INTERMEDIATE FILES FOR MUSIC PRINTING

                       November 28, 1990
                       (revised 5-28-93)
                       (revised 11-11-93)
                       (revised 9-15-94)
                       (revised 9-27-94)
                       (revised 2-25-95)
                       (revised 8-28-02)
                       (revised 1-06-04)
                       (revised 4-26-05)
                       (revised 3-18-06)
                       (revised 12-26-10)


I. File Structure

   Variable length ASCII records
   Fields separated by blanks

II. Types of I-files

    I-files come in four types: linear, page-specific, scrolling
    page-specific, and multi-page.

  (1) Linear.

    These files represent the first step toward generating
    real musical notation.  MuseData source files are organized
    as musical parts, which can be combined in various ways
    to produce musical scores, short scores, parts, and
    arrangements of parts.  Musical typesetting in this system
    is logically a two step process: (1) translate source data
    into the i-file format, and (2) combine the individual
    i-files into real musical notation for display, printing,
    or conversion to another format, e.g., SCORE format or
    one of the POSTSCRIPT formats.

    Linear i-files are not suitable for actual display.  The
    clef, key and time signatures occur only at the beginning
    of the file, and there are no line breaks or page breaks.
    The x-offset parameter increases indefinitely.

  (2) Page-specific.

    Page-specific files are the final result of combining
    one or more individual linear files into a form suitable
    for display, etc.  Musical data is organized as a set of
    musical systems, each of which may have one or more lines.

     1. In order to locate the output on pages of finite horizontal
        size, the systems and their musical lines are broken at
        various points.

     2. Getting the notes in various parts to line up properly
        requires that the horizontal positions of musical objects
        in some parts be adjusted.

     3. In order to achieve right-side justification, horizonal
        distances between musical objects may be also be altered.

     4. Clef, key and time signatures are placed at the beginning
        of every line.

     5. Ties, slurs, extended hyphens and other multi-measure
        super-objects may be broken up and continued across line
        and page boundaries.

  (3) Scrolling page-specific

    These files are similar to page-specific files but are designed
    for scrolling in a display output.  Conditions (1), (3), (4)
    and (5) above do not hold, since there are no system breaks.
    Only condition (2) must be met.  This, and the fact that the
    music is organized into a system of one or more lines are what
    distinguish scrolling page-specific files from their linear
    i-file sources.

  (4) Multi-page (page_specific) files

    This is a special case of the page-specific format, where
    several page-specific files are concatenated to form one
    large file.  The delineator is a record having the following
    properties:

      1. Column 1 must contain the letter "P"

      2. One of the following three conditions must be met.

         (a) Record length = 1
         (b) "space" character in column 3
         (c) The first 4 columns must be the word "Page"

         These are to insure that we don't confuse the page
         boundary with a "P" super-object.

    The last record in a multi-page file must be either a
    "P" record or the four character string "/eof".  Multi-page
    files are designed primarily for use as (one form of) input
    to the muse2ps conversion program.


III. Record Types

      Linear Files                     Page Specific (additional)
      ------------                     --------------------------
       Music Lines                      Systems
       Objects                          Music Lines (page specific)
       Sub-objects                      End of Music Line
       Text (form of sub-object)        System Bar
       Words (form of sub-object)       Page Text
       Attribute (form for sub-object)  Page delineators
       Super-objects
       Q-records (linear only)
       Tag records (linear only)

IV. Record Formats -- Linear Files

  A. Music Lines

     Field 1: Identifier = L (or small l = single line staff, e.g., for triangle)
     Field 2: 0 = single staff; # = vertical offset of second staff
     Field 3: vertical displacement of text line
     Field 4: notesize (zero if not specified)
     Field 5: part name (ASCII string)


  B. Objects

     Field 1: Identifier = J

     Field 2: type of object    B = bar line
                                C = clef
                                K = key signature
                                T = time signature
                                D = directive (e.g. time word, etc)
                                S = other symbol
                                N = note
                                R = rest
                                G = grace note
                                Q = cue note
                                F = figures
                                I = isolated directive or symbol (not
                                      associated with a note, rest, or figure)
                                M = mark (mostly for superobjects)
     Field 3: object code

           Bar line:   x = measure number if there be one, else zero

           Clef:       x = clef code
           Key:        x = key code
           Time sig:   x = time code (100 * tnum + tden)
           Directive:  0 = print always (parts and score)
                       bit 0 set = print in parts
                       bit 1 set = print in score
                       bit 2 set = print if top part in score
                       bit 3 set = print if bottom part in score
                          (bit 1 over-rides bits 2 and 3)

           Symbol:     0 = not identified
                       4 = extended rest
                       6 = whole measure rest

                       From the standpoint of typesetting for a score
                       the major difference between a directive and a
                       symbol is that a symbol will always be printed
                       whether or not the part stands alone or is part
                       of a score (e.g., letter dynamics), whereas a
                       directive can be "made invisible" in parts that
                       are not at the top or the bottom of a score
                       (e.g., the segno sign).

