
     Request for Comments (RFC) #733, "Standard for the Format of ARPA
     Network Text Message", by Crocker, Vittal, Pogran, and Henderson.
     The specification attempted to avoid major  changes  in  existing
     software, while permitting several new features.

          This document revises the specifications  in  RFC  #733,  in
     order  to  serve  the  needs  of the larger and more complex ARPA
     Internet.  Some of RFC #733's features failed  to  gain  adequate
     acceptance.   In  order to simplify the standard and the software
     that follows it, these features have been removed.   A  different
     addressing  scheme  is  used, to handle the case of inter-network
     mail; and the concept of re-transmission has been introduced.

          This specification is intended for use in the ARPA Internet.
     However, an attempt has been made to free it of any dependence on
     that environment, so that it can be applied to other network text
     message systems.

          The specification of RFC #733 took place over the course  of
     one  year, using the ARPANET mail environment, itself, to provide
     an on-going forum for discussing the capabilities to be included.
     More  than  twenty individuals, from across the country, partici-
     pated in  the  original  discussion.   The  development  of  this
     revised specification has, similarly, utilized network mail-based
     group discussion.  Both specification efforts  greatly  benefited
     from the comments and ideas of the participants.

          The syntax of the standard,  in  RFC  #733,  was  originally
     specified  in  the  Backus-Naur Form (BNF) meta-language.  Ken L.
     Harrenstien, of SRI International, was responsible for  re-coding
     the  BNF  into  an  augmented  BNF  that makes the representation
     smaller and easier to understand.

     August 13, 1982              - ii -                      RFC #822

     Standard for ARPA Internet Text Messages


     1.1.  SCOPE

          This standard specifies a syntax for text messages that  are
     sent  among  computer  users, within the framework of "electronic
     mail".  The standard supersedes  the  one  specified  in  ARPANET
     Request  for Comments #733, "Standard for the Format of ARPA Net-
     work Text Messages".

          In this context, messages are viewed as having  an  envelope
     and  contents.   The  envelope  contains  whatever information is
     needed to accomplish transmission  and  delivery.   The  contents
     compose  the object to be delivered to the recipient.  This stan-
     dard applies only to the format and some of the semantics of mes-
     sage  contents.   It contains no specification of the information
     in the envelope.

          However, some message systems may use information  from  the
     contents  to create the envelope.  It is intended that this stan-
     dard facilitate the acquisition of such information by programs.

          Some message systems may  store  messages  in  formats  that
     differ  from the one specified in this standard.  This specifica-
     tion is intended strictly as a definition of what message content
     format is to be passed BETWEEN hosts.

     Note:  This standard is NOT intended to dictate the internal for-
            mats  used  by sites, the specific message system features
            that they are expected to support, or any of  the  charac-
            teristics  of  user interface programs that create or read

          A distinction should be made between what the  specification
     REQUIRES  and  what  it ALLOWS.  Messages can be made complex and
     rich with formally-structured components of information or can be
     kept small and simple, with a minimum of such information.  Also,
     the standard simplifies the interpretation  of  differing  visual
     formats  in  messages;  only  the  visual  aspect of a message is
     affected and not the interpretation  of  information  within  it.
     Implementors may choose to retain such visual distinctions.

          The formal definition is divided into four levels.  The bot-
     tom level describes the meta-notation used in this document.  The
     second level describes basic lexical analyzers that  feed  tokens
     to  higher-level  parsers.   Next is an overall specification for
     messages; it permits distinguishing individual fields.   Finally,
     there is definition of the contents of several structured fields.

     August 13, 1982               - 1 -                      RFC #822

     Standard for ARPA Internet Text Messages


          Messages consist of lines of text.   No  special  provisions
     are  made for encoding drawings, facsimile, speech, or structured
     text.  No significant consideration has been given  to  questions
     of  data  compression  or to transmission and storage efficiency,
     and the standard tends to be free with the number  of  bits  con-
     sumed.   For  example,  field  names  are specified as free text,
     rather than special terse codes.

          A general "memo" framework is used.  That is, a message con-
     sists of some information in a rigid format, followed by the main
     part of the message, with a format that is not specified in  this
     document.   The  syntax of several fields of the rigidly-formated
     ("headers") section is defined in  this  specification;  some  of
     these fields must be included in all messages.

          The syntax  that  distinguishes  between  header  fields  is
     specified  separately  from  the  internal  syntax for particular
     fields.  This separation is intended to allow simple  parsers  to
     operate on the general structure of messages, without concern for
     the detailed structure of individual header fields.   Appendix  B
     is provided to facilitate construction of these parsers.

          In addition to the fields specified in this document, it  is
     expected  that  other fields will gain common use.  As necessary,
     the specifications for these "extension-fields" will be published
     through  the same mechanism used to publish this document.  Users
     may also  wish  to  extend  the  set  of  fields  that  they  use
     privately.  Such "user-defined fields" are permitted.

          The framework severely constrains document tone and  appear-
     ance and is primarily useful for most intra-organization communi-
     cations and  well-structured   inter-organization  communication.
     It  also  can  be used for some types of inter-process communica-
     tion, such as simple file transfer and remote job entry.  A  more
     robust  framework might allow for multi-font, multi-color, multi-
     dimension encoding of information.  A  less  robust  one,  as  is
     present  in  most  single-machine  message  systems,  would  more
     severely constrain the ability to add fields and the decision  to
     include specific fields.  In contrast with paper-based communica-
     tion, it is interesting to note that the RECEIVER  of  a  message
     can   exercise  an  extraordinary  amount  of  control  over  the
     message's appearance.  The amount of actual control available  to
     message  receivers  is  contingent upon the capabilities of their
     individual message systems.

     August 13, 1982               - 2 -                      RFC #822

     Standard for ARPA Internet Text Messages


          This specification uses an augmented Backus-Naur Form  (BNF)
     notation.  The differences from standard BNF involve naming rules
     and indicating repetition and "local" alternatives.

     2.1.  RULE NAMING

          Angle brackets ("<", ">") are not  used,  in  general.   The
     name  of  a rule is simply the name itself, rather than "<name>".
     Quotation-marks enclose literal text (which may be  upper  and/or
     lower  case).   Certain  basic  rules  are  in uppercase, such as
     SPACE, TAB, CRLF, DIGIT, ALPHA, etc.  Angle brackets are used  in
     rule  definitions,  and  in  the rest of this  document, whenever
     their presence will facilitate discerning the use of rule names.


          Elements separated by slash ("/") are alternatives.   There-
     fore "foo / bar" will accept foo or bar.


          Elements enclosed in parentheses are  treated  as  a  single
     element.   Thus,  "(elem  (foo  /  bar)  elem)"  allows the token
     sequences "elem foo elem" and "elem bar elem".

     2.4.  *RULE:  REPETITION

          The character "*" preceding an element indicates repetition.
     The full form is:


     indicating at least <l> and at most <m> occurrences  of  element.
     Default values are 0 and infinity so that "*(element)" allows any
     number, including zero; "1*element" requires at  least  one;  and
     "1*2element" allows one or two.

     2.5.  [RULE]:  OPTIONAL

          Square brackets enclose optional elements; "[foo  bar]"   is
     equivalent to "*1(foo bar)".


          "<n>(element)" is equivalent to "<n>*<n>(element)"; that is,
     exactly  <n>  occurrences  of (element). Thus 2DIGIT is a 2-digit
     number, and 3ALPHA is a string of three alphabetic characters.

     August 13, 1982               - 3 -                      RFC #822

     Standard for ARPA Internet Text Messages

     2.7.  #RULE:  LISTS

          A construct "#" is defined, similar to "*", as follows:


     indicating at least <l> and at most <m> elements, each  separated
     by  one  or more commas (","). This makes the usual form of lists
     very easy; a rule such as '(element *("," element))' can be shown
     as  "1#element".   Wherever this construct is used, null elements
     are allowed, but do not  contribute  to  the  count  of  elements
     present.   That  is,  "(element),,(element)"  is  permitted,  but
     counts as only two elements.  Therefore, where at least one  ele-
     ment  is required, at least one non-null element must be present.
     Default values are 0 and infinity so that "#(element)" allows any
     number,  including  zero;  "1#element" requires at least one; and
     "1#2element" allows one or two.

     2.8.  ; COMMENTS

          A semi-colon, set off some distance to  the  right  of  rule
     text,  starts  a comment that continues to the end of line.  This
     is a simple way of including useful notes in  parallel  with  the

     August 13, 1982               - 4 -                      RFC #822

     Standard for ARPA Internet Text Messages



          A message consists of header fields and, optionally, a body.
     The  body  is simply a sequence of lines containing ASCII charac-
     ters.  It is separated from the headers by a null line  (i.e.,  a
     line with nothing preceding the CRLF).


        Each header field can be viewed as a single, logical  line  of
        ASCII  characters,  comprising  a field-name and a field-body.
        For convenience, the field-body  portion  of  this  conceptual
        entity  can be split into a multiple-line representation; this
        is called "folding".  The general rule is that wherever  there
        may  be  linear-white-space  (NOT  simply  LWSP-chars), a CRLF
        immediately followed by AT LEAST one LWSP-char may instead  be
        inserted.  Thus, the single line

            To:  "Joe & J. Harvey" <ddd @Org>, JJV @ BBN

        can be represented as:

            To:  "Joe & J. Harvey" <ddd @ Org>,


            To:  "Joe & J. Harvey"
                            <ddd@ Org>, JJV


            To:  "Joe &
             J. Harvey" <ddd @ Org>, JJV @ BBN

             The process of moving  from  this  folded   multiple-line
        representation  of a header field to its single line represen-
        tation is called "unfolding".  Unfolding  is  accomplished  by
        regarding   CRLF   immediately  followed  by  a  LWSP-char  as
        equivalent to the LWSP-char.

        Note:  While the standard  permits  folding  wherever  linear-
               white-space is permitted, it is recommended that struc-
               tured fields, such as those containing addresses, limit
               folding  to higher-level syntactic breaks.  For address
               fields, it  is  recommended  that  such  folding  occur

     August 13, 1982               - 5 -                      RFC #822

     Standard for ARPA Internet Text Messages

               between addresses, after the separating comma.


        Once a field has been unfolded, it may be viewed as being com-
        posed of a field-name followed by a colon (":"), followed by a
        field-body, and  terminated  by  a  carriage-return/line-feed.
        The  field-name must be composed of printable ASCII characters
        (i.e., characters that  have  values  between  33.  and  126.,
        decimal, except colon).  The field-body may be composed of any
        ASCII characters, except CR or LF.  (While CR and/or LF may be
        present  in the actual text, they are removed by the action of
        unfolding the field.)

        Certain field-bodies of headers may be  interpreted  according
        to  an  internal  syntax  that some systems may wish to parse.
        These  fields  are  called  "structured   fields".    Examples
        include  fields containing dates and addresses.  Other fields,
        such as "Subject"  and  "Comments",  are  regarded  simply  as
        strings of text.

        Note:  Any field which has a field-body  that  is  defined  as
               other  than  simply <text> is to be treated as a struc-
               tured field.

               Field-names, unstructured field bodies  and  structured
               field bodies each are scanned by their own, independent
               "lexical" analyzers.


        For some fields, such as "Subject" and "Comments",  no  struc-
        turing  is assumed, and they are treated simply as <text>s, as
        in the message body.  Rules of folding apply to these  fields,
        so  that  such  field  bodies  which occupy several lines must
        therefore have the second and successive lines indented by  at
        least one LWSP-char.


        To aid in the creation and reading of structured  fields,  the
        free  insertion   of linear-white-space (which permits folding
        by inclusion of CRLFs)  is  allowed  between  lexical  tokens.
        Rather  than  obscuring  the  syntax  specifications for these
        structured fields with explicit syntax for this  linear-white-
        space, the existence of another "lexical" analyzer is assumed.
        This analyzer does not apply  for  unstructured  field  bodies
        that  are  simply  strings  of  text, as described above.  The
        analyzer provides  an  interpretation  of  the  unfolded  text

     August 13, 1982               - 6 -                      RFC #822

     Standard for ARPA Internet Text Messages

        composing  the body of the field as a sequence of lexical sym-

        These symbols are:

                     -  individual special characters
                     -  quoted-strings
                     -  domain-literals
                     -  comments
                     -  atoms

        The first four of these symbols  are  self-delimiting.   Atoms
        are not; they are delimited by the self-delimiting symbols and
        by  linear-white-space.   For  the  purposes  of  regenerating
        sequences  of  atoms  and quoted-strings, exactly one SPACE is
        assumed to exist, and should be used, between them.  (Also, in
        the "Clarifications" section on "White Space", below, note the
        rules about treatment of multiple contiguous LWSP-chars.)

        So, for example, the folded body of an address field

            ":sysmail"@  Some-Group. Some-Org,
            Muhammed.(I am  the greatest) Ali @(the)Vegas.WBA

     August 13, 1982               - 7 -                      RFC #822

     Standard for ARPA Internet Text Messages

        is analyzed into the following lexical symbols and types:

                    :sysmail              quoted string
                    @                     special
                    Some-Group            atom
                    .                     special
                    Some-Org              atom
                    ,                     special
                    Muhammed              atom
                    .                     special
                    (I am  the greatest)  comment
                    Ali                   atom
                    @                     atom
                    (the)                 comment
                    Vegas                 atom
                    .                     special
                    WBA                   atom

        The canonical representations for the data in these  addresses
        are the following strings:




        Note:  For purposes of display, and when passing  such  struc-
               tured information to other systems, such as mail proto-
               col  services,  there  must  be  NO  linear-white-space
               between  <word>s  that are separated by period (".") or
               at-sign ("@") and exactly one SPACE between  all  other
               <word>s.  Also, headers should be in a folded form.

     August 13, 1982               - 8 -                      RFC #822

     Standard for ARPA Internet Text Messages


          These rules show a field meta-syntax, without regard for the
     particular  type  or internal syntax.  Their purpose is to permit
     detection of fields; also, they present to  higher-level  parsers
     an image of each field as fitting on one line.

     field       =  field-name ":" [ field-body ] CRLF

     field-name  =  1*<any CHAR, excluding CTLs, SPACE, and ":">

     field-body  =  field-body-contents
                    [CRLF LWSP-char field-body]

     field-body-contents =
                   <the ASCII characters making up the field-body, as
                    defined in the following sections, and consisting
                    of combinations of atom, quoted-string, and
                    specials tokens, or else consisting of texts>

     August 13, 1982               - 9 -                      RFC #822

     Standard for ARPA Internet Text Messages


          The following rules are used to define an underlying lexical
     analyzer,  which  feeds  tokens to higher level parsers.  See the
     ANSI references, in the Bibliography.

                                                 ; (  Octal, Decimal.)
     CHAR        =  <any ASCII character>        ; (  0-177,  0.-127.)
     ALPHA       =  <any ASCII alphabetic character>
                                                 ; (101-132, 65.- 90.)
                                                 ; (141-172, 97.-122.)
     DIGIT       =  <any ASCII decimal digit>    ; ( 60- 71, 48.- 57.)
     CTL         =  <any ASCII control           ; (  0- 37,  0.- 31.)
                     character and DEL>          ; (    177,     127.)
     CR          =  <ASCII CR, carriage return>  ; (     15,      13.)
     LF          =  <ASCII LF, linefeed>         ; (     12,      10.)
     SPACE       =  <ASCII SP, space>            ; (     40,      32.)
     HTAB        =  <ASCII HT, horizontal-tab>   ; (     11,       9.)
     <">         =  <ASCII quote mark>           ; (     42,      34.)
     CRLF        =  CR LF

     LWSP-char   =  SPACE / HTAB                 ; semantics = SPACE

     linear-white-space =  1*([CRLF] LWSP-char)  ; semantics = SPACE
                                                 ; CRLF => folding

     specials    =  "(" / ")" / "<" / ">" / "@"  ; Must be in quoted-
                 /  "," / ";" / ":" / "\" / <">  ;  string, to use
                 /  "." / "[" / "]"              ;  within a word.

     delimiters  =  specials / linear-white-space / comment

     text        =  <any CHAR, including bare    ; => atoms, specials,
                     CR & bare LF, but NOT       ;  comments and
                     including CRLF>             ;  quoted-strings are
                                                 ;  NOT recognized.

     atom        =  1*<any CHAR except specials, SPACE and CTLs>

     quoted-string = <"> *(qtext/quoted-pair) <">; Regular qtext or
                                                 ;   quoted chars.

     qtext       =  <any CHAR excepting <">,     ; => may be folded
                     "\" & CR, and including

     domain-literal =  "[" *(dtext / quoted-pair) "]"

     August 13, 1982              - 10 -                      RFC #822

     Standard for ARPA Internet Text Messages

     dtext       =  <any CHAR excluding "[",     ; => may be folded
                     "]", "\" & CR, & including

     comment     =  "(" *(ctext / quoted-pair / comment) ")"

     ctext       =  <any CHAR excluding "(",     ; => may be folded
                     ")", "\" & CR, & including

     quoted-pair =  "\" CHAR                     ; may quote any char

     phrase      =  1*word                       ; Sequence of words

     word        =  atom / quoted-string


     3.4.1.  QUOTING

        Some characters are reserved for special interpretation,  such
        as  delimiting lexical tokens.  To permit use of these charac-
        ters as uninterpreted data, a quoting mechanism  is  provided.
        To quote a character, precede it with a backslash ("\").

        This mechanism is not fully general.  Characters may be quoted
        only  within  a subset of the lexical constructs.  In particu-
        lar, quoting is limited to use within:

                             -  quoted-string
                             -  domain-literal
                             -  comment

        Within these constructs, quoting is REQUIRED for  CR  and  "\"
        and for the character(s) that delimit the token (e.g., "(" and
        ")" for a comment).  However, quoting  is  PERMITTED  for  any

        Note:  In particular, quoting is NOT permitted  within  atoms.
               For  example  when  the local-part of an addr-spec must
               contain a special character, a quoted  string  must  be
               used.  Therefore, a specification such as:

                            Full\ Name@Domain

               is not legal and must be specified as:

                            "Full Name"@Domain

     August 13, 1982              - 11 -                      RFC #822

     Standard for ARPA Internet Text Messages

     3.4.2.  WHITE SPACE

        Note:  In structured field bodies, multiple linear space ASCII
               characters  (namely  HTABs  and  SPACEs) are treated as
               single spaces and may freely surround any  symbol.   In
               all header fields, the only place in which at least one
               LWSP-char is REQUIRED is at the beginning of  continua-
               tion lines in a folded field.

        When passing text to processes  that  do  not  interpret  text
        according to this standard (e.g., mail protocol servers), then
        NO linear-white-space characters should occur between a period
        (".") or at-sign ("@") and a <word>.  Exactly ONE SPACE should
        be used in place of arbitrary linear-white-space  and  comment

        Note:  Within systems conforming to this standard, wherever  a
               member of the list of delimiters is allowed, LWSP-chars
               may also occur before and/or after it.

        Writers of  mail-sending  (i.e.,  header-generating)  programs
        should realize that there is no network-wide definition of the
        effect of ASCII HT (horizontal-tab) characters on the  appear-
        ance  of  text  at another network host; therefore, the use of
        tabs in message headers, though permitted, is discouraged.


        A comment is a set of ASCII characters, which is  enclosed  in
        matching  parentheses  and which is not within a quoted-string
        The comment construct permits message originators to add  text
        which  will  be  useful  for  human readers, but which will be
        ignored by the formal semantics.  Comments should be  retained
        while  the  message  is subject to interpretation according to
        this standard.  However, comments  must  NOT  be  included  in
        other  cases,  such  as  during  protocol  exchanges with mail

        Comments nest, so that if an unquoted left parenthesis  occurs
        in  a  comment  string,  there  must  also be a matching right
        parenthesis.  When a comment acts as the delimiter  between  a
        sequence of two lexical symbols, such as two atoms, it is lex-
        ically equivalent with a single SPACE,  for  the  purposes  of
        regenerating  the  sequence, such as when passing the sequence
        onto a mail protocol server.  Comments are  detected  as  such
        only within field-bodies of structured fields.

        If a comment is to be "folded" onto multiple lines,  then  the
        syntax  for  folding  must  be  adhered to.  (See the "Lexical

     August 13, 1982              - 12 -                      RFC #822

     Standard for ARPA Internet Text Messages

        Analysis of Messages" section on "Folding Long Header  Fields"
        above,  and  the  section on "Case Independence" below.)  Note
        that  the  official  semantics  therefore  do  not  "see"  any
        unquoted CRLFs that are in comments, although particular pars-
        ing programs may wish to note their presence.  For these  pro-
        grams,  it would be reasonable to interpret a "CRLF LWSP-char"
        as being a CRLF that is part of the comment; i.e., the CRLF is
        kept  and  the  LWSP-char is discarded.  Quoted CRLFs (i.e., a
        backslash followed by a CR followed by a  LF)  still  must  be
        followed by at least one LWSP-char.


        The quote character (backslash) and  characters  that  delimit
        syntactic  units  are not, generally, to be taken as data that
        are part of the delimited or quoted unit(s).   In  particular,
        the   quotation-marks   that   define   a  quoted-string,  the
        parentheses that define  a  comment  and  the  backslash  that
        quotes  a  following  character  are  NOT  part of the quoted-
        string, comment or quoted character.  A quotation-mark that is
        to  be  part  of  a quoted-string, a parenthesis that is to be
        part of a comment and a backslash that is to be part of either
        must  each be preceded by the quote-character backslash ("\").
        Note that the syntax allows any character to be quoted  within
        a  quoted-string  or  comment; however only certain characters
        MUST be quoted to be included as data.  These  characters  are
        the  ones that are not part of the alternate text group (i.e.,
        ctext or qtext).

        The one exception to this rule  is  that  a  single  SPACE  is
        assumed  to  exist  between  contiguous words in a phrase, and
        this interpretation is independent of  the  actual  number  of
        LWSP-chars  that  the  creator  places  between the words.  To
        include more than one SPACE, the creator must make  the  LWSP-
        chars be part of a quoted-string.

        Quotation marks that delimit a quoted string  and  backslashes
        that  quote  the  following character should NOT accompany the
        quoted-string when the string is passed to processes  that  do
        not interpret data according to this specification (e.g., mail
        protocol servers).

     3.4.5.  QUOTED-STRINGS

        Where permitted (i.e., in words in structured fields)  quoted-
        strings  are  treated  as a single symbol.  That is, a quoted-
        string is equivalent to an atom, syntactically.  If a  quoted-
        string  is to be "folded" onto multiple lines, then the syntax
        for folding must be adhered to.  (See the "Lexical Analysis of

     August 13, 1982              - 13 -                      RFC #822

     Standard for ARPA Internet Text Messages

        Messages"  section  on "Folding Long Header Fields" above, and
        the section on "Case  Independence"  below.)   Therefore,  the
        official  semantics  do  not  "see" any bare CRLFs that are in
        quoted-strings; however particular parsing programs  may  wish
        to  note  their presence.  For such programs, it would be rea-
        sonable to interpret a "CRLF LWSP-char" as being a CRLF  which
        is  part  of the quoted-string; i.e., the CRLF is kept and the
        LWSP-char is discarded.  Quoted CRLFs (i.e., a backslash  fol-
        lowed  by  a CR followed by a LF) are also subject to rules of
        folding, but the presence of the quoting character (backslash)
        explicitly  indicates  that  the  CRLF  is  data to the quoted
        string.  Stripping off the first following LWSP-char  is  also
        appropriate when parsing quoted CRLFs.


        There is one type of bracket which must occur in matched pairs
        and may have pairs nested within each other:

            o   Parentheses ("(" and ")") are used  to  indicate  com-

        There are three types of brackets which must occur in  matched
        pairs, and which may NOT be nested:

            o   Colon/semi-colon (":" and ";") are   used  in  address
                specifications  to  indicate that the included list of
                addresses are to be treated as a group.

            o   Angle brackets ("<" and ">")  are  generally  used  to
                indicate  the  presence of a one machine-usable refer-
                ence (e.g., delimiting mailboxes), possibly  including
                source-routing to the machine.

            o   Square brackets ("[" and "]") are used to indicate the
                presence  of  a  domain-literal, which the appropriate
                name-domain  is  to  use  directly,  bypassing  normal
                name-resolution mechanisms.


        Except as noted, alphabetic strings may be represented in  any
        combination of upper and lower case.  The only syntactic units

     August 13, 1982              - 14 -                      RFC #822

     Standard for ARPA Internet Text Messages

        which requires preservation of case information are:

                    -  text
                    -  qtext
                    -  dtext
                    -  ctext
                    -  quoted-pair
                    -  local-part, except "Postmaster"

        When matching any other syntactic unit, case is to be ignored.
        For  example, the field-names "From", "FROM", "from", and even
        "FroM" are semantically equal and should all be treated ident-

        When generating these units, any mix of upper and  lower  case
        alphabetic  characters  may  be  used.  The case shown in this
        specification is suggested for message-creating processes.

        Note:  The reserved local-part address unit, "Postmaster",  is
               an  exception.   When  the  value "Postmaster" is being
               interpreted, it must be  accepted  in  any  mixture  of
               case, including "POSTMASTER", and "postmaster".


        Each header field may be represented on exactly one line  con-
        sisting  of the name of the field and its body, and terminated
        by a CRLF; this is what the parser sees.  For readability, the
        field-body  portion of long header fields may be "folded" onto
        multiple lines of the actual field.  "Long" is commonly inter-
        preted  to  mean greater than 65 or 72 characters.  The former
        length serves as a limit, when the message is to be viewed  on
        most  simple terminals which use simple display software; how-
        ever, the limit is not imposed by this standard.

        Note:  Some display software often can selectively fold lines,
               to  suit  the display terminal.  In such cases, sender-
               provided  folding  can  interfere  with   the   display


        ASCII BS characters (Backspace, decimal 8) may be included  in
        texts and quoted-strings to effect overstriking.  However, any
        use of backspaces which effects an overstrike to the  left  of
        the beginning of the text or quoted-string is prohibited.

     August 13, 1982              - 15 -                      RFC #822

     Standard for ARPA Internet Text Messages


        During transmission through heterogeneous networks, it may  be
        necessary  to  force data to conform to a network's local con-
        ventions.  For example, it may be required that a CR  be  fol-
        lowed  either by LF, making a CRLF, or by <null>, if the CR is
        to stand alone).  Such transformations are reversed, when  the
        message exits that network.

        When  crossing  network  boundaries,  the  message  should  be
        treated  as  passing  through  two modules.  It will enter the
        first module containing whatever network-specific  transforma-
        tions  that  were  necessary  to  permit migration through the
        "current" network.  It then passes through the modules:

            o   Transformation Reversal

                The "current" network's idiosyncracies are removed and
                the  message  is returned to the canonical form speci-
                fied in this standard.

            o   Transformation

                The "next" network's local idiosyncracies are  imposed
                on the message.

                    From   ==>  | Remove Net-A   |
                    Net-A       | idiosyncracies |
                                  with standard
                                | Impose Net-B   |  ==>  To
                                | idiosyncracies |       Net-B

     August 13, 1982              - 16 -                      RFC #822

     Standard for ARPA Internet Text Messages


     4.1.  SYNTAX

     Note:  Due to an artifact of the notational conventions, the syn-
            tax  indicates that, when present, some fields, must be in
            a particular order.  Header fields  are  NOT  required  to
            occur  in  any  particular  order, except that the message
            body must occur AFTER  the  headers.   It  is  recommended
            that,  if  present,  headers be sent in the order "Return-
            Path", "Received", "Date",  "From",  "Subject",  "Sender",
            "To", "cc", etc.

            This specification permits multiple  occurrences  of  most
            fields.   Except  as  noted,  their  interpretation is not
            specified here, and their use is discouraged.

          The following syntax for the bodies of various fields should
     be  thought  of  as  describing  each field body as a single long
     string (or line).  The "Lexical Analysis of Message"  section  on
     "Long  Header Fields", above, indicates how such long strings can
     be represented on more than one line in  the  actual  transmitted

     message     =  fields *( CRLF *text )       ; Everything after
                                                 ;  first null line
                                                 ;  is message body

     fields      =    dates                      ; Creation time,
                      source                     ;  author id & one
                    1*destination                ;  address required
                     *optional-field             ;  others optional

     source      = [  trace ]                    ; net traversals
                      originator                 ; original mail
                   [  resent ]                   ; forwarded

     trace       =    return                     ; path to sender
                    1*received                   ; receipt tags

     return      =  "Return-path" ":" route-addr ; return address

     received    =  "Received"    ":"            ; one per relay
                       ["from" domain]           ; sending host
                       ["by"   domain]           ; receiving host
                       ["via"  atom]             ; physical path
                      *("with" atom)             ; link/mail protocol
                       ["id"   msg-id]           ; receiver msg id
                       ["for"  addr-spec]        ; initial form

     August 13, 1982              - 17 -                      RFC #822

     Standard for ARPA Internet Text Messages

                        ";"    date-time         ; time received

     originator  =   authentic                   ; authenticated addr
                   [ "Reply-To"   ":" 1#address] )

     authentic   =   "From"       ":"   mailbox  ; Single author
                 / ( "Sender"     ":"   mailbox  ; Actual submittor
                     "From"       ":" 1#mailbox) ; Multiple authors
                                                 ;  or not sender

     resent      =   resent-authentic
                   [ "Resent-Reply-To"  ":" 1#address] )

     resent-authentic =
                 =   "Resent-From"      ":"   mailbox
                 / ( "Resent-Sender"    ":"   mailbox
                     "Resent-From"      ":" 1#mailbox  )

     dates       =   orig-date                   ; Original
                   [ resent-date ]               ; Forwarded

     orig-date   =  "Date"        ":"   date-time

     resent-date =  "Resent-Date" ":"   date-time

     destination =  "To"          ":" 1#address  ; Primary
                 /  "Resent-To"   ":" 1#address
                 /  "cc"          ":" 1#address  ; Secondary
                 /  "Resent-cc"   ":" 1#address
                 /  "bcc"         ":"  #address  ; Blind carbon
                 /  "Resent-bcc"  ":"  #address

     optional-field =
                 /  "Message-ID"        ":"   msg-id
                 /  "Resent-Message-ID" ":"   msg-id
                 /  "In-Reply-To"       ":"  *(phrase / msg-id)
                 /  "References"        ":"  *(phrase / msg-id)
                 /  "Keywords"          ":"  #phrase
                 /  "Subject"           ":"  *text
                 /  "Comments"          ":"  *text
                 /  "Encrypted"         ":" 1#2word
                 /  extension-field              ; To be defined
                 /  user-defined-field           ; May be pre-empted

     msg-id      =  "<" addr-spec ">"            ; Unique message id

     August 13, 1982              - 18 -                      RFC #822

     Standard for ARPA Internet Text Messages

     extension-field =
                   <Any field which is defined in a document
                    published as a formal extension to this
                    specification; none will have names beginning
                    with the string "X-">

     user-defined-field =
                   <Any field which has not been defined
                    in this specification or published as an
                    extension to this specification; names for
                    such fields must be unique and may be
                    pre-empted by published extensions>

     4.2.  FORWARDING

          Some systems permit mail recipients to  forward  a  message,
     retaining  the original headers, by adding some new fields.  This
     standard supports such a service, through the "Resent-" prefix to
     field names.

          Whenever the string "Resent-" begins a field name, the field
     has  the  same  semantics as a field whose name does not have the
     prefix.  However, the message is assumed to have  been  forwarded
     by  an original recipient who attached the "Resent-" field.  This
     new field is treated as being more recent  than  the  equivalent,
     original  field.   For  example, the "Resent-From", indicates the
     person that forwarded the message, whereas the "From" field indi-
     cates the original author.

          Use of such precedence  information  depends  upon  partici-
     pants'  communication needs.  For example, this standard does not
     dictate when a "Resent-From:" address should receive replies,  in
     lieu of sending them to the "From:" address.

     Note:  In general, the "Resent-" fields should be treated as con-
            taining  a  set  of information that is independent of the
            set of original fields.  Information for  one  set  should
            not  automatically be taken from the other.  The interpre-
            tation of multiple "Resent-" fields, of the same type,  is

          In the remainder of this specification, occurrence of  legal
     "Resent-"  fields  are treated identically with the occurrence of

     August 13, 1982              - 19 -                      RFC #822

     Standard for ARPA Internet Text Messages

     fields whose names do not contain this prefix.

     4.3.  TRACE FIELDS

          Trace information is used to provide an audit trail of  mes-
     sage  handling.   In  addition,  it indicates a route back to the
     sender of the message.

          The list of known "via" and  "with"  values  are  registered
     with  the  Network  Information  Center, SRI International, Menlo
     Park, California.

     4.3.1.  RETURN-PATH

        This field  is  added  by  the  final  transport  system  that
        delivers  the message to its recipient.  The field is intended
        to contain definitive information about the address and  route
        back to the message's originator.

        Note:  The "Reply-To" field is added  by  the  originator  and
               serves  to  direct  replies,  whereas the "Return-Path"
               field is used to identify a path back to  the  origina-

        While the syntax  indicates  that  a  route  specification  is
        optional,  every attempt should be made to provide that infor-
        mation in this field.

     4.3.2.  RECEIVED

        A copy of this field is added by each transport  service  that
        relays the message.  The information in the field can be quite
        useful for tracing transport problems.

        The names of the sending  and  receiving  hosts  and  time-of-
        receipt may be specified.  The "via" parameter may be used, to
        indicate what physical mechanism the message  was  sent  over,
        such  as  Arpanet or Phonenet, and the "with" parameter may be
        used to indicate the mail-,  or  connection-,  level  protocol
        that  was  used, such as the SMTP mail protocol, or X.25 tran-
        sport protocol.

        Note:  Several "with" parameters may  be  included,  to  fully
               specify the set of protocols that were used.

        Some transport services queue mail; the internal message iden-
        tifier that is assigned to the message may be noted, using the
        "id" parameter.  When the  sending  host  uses  a  destination
        address specification that the receiving host reinterprets, by

     August 13, 1982              - 20 -                      RFC #822

     Standard for ARPA Internet Text Messages

        expansion or transformation, the receiving host  may  wish  to
        record  the original specification, using the "for" parameter.
        For example, when a copy of mail is sent to the  member  of  a
        distribution  list,  this  parameter may be used to record the
        original address that was used to specify the list.


          The standard allows only a subset of the combinations possi-
     ble  with the From, Sender, Reply-To, Resent-From, Resent-Sender,
     and Resent-Reply-To fields.  The limitation is intentional.

     4.4.1.  FROM / RESENT-FROM

        This field contains the identity of the person(s)  who  wished
        this  message to be sent.  The message-creation process should
        default this field  to  be  a  single,  authenticated  machine
        address,  indicating  the  AGENT  (person,  system or process)
        entering the message.  If this is not done, the "Sender" field
        MUST  be  present.  If the "From" field IS defaulted this way,
        the "Sender" field is  optional  and  is  redundant  with  the
        "From"  field.   In  all  cases, addresses in the "From" field
        must be machine-usable (addr-specs) and may not contain  named
        lists (groups).


        This field contains the authenticated identity  of  the  AGENT
        (person,  system  or  process)  that sends the message.  It is
        intended for use when the sender is not the author of the mes-
        sage,  or  to  indicate  who among a group of authors actually
        sent the message.  If the contents of the "Sender" field would
        be  completely  redundant  with  the  "From"  field,  then the
        "Sender" field need not be present and its use is  discouraged
        (though  still legal).  In particular, the "Sender" field MUST
        be present if it is NOT the same as the "From" Field.

        The Sender mailbox  specification  includes  a  word  sequence
        which  must correspond to a specific agent (i.e., a human user
        or a computer program) rather than a standard  address.   This
        indicates  the  expectation  that  the field will identify the
        single AGENT (person,  system,  or  process)  responsible  for
        sending  the mail and not simply include the name of a mailbox
        from which the mail was sent.  For example in the  case  of  a
        shared login name, the name, by itself, would not be adequate.
        The local-part address unit, which refers to  this  agent,  is
        expected to be a computer system term, and not (for example) a
        generalized person reference which can  be  used  outside  the
        network text message context.

     August 13, 1982              - 21 -                      RFC #822

     Standard for ARPA Internet Text Messages

        Since the critical function served by the  "Sender"  field  is
        identification  of  the agent responsible for sending mail and
        since computer programs cannot be held accountable  for  their
        behavior, it is strongly recommended that when a computer pro-
        gram generates a message, the HUMAN  who  is  responsible  for
        that program be referenced as part of the "Sender" field mail-
        box specification.


        This field provides a general  mechanism  for  indicating  any
        mailbox(es)  to which responses are to be sent.  Three typical
        uses for this feature can  be  distinguished.   In  the  first
        case,  the  author(s) may not have regular machine-based mail-
        boxes and therefore wish(es) to indicate an alternate  machine
        address.   In  the  second case, an author may wish additional
        persons to be made aware of, or responsible for,  replies.   A
        somewhat  different  use  may be of some help to "text message
        teleconferencing" groups equipped with automatic  distribution
        services:   include the address of that service in the "Reply-
        To" field of all messages  submitted  to  the  teleconference;
        then  participants  can  "reply"  to conference submissions to
        guarantee the correct distribution of any submission of  their

        Note:  The "Return-Path" field is added by the mail  transport
               service,  at the time of final deliver.  It is intended
               to identify a path back to the orginator  of  the  mes-
               sage.   The  "Reply-To"  field  is added by the message
               originator and is intended to direct replies.


        For systems which automatically  generate  address  lists  for
        replies to messages, the following recommendations are made:

            o   The "Sender" field mailbox should be sent  notices  of
                any  problems in transport or delivery of the original
                messages.  If there is no  "Sender"  field,  then  the
                "From" field mailbox should be used.

            o   The  "Sender"  field  mailbox  should  NEVER  be  used
                automatically, in a recipient's reply message.

            o   If the "Reply-To" field exists, then the reply  should
                go to the addresses indicated in that field and not to
                the address(es) indicated in the "From" field.

     August 13, 1982              - 22 -                      RFC #822

     Standard for ARPA Internet Text Messages

            o   If there is a "From" field, but no  "Reply-To"  field,
                the  reply should be sent to the address(es) indicated
                in the "From" field.

        Sometimes, a recipient may actually wish to  communicate  with
        the  person  that  initiated  the  message  transfer.  In such
        cases, it is reasonable to use the "Sender" address.

        This recommendation is intended  only  for  automated  use  of
        originator-fields  and is not intended to suggest that replies
        may not also be sent to other recipients of messages.   It  is
        up  to  the  respective  mail-handling programs to decide what
        additional facilities will be provided.

        Examples are provided in Appendix A.


     4.5.1.  TO / RESENT-TO

        This field contains the identity of the primary recipients  of
        the message.

     4.5.2.  CC / RESENT-CC

        This field contains the identity of  the  secondary  (informa-
        tional) recipients of the message.

     4.5.3.  BCC / RESENT-BCC

        This field contains the identity of additional  recipients  of
        the  message.   The contents of this field are not included in
        copies of the message sent to the primary and secondary  reci-
        pients.   Some  systems  may choose to include the text of the
        "Bcc" field only in the author(s)'s  copy,  while  others  may
        also include it in the text sent to all those indicated in the
        "Bcc" list.



             This field contains a unique identifier  (the  local-part
        address  unit)  which  refers to THIS version of THIS message.
        The uniqueness of the message identifier is guaranteed by  the
        host  which  generates  it.  This identifier is intended to be
        machine readable and not necessarily meaningful to humans.   A
        message  identifier pertains to exactly one instantiation of a
        particular message; subsequent revisions to the message should

     August 13, 1982              - 23 -                      RFC #822

     Standard for ARPA Internet Text Messages

        each receive new message identifiers.

     4.6.2.  IN-REPLY-TO

             The contents of this field identify  previous  correspon-
        dence  which this message answers.  Note that if message iden-
        tifiers are used in this  field,  they  must  use  the  msg-id
        specification format.

     4.6.3.  REFERENCES

             The contents of this field identify other  correspondence
        which  this message references.  Note that if message identif-
        iers are used, they must use the msg-id specification format.

     4.6.4.  KEYWORDS

             This field contains keywords  or  phrases,  separated  by

     4.7.  OTHER FIELDS

     4.7.1.  SUBJECT

             This is intended to provide a summary,  or  indicate  the
        nature, of the message.


             Permits adding text comments  onto  the  message  without
        disturbing the contents of the message's body.

     4.7.3.  ENCRYPTED

             Sometimes,  data  encryption  is  used  to  increase  the
        privacy  of  message  contents.   If the body of a message has
        been encrypted, to keep its contents private, the  "Encrypted"
        field  can be used to note the fact and to indicate the nature
        of the encryption.  The first <word> parameter  indicates  the
        software  used  to  encrypt the body, and the second, optional
        <word> is intended to  aid  the  recipient  in  selecting  the
        proper  decryption  key.   This  code word may be viewed as an
        index to a table of keys held by the recipient.

        Note:  Unfortunately, headers must contain envelope,  as  well
               as  contents,  information.  Consequently, it is neces-
               sary that they remain unencrypted, so that  mail  tran-
               sport   services   may   access   them.   Since  names,
               addresses, and "Subject"  field  contents  may  contain

     August 13, 1982              - 24 -                      RFC #822

     Standard for ARPA Internet Text Messages

               sensitive  information,  this  requirement limits total
               message privacy.

             Names of encryption software are registered with the Net-
        work  Information Center, SRI International, Menlo Park, Cali-

     4.7.4.  EXTENSION-FIELD

             A limited number of common fields have  been  defined  in
        this  document.   As  network mail requirements dictate, addi-
        tional fields may be standardized.   To  provide  user-defined
        fields  with  a  measure  of  safety,  in name selection, such
        extension-fields will never have names  that  begin  with  the
        string "X-".

             Names of Extension-fields are registered with the Network
        Information Center, SRI International, Menlo Park, California.


             Individual users of network mail are free to  define  and
        use  additional  header  fields.   Such fields must have names
        which are not already used in the current specification or  in
        any definitions of extension-fields, and the overall syntax of
        these user-defined-fields must conform to this specification's
        rules   for   delimiting  and  folding  fields.   Due  to  the
        extension-field  publishing  process,  the  name  of  a  user-
        defined-field may be pre-empted

        Note:  The prefatory string "X-" will never  be  used  in  the
               names  of Extension-fields.  This provides user-defined
               fields with a protected set of names.

     August 13, 1982              - 25 -                      RFC #822

     Standard for ARPA Internet Text Messages


     5.1.  SYNTAX

     date-time   =  [ day "," ] date time        ; dd mm yy
                                                 ;  hh:mm:ss zzz

     day         =  "Mon"  / "Tue" /  "Wed"  / "Thu"
                 /  "Fri"  / "Sat" /  "Sun"

     date        =  1*2DIGIT month 2DIGIT        ; day month year
                                                 ;  e.g. 20 Jun 82

     month       =  "Jan"  /  "Feb" /  "Mar"  /  "Apr"
                 /  "May"  /  "Jun" /  "Jul"  /  "Aug"
                 /  "Sep"  /  "Oct" /  "Nov"  /  "Dec"

     time        =  hour zone                    ; ANSI and Military

     hour        =  2DIGIT ":" 2DIGIT [":" 2DIGIT]
                                                 ; 00:00:00 - 23:59:59

     zone        =  "UT"  / "GMT"                ; Universal Time
                                                 ; North American : UT
                 /  "EST" / "EDT"                ;  Eastern:  - 5/ - 4
                 /  "CST" / "CDT"                ;  Central:  - 6/ - 5
                 /  "MST" / "MDT"                ;  Mountain: - 7/ - 6
                 /  "PST" / "PDT"                ;  Pacific:  - 8/ - 7
                 /  1ALPHA                       ; Military: Z = UT;
                                                 ;  A:-1; (J not used)
                                                 ;  M:-12; N:+1; Y:+12
                 / ( ("+" / "-") 4DIGIT )        ; Local differential
                                                 ;  hours+min. (HHMM)

     5.2.  SEMANTICS

          If included, day-of-week must be the day implied by the date

          Time zone may be indicated in several ways.  "UT" is Univer-
     sal  Time  (formerly called "Greenwich Mean Time"); "GMT" is per-
     mitted as a reference to Universal Time.  The  military  standard
     uses  a  single  character for each zone.  "Z" is Universal Time.
     "A" indicates one hour earlier, and "M" indicates 12  hours  ear-
     lier;  "N"  is  one  hour  later, and "Y" is 12 hours later.  The
     letter "J" is not used.  The other remaining two forms are  taken
     from ANSI standard X3.51-1975.  One allows explicit indication of
     the amount of offset from UT; the other uses  common  3-character
     strings for indicating time zones in North America.

     August 13, 1982              - 26 -                      RFC #822

     Standard for ARPA Internet Text Messages


     6.1.  SYNTAX

     address     =  mailbox                      ; one addressee
                 /  group                        ; named list

     group       =  phrase ":" [#mailbox] ";"

     mailbox     =  addr-spec                    ; simple address
                 /  phrase route-addr            ; name & addr-spec

     route-addr  =  "<" [route] addr-spec ">"

     route       =  1#("@" domain) ":"           ; path-relative

     addr-spec   =  local-part "@" domain        ; global address

     local-part  =  word *("." word)             ; uninterpreted
                                                 ; case-preserved

     domain      =  sub-domain *("." sub-domain)

     sub-domain  =  domain-ref / domain-literal

     domain-ref  =  atom                         ; symbolic reference

     6.2.  SEMANTICS

          A mailbox receives mail.  It is a  conceptual  entity  which
     does  not necessarily pertain to file storage.  For example, some
     sites may choose to print mail on their line printer and  deliver
     the output to the addressee's desk.

          A mailbox specification comprises a person, system  or  pro-
     cess name reference, a domain-dependent string, and a name-domain
     reference.  The name reference is optional and is usually used to
     indicate  the  human name of a recipient.  The name-domain refer-
     ence specifies a sequence of sub-domains.   The  domain-dependent
     string is uninterpreted, except by the final sub-domain; the rest
     of the mail service merely transmits it as a literal string.

     6.2.1.  DOMAINS

        A name-domain is a set of registered (mail)  names.   A  name-
        domain  specification  resolves  to  a subordinate name-domain
        specification  or  to  a  terminal  domain-dependent   string.
        Hence,  domain  specification  is  extensible,  permitting any
        number of registration levels.

     August 13, 1982              - 27 -                      RFC #822

     Standard for ARPA Internet Text Messages

        Name-domains model a global, logical, hierarchical  addressing
        scheme.   The  model is logical, in that an address specifica-
        tion is related to name registration and  is  not  necessarily
        tied  to  transmission  path.   The  model's  hierarchy  is  a
        directed graph, called an in-tree, such that there is a single
        path  from  the root of the tree to any node in the hierarchy.
        If more than one path actually exists, they are considered  to
        be different addresses.

        The root node is common to all addresses; consequently, it  is
        not  referenced.   Its  children  constitute "top-level" name-
        domains.  Usually, a service has access to its own full domain
        specification and to the names of all top-level name-domains.

        The "top" of the domain addressing hierarchy -- a child of the
        root  --  is  indicated  by  the right-most field, in a domain
        specification.  Its child is specified to the left, its  child
        to the left, and so on.

        Some groups provide formal registration services;  these  con-
        stitute   name-domains   that  are  independent  logically  of
        specific machines.  In addition, networks and machines  impli-
        citly  compose name-domains, since their membership usually is
        registered in name tables.

        In the case of formal registration, an organization implements
        a  (distributed)  data base which provides an address-to-route
        mapping service for addresses of the form:


        Note that "organization" is a logical  entity,  separate  from
        any particular communication network.

        A mechanism for accessing "organization" is universally avail-
        able.   That mechanism, in turn, seeks an instantiation of the
        registry; its location is not indicated in the address specif-
        ication.   It  is assumed that the system which operates under
        the name "organization" knows how to find a subordinate regis-
        try.  The registry will then use the "person" string to deter-
        mine where to send the mail specification.

        The latter,  network-oriented  case  permits  simple,  direct,
        attachment-related address specification, such as:


        Once the network is accessed, it is expected  that  a  message
        will  go  directly  to the host and that the host will resolve

     August 13, 1982              - 28 -                      RFC #822

     Standard for ARPA Internet Text Messages

        the user name, placing the message in the user's mailbox.


        Since any number of  levels  is  possible  within  the  domain
        hierarchy,  specification  of  a  fully  qualified address can
        become inconvenient.  This standard permits abbreviated domain
        specification, in a special case:

            For the address of  the  sender,  call  the  left-most
            sub-domain  Level  N.   In a header address, if all of
            the sub-domains above (i.e., to the right of) Level  N
            are  the same as those of the sender, then they do not
            have to appear in the specification.   Otherwise,  the
            address must be fully qualified.

            This feature is subject  to  approval  by  local  sub-
            domains.   Individual  sub-domains  may  require their
            member systems, which originate mail, to provide  full
            domain  specification only.  When permitted, abbrevia-
            tions may be present  only  while  the  message  stays
            within the sub-domain of the sender.

            Use of this mechanism requires the sender's sub-domain
            to reserve the names of all top-level domains, so that
            full specifications can be distinguished from abbrevi-
            ated specifications.

        For example, if a sender's address is:


        and one recipient's address is:


        and another's is:


        then ".registry-1.organization-X" need not be specified in the
        the  message,  but  "registry-C.registry-2"  DOES  have  to be
        specified.  That is, the first two addresses may  be  abbrevi-
        ated, but the third address must be fully specified.

        When a message crosses a domain boundary, all  addresses  must
        be  specified  in  the  full format, ending with the top-level
        name-domain in the right-most field.  It is the responsibility
        of  mail  forwarding services to ensure that addresses conform

     August 13, 1982              - 29 -                      RFC #822

     Standard for ARPA Internet Text Messages

        with this requirement.  In the case of abbreviated  addresses,
        the  relaying  service must make the necessary expansions.  It
        should be noted that it often is difficult for such a  service
        to locate all occurrences of address abbreviations.  For exam-
        ple, it will not be possible to find such abbreviations within
        the  body  of  the  message.   The "Return-Path" field can aid
        recipients in recovering from these errors.

        Note:  When passing any portion of an addr-spec onto a process
               which  does  not interpret data according to this stan-
               dard (e.g., mail protocol servers).  There must  be  NO
               LWSP-chars  preceding  or  following the at-sign or any
               delimiting period ("."), such as  shown  in  the  above
               examples,   and   only  ONE  SPACE  between  contiguous

     6.2.3.  DOMAIN TERMS

        A domain-ref must be THE official name of a registry, network,
        or  host.   It  is  a  symbolic  reference, within a name sub-
        domain.  At times, it is necessary to bypass standard  mechan-
        isms  for  resolving  such  references,  using  more primitive
        information, such as a network host address  rather  than  its
        associated host name.

        To permit such references, this standard provides the  domain-
        literal  construct.   Its contents must conform with the needs
        of the sub-domain in which it is interpreted.

        Domain-literals which refer to domains within the ARPA  Inter-
        net  specify  32-bit  Internet addresses, in four 8-bit fields
        noted in decimal, as described in Request for  Comments  #820,
        "Assigned Numbers."  For example:


               is  permitted  only  as  a means of bypassing temporary
               system limitations, such as name tables which  are  not

        The names of "top-level" domains, and  the  names  of  domains
        under  in  the  ARPA Internet, are registered with the Network
        Information Center, SRI International, Menlo Park, California.


        The local-part of an  addr-spec  in  a  mailbox  specification
        (i.e.,  the  host's  name for the mailbox) is understood to be

     August 13, 1982              - 30 -                      RFC #822

     Standard for ARPA Internet Text Messages

        whatever the receiving mail protocol server allows.  For exam-
        ple,  some systems do not understand mailbox references of the
        form "P. D. Q. Bach", but others do.

        This specification treats periods (".") as lexical separators.
        Hence,  their  presence  in  local-parts which are not quoted-
        strings, is detected.   However,  such  occurrences  carry  NO
        semantics.  That is, if a local-part has periods within it, an
        address parser will divide the local-part into several tokens,
        but  the  sequence  of  tokens will be treated as one uninter-
        preted unit.  The sequence  will  be  re-assembled,  when  the
        address is passed outside of the system such as to a mail pro-
        tocol service.

        For example, the address:


        is legal and does not require the local-part to be  surrounded
        with  quotation-marks.   (However,  "First  Last" DOES require
        quoting.)  The local-part of the address, when passed  outside
        of  the  mail  system,  within  the  Registry.Org  domain,  is
        "First.Last", again without quotation marks.


        In some cases, the boundary between local-part and domain  can
        be  flexible.  The local-part may be a simple string, which is
        used for the final determination of the  recipient's  mailbox.
        All  other  levels  of  reference  are, therefore, part of the

        For some systems, in the case of abbreviated reference to  the
        local  and  subordinate  sub-domains,  it  may  be possible to
        specify only one reference within the domain  part  and  place
        the  other,  subordinate  name-domain  references  within  the
        local-part.  This would appear as:


        Such a specification would be acceptable  to  address  parsers
        which  conform  to  RFC  #733,  but  do not support this newer
        Internet standard.  While contrary to the intent of this stan-
        dard, the form is legal.

        Also, some sub-domains have a specification syntax which  does
        not conform to this standard.  For example:


     August 13, 1982              - 31 -                      RFC #822

     Standard for ARPA Internet Text Messages

        uses a different parsing  sequence  for  local-part  than  for

        Note:  As a rule,  the  domain  specification  should  contain
               fields  which  are  encoded  according to the syntax of
               this standard and which contain  generally-standardized
               information.   The local-part specification should con-
               tain only that portion of the  address  which  deviates
               from the form or intention of the domain field.


        An individual may have several mailboxes and wish  to  receive
        mail  at  whatever  mailbox  is  convenient  for the sender to
        access.  This standard does not provide a means of  specifying
        "any member of" a list of mailboxes.

        A set of individuals may wish to receive mail as a single unit
        (i.e.,  a  distribution  list).  The <group> construct permits
        specification of such a list.  Recipient mailboxes are  speci-
        fied  within  the  bracketed  part (":" - ";").  A copy of the
        transmitted message is to be  sent  to  each  mailbox  listed.
        This  standard  does  not  permit  recursive  specification of
        groups within groups.

        While a list must be named, it is not required that  the  con-
        tents  of  the  list be included.  In this case, the <address>
        serves only as an indication of group distribution  and  would
        appear in the form:


        Some mail  services  may  provide  a  group-list  distribution
        facility,  accepting  a single mailbox reference, expanding it
        to the full distribution list, and relaying the  mail  to  the
        list's  members.   This standard provides no additional syntax
        for indicating such a  service.   Using  the  <group>  address
        alternative,  while listing one mailbox in it, can mean either
        that the mailbox reference will be expanded to a list or  that
        there is a group with one member.


        At times, a  message  originator  may  wish  to  indicate  the
        transmission  path  that  a  message  should  follow.  This is
        called source routing.  The normal addressing scheme, used  in
        an  addr-spec,  is  carefully separated from such information;
        the <route> portion of a route-addr is provided for such occa-
        sions.  It specifies the sequence of hosts and/or transmission

     August 13, 1982              - 32 -                      RFC #822

     Standard for ARPA Internet Text Messages

        services that are  to  be  traversed.   Both  domain-refs  and
        domain-literals may be used.

        Note:  The use of source routing is discouraged.   Unless  the
               sender has special need of path restriction, the choice
               of transmission route should be left to the mail  tran-
               sport service.


          It often is necessary to send mail to a site, without  know-
     ing  any  of its valid addresses.  For example, there may be mail
     system dysfunctions, or a user may wish to find  out  a  person's
     correct address, at that site.

          This standard specifies a single, reserved  mailbox  address
     (local-part)  which  is  to  be valid at each site.  Mail sent to
     that address is to be routed to  a  person  responsible  for  the
     site's mail system or to a person with responsibility for general
     site operation.  The name of the reserved local-part address is:


     so that "Postmaster@domain" is required to be valid.

     Note:  This reserved local-part must be  matched  without  sensi-
            tivity to alphabetic case, so that "POSTMASTER", "postmas-
            ter", and even "poStmASteR" is to be accepted.

     August 13, 1982              - 33 -                      RFC #822

     Standard for ARPA Internet Text Messages


     ANSI.  "USA Standard Code  for  Information  Interchange,"  X3.4.
        American  National Standards Institute: New York (1968).  Also
        in:  Feinler, E.  and J. Postel, eds., "ARPANET Protocol Hand-
        book", NIC 7104.

     ANSI.  "Representations of Universal Time, Local  Time  Differen-
        tials,  and United States Time Zone References for Information
        Interchange," X3.51-1975.  American National Standards  Insti-
        tute:  New York (1975).

     Bemer, R.W., "Time and the Computer."  In:  Interface  Age  (Feb.

     Bennett, C.J.  "JNT Mail Protocol".  Joint Network Team,  Ruther-
        ford and Appleton Laboratory:  Didcot, England.

     Bhushan, A.K., Pogran, K.T., Tomlinson,  R.S.,  and  White,  J.E.
        "Standardizing  Network  Mail  Headers,"   ARPANET Request for
        Comments No. 561, Network Information Center  No.  18516;  SRI
        International:  Menlo Park (September 1973).

     Birrell, A.D., Levin, R.,  Needham,  R.M.,  and  Schroeder,  M.D.
        "Grapevine:  An Exercise in Distributed Computing," Communica-
        tions of the ACM 25, 4 (April 1982), 260-274.

     Crocker,  D.H.,  Vittal,  J.J.,  Pogran,  K.T.,  Henderson,  D.A.
        "Standard  for  the  Format  of  ARPA  Network  Text Message,"
        ARPANET Request for  Comments  No.  733,  Network  Information
        Center  No.  41952.   SRI International:  Menlo Park (November

     Feinler, E.J. and Postel, J.B.  ARPANET Protocol  Handbook,  Net-
        work  Information  Center  No.  7104   (NTIS AD A003890).  SRI
        International:  Menlo Park (April 1976).

     Harary, F.   "Graph  Theory".   Addison-Wesley:   Reading,  Mass.

     Levin, R. and Schroeder, M.  "Transport  of  Electronic  Messages
        through  a  Network,"   TeleInformatics  79, pp. 29-33.  North
        Holland (1979).  Also  as  Xerox  Palo  Alto  Research  Center
        Technical Report CSL-79-4.

     Myer, T.H. and Henderson, D.A.  "Message Transmission  Protocol,"
        ARPANET  Request  for  Comments,  No. 680, Network Information
        Center No. 32116.  SRI International:  Menlo Park (1975).

     August 13, 1982              - 34 -                      RFC #822

     Standard for ARPA Internet Text Messages

     NBS.  "Specification of Message Format for Computer Based Message
        Systems, Recommended Federal Information Processing Standard."
        National  Bureau   of   Standards:    Gaithersburg,   Maryland
        (October 1981).

     NIC.  Internet Protocol Transition Workbook.  Network Information
        Center,   SRI-International,  Menlo  Park,  California  (March

     Oppen, D.C. and Dalal, Y.K.  "The Clearinghouse:  A Decentralized
        Agent  for  Locating  Named  Objects in a Distributed Environ-
        ment," OPD-T8103.  Xerox Office Products Division:  Palo Alto,
        CA. (October 1981).

     Postel, J.B.  "Assigned Numbers,"  ARPANET Request for  Comments,
        No. 820.  SRI International:  Menlo Park (August 1982).

     Postel, J.B.  "Simple Mail Transfer  Protocol,"  ARPANET  Request
        for Comments, No. 821.  SRI International:  Menlo Park (August

     Shoch, J.F.  "Internetwork naming, addressing  and  routing,"  in
        Proc. 17th IEEE Computer Society International Conference, pp.
        72-79, Sept. 1978, IEEE Cat. No. 78 CH 1388-8C.

     Su, Z. and Postel, J.  "The Domain Naming Convention for Internet
        User  Applications,"  ARPANET  Request  for Comments, No. 819.
        SRI International:  Menlo Park (August 1982).

     August 13, 1982              - 35 -                      RFC #822

     Standard for ARPA Internet Text Messages



     A.1.  ADDRESSES

     A.1.1.  Alfred Neuman <Neuman@BBN-TENEXA>

     A.1.2.  Neuman@BBN-TENEXA

             These two "Alfred Neuman" examples have identical  seman-
        tics, as far as the operation of the local host's mail sending
        (distribution) program (also sometimes  called  its  "mailer")
        and  the remote host's mail protocol server are concerned.  In
        the first example, the  "Alfred  Neuman"  is  ignored  by  the
        mailer,  as "Neuman@BBN-TENEXA" completely specifies the reci-
        pient.  The second example contains  no  superfluous  informa-
        tion,  and,  again,  "Neuman@BBN-TENEXA" is the intended reci-

        Note:  When the message crosses name-domain  boundaries,  then
               these specifications must be changed, so as to indicate
               the remainder of the hierarchy, starting with  the  top

     A.1.3.  "George, Ted" <Shared@Group.Arpanet>

             This form might be used to indicate that a single mailbox
        is  shared  by several users.  The quoted string is ignored by
        the originating host's mailer, because  "Shared@Group.Arpanet"
        completely specifies the destination mailbox.

     A.1.4.  Wilt . (the  Stilt) Chamberlain@NBA.US

             The "(the  Stilt)" is a comment, which is NOT included in
        the  destination  mailbox  address  handed  to the originating
        system's mailer.  The local-part of the address is the  string
        "Wilt.Chamberlain", with NO space between the first and second

     A.1.5.  Address Lists

     Gourmets:  Pompous Person <WhoZiWhatZit@Cordon-Bleu>,
                Childs@WGBH.Boston, Galloping Gourmet@
                ANT.Down-Under (Australian National Television),
       Cruisers:  Port@Portugal, Jones@SEA;,

     August 13, 1982              - 36 -                      RFC #822

     Standard for ARPA Internet Text Messages

        This group list example points out the use of comments and the
        mixing of addresses and groups.


     A.2.1.  Author-sent

             George Jones logs into his host  as  "Jones".   He  sends
        mail himself.

            From:  Jones@Group.Org


            From:  George Jones <Jones@Group.Org>

     A.2.2.  Secretary-sent

             George Jones logs in as Jones on his  host.   His  secre-
        tary,  who logs in as Secy sends mail for him.  Replies to the
        mail should go to George.

            From:    George Jones <Jones@Group>
            Sender:  Secy@Other-Group

     A.2.3.  Secretary-sent, for user of shared directory

             George Jones' secretary sends mail  for  George.  Replies
        should go to George.

            From:     George Jones<Shared@Group.Org>
            Sender:   Secy@Other-Group

        Note that there need not be a space between  "Jones"  and  the
        "<",  but  adding a space enhances readability (as is the case
        in other examples.

     A.2.4.  Committee activity, with one author

             George is a member of a committee.  He wishes to have any
        replies to his message go to all committee members.

            From:     George Jones <Jones@Host.Net>
            Sender:   Jones@Host
            Reply-To: The Committee: Jones@Host.Net,

        Note  that  if  George  had  not  included  himself   in   the

     August 13, 1982              - 37 -                      RFC #822

     Standard for ARPA Internet Text Messages

        enumeration  of  The  Committee,  he  would not have gotten an
        implicit reply; the presence of the  "Reply-to"  field  SUPER-
        SEDES the sending of a reply to the person named in the "From"

     A.2.5.  Secretary acting as full agent of author

             George Jones asks his secretary  (Secy@Host)  to  send  a
        message for him in his capacity as Group.  He wants his secre-
        tary to handle all replies.

            From:     George Jones <Group@Host>
            Sender:   Secy@Host
            Reply-To: Secy@Host

     A.2.6.  Agent for user without online mailbox

             A friend  of  George's,  Sarah,  is  visiting.   George's
        secretary  sends  some  mail to a friend of Sarah in computer-
        land.  Replies should go to George, whose mailbox is Jones  at

            From:     Sarah Friendly <Secy@Registry>
            Sender:   Secy-Name <Secy@Registry>
            Reply-To: Jones@Registry.

     A.2.7.  Agent for member of a committee

             George's secretary sends out a message which was authored
        jointly by all the members of a committee.  Note that the name
        of the committee cannot be specified, since <group> names  are
        not permitted in the From field.

            From:   Jones@Host,
            Sender: Secy@SHost

     August 13, 1982              - 38 -                      RFC #822

     Standard for ARPA Internet Text Messages


     A.3.1.  Minimum required

     Date:     26 Aug 76 1429 EDT        Date:     26 Aug 76 1429 EDT
     From:     Jones@Registry.Org   or   From:     Jones@Registry.Org
     Bcc:                                To:       Smith@Registry.Org

        Note that the "Bcc" field may be empty, while the  "To"  field
        is required to have at least one address.

     A.3.2.  Using some of the additional fields

     Date:     26 Aug 76 1430 EDT
     From:     George Jones<Group@Host>
     Sender:   Secy@SHOST
     To:       "Al Neuman"@Mad-Host,
     Message-ID:  <some.string@SHOST>

     A.3.3.  About as complex as you're going to get

     Date     :  27 Aug 76 0932 PDT
     From     :  Ken Davis <KDavis@This-Host.This-net>
     Subject  :  Re: The Syntax in the RFC
     Sender   :  KSecy@Other-Host
     Reply-To :  Sam.Irving@Reg.Organization
     To       :  George Jones <Group@Some-Reg.An-Org>,
     cc       :  Important folk:
                   Tom Softwood <Balsa@Tree.Root>,
                   "Sam Irving"@Other-Host;,
                 Standard Distribution:
     Comment  :  Sam is away on business. He asked me to handle
                 his mail for him.  He'll be able to provide  a
                 more  accurate  explanation  when  he  returns
                 next week.
     In-Reply-To: <some.string@DBM.Group>, George's message
     X-Special-action:  This is a sample of user-defined field-
                 names.  There could also be a field-name
                 "Special-action", but its name might later be
     Message-ID: <4231.629.XYzi-What@Other-Host>

     August 13, 1982              - 39 -                      RFC #822

     Standard for ARPA Internet Text Messages


          Some mail-reading software systems may wish to perform  only
     minimal  processing,  ignoring  the internal syntax of structured
     field-bodies and treating them the  same  as  unstructured-field-
     bodies.  Such software will need only to distinguish:

         o   Header fields from the message body,

         o   Beginnings of fields from lines which continue fields,

         o   Field-names from field-contents.

          The abbreviated set of syntactic rules  which  follows  will
     suffice  for  this  purpose.  It describes a limited view of mes-
     sages and is a subset of the syntactic rules provided in the main
     part of this specification.  One small exception is that the con-
     tents of field-bodies consist only of text:

     B.1.  SYNTAX

     message         =   *field *(CRLF *text)

     field           =    field-name ":" [field-body] CRLF

     field-name      =  1*<any CHAR, excluding CTLs, SPACE, and ":">

     field-body      =   *text [CRLF LWSP-char field-body]

     B.2.  SEMANTICS

          Headers occur before the message body and are terminated  by
     a null line (i.e., two contiguous CRLFs).

          A line which continues a header field begins with a SPACE or
     HTAB  character,  while  a  line  beginning a field starts with a
     printable character which is not a colon.

          A field-name consists of one or  more  printable  characters
     (excluding  colon,  space, and control-characters).  A field-name
     MUST be contained on one line.  Upper and lower case are not dis-
     tinguished when comparing field-names.

     August 13, 1982              - 40 -                      RFC #822

     Standard for ARPA Internet Text Messages


          The following summarizes the differences between this  stan-
     dard  and the one specified in Arpanet Request for Comments #733,
     "Standard for the Format of ARPA  Network  Text  Messages".   The
     differences  are  listed  in the order of their occurrence in the
     current specification.


     C.1.1.  FIELD NAMES

        These now must be a sequence of  printable  characters.   They
        may not contain any LWSP-chars.


     C.2.1.  SPECIALS

        The characters period ("."), left-square  bracket  ("["),  and
        right-square  bracket ("]") have been added.  For presentation
        purposes, and when passing a specification to  a  system  that
        does  not conform to this standard, periods are to be contigu-
        ous with their surrounding lexical tokens.   No  linear-white-
        space  is  permitted  between them.  The presence of one LWSP-
        char between other tokens is still directed.

     C.2.2.  ATOM

        Atoms may not contain SPACE.

     C.2.3.  SPECIAL TEXT

        ctext and qtext have had backslash ("\") added to the list  of
        prohibited characters.

     C.2.4.  DOMAINS

        The lexical tokens  <domain-literal>  and  <dtext>  have  been


     C.3.1.  TRACE

        The "Return-path:" and "Received:" fields have been specified.

     August 13, 1982              - 41 -                      RFC #822

     Standard for ARPA Internet Text Messages

     C.3.2.  FROM

        The "From" field must contain machine-usable addresses  (addr-
        spec).   Multiple  addresses may be specified, but named-lists
        (groups) may not.

     C.3.3.  RESENT

        The meta-construct of prefacing field names  with  the  string
        "Resent-"  has been added, to indicate that a message has been
        forwarded by an intermediate recipient.

     C.3.4.  DESTINATION

        A message must contain at least one destination address field.
        "To" and "CC" are required to contain at least one address.

     C.3.5.  IN-REPLY-TO

        The field-body is no longer a comma-separated list, although a
        sequence is still permitted.

     C.3.6.  REFERENCE

        The field-body is no longer a comma-separated list, although a
        sequence is still permitted.

     C.3.7.  ENCRYPTED

        A field has been specified that permits  senders  to  indicate
        that the body of a message has been encrypted.


        Extension fields are prohibited from beginning with the  char-
        acters "X-".



        Fewer optional forms are permitted  and  the  list  of  three-
        letter time zones has been shortened.


     August 13, 1982              - 42 -                      RFC #822

     Standard for ARPA Internet Text Messages

     C.5.1.  ADDRESS

        The use of quoted-string, and the ":"-atom-":" construct, have
        been  removed.   An  address  now  is  either a single mailbox
        reference or is a named list of addresses.  The  latter  indi-
        cates a group distribution.

     C.5.2.  GROUPS

        Group lists are now required to to have a name.   Group  lists
        may not be nested.

     C.5.3.  MAILBOX

        A mailbox specification  may  indicate  a  person's  name,  as
        before.   Such  a  named  list  no longer may specify multiple
        mailboxes and may not be nested.


        Addresses now are taken to be absolute, global specifications,
        independent  of transmission paths.  The <route> construct has
        been provided, to permit explicit specification  of  transmis-
        sion  path.   RFC  #733's  use  of multiple at-signs ("@") was
        intended as a general syntax  for  indicating  routing  and/or
        hierarchical addressing.  The current standard separates these
        specifications and only one at-sign is permitted.

     C.5.5.  AT-SIGN

        The string " at " no longer is used as an  address  delimiter.
        Only at-sign ("@") serves the function.

     C.5.6.  DOMAINS

        Hierarchical, logical name-domains have been added.


     The local-part "Postmaster" has been reserved, so that users  can
     be guaranteed at least one valid address at a site.

     August 13, 1982              - 43 -                      RFC #822

     Standard for ARPA Internet Text Messages


     address     =  mailbox                      ; one addressee
                 /  group                        ; named list
     addr-spec   =  local-part "@" domain        ; global address
     ALPHA       =  <any ASCII alphabetic character>
                                                 ; (101-132, 65.- 90.)
                                                 ; (141-172, 97.-122.)
     atom        =  1*<any CHAR except specials, SPACE and CTLs>
     authentic   =   "From"       ":"   mailbox  ; Single author
                 / ( "Sender"     ":"   mailbox  ; Actual submittor
                     "From"       ":" 1#mailbox) ; Multiple authors
                                                 ;  or not sender
     CHAR        =  <any ASCII character>        ; (  0-177,  0.-127.)
     comment     =  "(" *(ctext / quoted-pair / comment) ")"
     CR          =  <ASCII CR, carriage return>  ; (     15,      13.)
     CRLF        =  CR LF
     ctext       =  <any CHAR excluding "(",     ; => may be folded
                     ")", "\" & CR, & including
     CTL         =  <any ASCII control           ; (  0- 37,  0.- 31.)
                     character and DEL>          ; (    177,     127.)
     date        =  1*2DIGIT month 2DIGIT        ; day month year
                                                 ;  e.g. 20 Jun 82
     dates       =   orig-date                   ; Original
                   [ resent-date ]               ; Forwarded
     date-time   =  [ day "," ] date time        ; dd mm yy
                                                 ;  hh:mm:ss zzz
     day         =  "Mon"  / "Tue" /  "Wed"  / "Thu"
                 /  "Fri"  / "Sat" /  "Sun"
     delimiters  =  specials / linear-white-space / comment
     destination =  "To"          ":" 1#address  ; Primary
                 /  "Resent-To"   ":" 1#address
                 /  "cc"          ":" 1#address  ; Secondary
                 /  "Resent-cc"   ":" 1#address
                 /  "bcc"         ":"  #address  ; Blind carbon
                 /  "Resent-bcc"  ":"  #address
     DIGIT       =  <any ASCII decimal digit>    ; ( 60- 71, 48.- 57.)
     domain      =  sub-domain *("." sub-domain)
     domain-literal =  "[" *(dtext / quoted-pair) "]"
     domain-ref  =  atom                         ; symbolic reference
     dtext       =  <any CHAR excluding "[",     ; => may be folded
                     "]", "\" & CR, & including
     extension-field =
                   <Any field which is defined in a document
                    published as a formal extension to this
                    specification; none will have names beginning
                    with the string "X-">

     August 13, 1982              - 44 -                      RFC #822

     Standard for ARPA Internet Text Messages

     field       =  field-name ":" [ field-body ] CRLF
     fields      =    dates                      ; Creation time,
                      source                     ;  author id & one
                    1*destination                ;  address required
                     *optional-field             ;  others optional
     field-body  =  field-body-contents
                    [CRLF LWSP-char field-body]
     field-body-contents =
                   <the ASCII characters making up the field-body, as
                    defined in the following sections, and consisting
                    of combinations of atom, quoted-string, and
                    specials tokens, or else consisting of texts>
     field-name  =  1*<any CHAR, excluding CTLs, SPACE, and ":">
     group       =  phrase ":" [#mailbox] ";"
     hour        =  2DIGIT ":" 2DIGIT [":" 2DIGIT]
                                                 ; 00:00:00 - 23:59:59
     HTAB        =  <ASCII HT, horizontal-tab>   ; (     11,       9.)
     LF          =  <ASCII LF, linefeed>         ; (     12,      10.)
     linear-white-space =  1*([CRLF] LWSP-char)  ; semantics = SPACE
                                                 ; CRLF => folding
     local-part  =  word *("." word)             ; uninterpreted
                                                 ; case-preserved
     LWSP-char   =  SPACE / HTAB                 ; semantics = SPACE
     mailbox     =  addr-spec                    ; simple address
                 /  phrase route-addr            ; name & addr-spec
     message     =  fields *( CRLF *text )       ; Everything after
                                                 ;  first null line
                                                 ;  is message body
     month       =  "Jan"  /  "Feb" /  "Mar"  /  "Apr"
                 /  "May"  /  "Jun" /  "Jul"  /  "Aug"
                 /  "Sep"  /  "Oct" /  "Nov"  /  "Dec"
     msg-id      =  "<" addr-spec ">"            ; Unique message id
     optional-field =
                 /  "Message-ID"        ":"   msg-id
                 /  "Resent-Message-ID" ":"   msg-id
                 /  "In-Reply-To"       ":"  *(phrase / msg-id)
                 /  "References"        ":"  *(phrase / msg-id)
                 /  "Keywords"          ":"  #phrase
                 /  "Subject"           ":"  *text
                 /  "Comments"          ":"  *text
                 /  "Encrypted"         ":" 1#2word
                 /  extension-field              ; To be defined
                 /  user-defined-field           ; May be pre-empted
     orig-date   =  "Date"        ":"   date-time
     originator  =   authentic                   ; authenticated addr
                   [ "Reply-To"   ":" 1#address] )
     phrase      =  1*word                       ; Sequence of words

     August 13, 1982              - 45 -                      RFC #822

     Standard for ARPA Internet Text Messages

     qtext       =  <any CHAR excepting <">,     ; => may be folded
                     "\" & CR, and including
     quoted-pair =  "\" CHAR                     ; may quote any char
     quoted-string = <"> *(qtext/quoted-pair) <">; Regular qtext or
                                                 ;   quoted chars.
     received    =  "Received"    ":"            ; one per relay
                       ["from" domain]           ; sending host
                       ["by"   domain]           ; receiving host
                       ["via"  atom]             ; physical path
                      *("with" atom)             ; link/mail protocol
                       ["id"   msg-id]           ; receiver msg id
                       ["for"  addr-spec]        ; initial form
                        ";"    date-time         ; time received

     resent      =   resent-authentic
                   [ "Resent-Reply-To"  ":" 1#address] )
     resent-authentic =
                 =   "Resent-From"      ":"   mailbox
                 / ( "Resent-Sender"    ":"   mailbox
                     "Resent-From"      ":" 1#mailbox  )
     resent-date =  "Resent-Date" ":"   date-time
     return      =  "Return-path" ":" route-addr ; return address
     route       =  1#("@" domain) ":"           ; path-relative
     route-addr  =  "<" [route] addr-spec ">"
     source      = [  trace ]                    ; net traversals
                      originator                 ; original mail
                   [  resent ]                   ; forwarded
     SPACE       =  <ASCII SP, space>            ; (     40,      32.)
     specials    =  "(" / ")" / "<" / ">" / "@"  ; Must be in quoted-
                 /  "," / ";" / ":" / "\" / <">  ;  string, to use
                 /  "." / "[" / "]"              ;  within a word.
     sub-domain  =  domain-ref / domain-literal
     text        =  <any CHAR, including bare    ; => atoms, specials,
                     CR & bare LF, but NOT       ;  comments and
                     including CRLF>             ;  quoted-strings are
                                                 ;  NOT recognized.
     time        =  hour zone                    ; ANSI and Military
     trace       =    return                     ; path to sender
                    1*received                   ; receipt tags
     user-defined-field =
                   <Any field which has not been defined
                    in this specification or published as an
                    extension to this specification; names for
                    such fields must be unique and may be
                    pre-empted by published extensions>
     word        =  atom / quoted-string

     August 13, 1982              - 46 -                      RFC #822

     Standard for ARPA Internet Text Messages

     zone        =  "UT"  / "GMT"                ; Universal Time
                                                 ; North American : UT
                 /  "EST" / "EDT"                ;  Eastern:  - 5/ - 4
                 /  "CST" / "CDT"                ;  Central:  - 6/ - 5
                 /  "MST" / "MDT"                ;  Mountain: - 7/ - 6
                 /  "PST" / "PDT"                ;  Pacific:  - 8/ - 7
                 /  1ALPHA                       ; Military: Z = UT;
     <">         =  <ASCII quote mark>           ; (     42,      34.)

     August 13, 1982              - 47 -                      RFC #822