           Notes:      x = note type
           Rests:      x = rest type
           Grace note: x = note type
           Cue note:   x = note type
           Figure:     0
           Isolated:   same as for Directives
           Mark:       0

     Field 4: horizontal displacement measured in 32-bit precision
                from beginning of piece

           For objects which are groups of notes (chords),
           the x co-ordinate is the position of the notes
           which fall on the proper side of the stem.  For
           stem up, the object's position is that of the
           note heads to the left of the stem; for stem down,
           the object's position is that of the note heads
           to the right of the stem.

     Field 5: vertical displacement from top staff line

           For objects which are groups of notes (chords), the
           y co-ordinate is the position of the note which is
           farthest from the note-end of the stem (and hence
           closest to a beam, if there be one).  This is impor-
           tant for the proper setting of beams at print time.

           For objects to be placed on the second (or greater)
           staff of the grand staff, 1000 (2000) will be added
           to the displacement.  This means that the print
           program must check the range of the displacement before
           placing the object.

           For Bar type objects, which always have a
           y co-ordinate of 0, this field contains instead
           the bar line code:
                                 1 = single light bar
                                 2 = single heave bar
                                 3 = dotted bar
                                 5 = double light bar
                                 6 = light-heavy double bar
                                 9 = heavy-light double bar
                                10 = heavy-heavy double bar

               In case the bar object extends to more than one staff
               an amount equal to 1000 times (n-1) where n is the
               number of staves is added to the bar line code

               Note: If the stage2 source file contains a print suggestion
                     modifying how mskpage deals with the music in the
                     measure following a bar line, this suggestion is
                     communicated to the mskpage program in this field.
                     The suggestion code is multiplied by 1,000,000 (one
                     million) and added to Field 5.

                     Print suggestions for dealing with music in a measure:

                       1 = do not modify the spacings in this measure when
                           justifying a line of music in a score (used only
                           by mskpage).

                       3 = same thing

                     If the stage2 source file contains a print suggestion
                     requesting that a particular bar line within a movement
                     should be right justified, this suggestion is
                     communicated to the mskpage program in this field by
                     adding 10,000,000 (ten million) to Field 5.

                     These print suggestions apply only to non-page-specific
                     files and are removed by mskpage (and mkpart) after they
                     are utilized.  They do not appear in the page-specific
                     output.

     Field 6: print code (or if < 32, number of sub-objects)

     Field 7: space node number

           1. Multiple rests -- number = measures of rest

           2. Everything else -- number = division number

              This number will indicate the division within the
            measure to which the object belongs.  The measure
            is divided into 6912 (27 x 256) parts.  Divisions
            run from 1 to 6912.  It is possible for several
            objects to have the same space node number.  These
            objects will have separate numbers in the page-
            specific format.

     Field 8: distance increment flag.  This flag describes the
              distance from the preceding object, as well as an
              attribute of the node:

                 0 = preceding distance should remain fixed
                      at print time

                 # = preceding distance may vary at print time;
                      parameter is the duration of the preceding
                      node, measured in units such that
                      576 = quarter note.

             10000 = object will be centered between bar lines
                      (used only with whole measure rests)

     Field 9: number of super-objects associated with this object

     Fields 9a,b...,: super-object numbers

  C. Sub-objects (these must follow directly after their associated
                    objects)

     Field 1: Identifier = K or C (or k for "silent" Sub-objects)

                "C" is used in place of "K" to indicate that the
                sub-object should be displayed in color rather
                than black.  This feature is used for highlighting.

     Field 2: horiz. displacement, relative to object co-ordinate
     Field 3: vertical displacement, relative to object co-ordinate
     Field 4: print code of sub-object

  D. Text (a form of sub-object attached to note or rest)

     Field 1: Identifier = T
     Field 2: horizontal displacement of text, relative to object

                Note: Non-page-specific i-files may have a second
                parameter attached to field 2 (separated by the "|"
                character).  The second parameter is the "ideal"
                displacement of text, relative to the object note
                under the assumption that we don't have to worry
                about clashes with adjacent notes.

     Field 3: vertical code (text line number, between 1 and 10)
                The vertical code may be followed by the "|" character
                and a y offset.  The interpretation of the offset is
                a displacement from the normal position of the line
                (see page-specific lines below).
     Field 4: ASCII string
                If this field = "~", this means that this note is the
                termination of a forward underline.

     Field 5: indication of forward hyphen or underline character

                - = forward hyphen
                _ = forward underline character
                . =     "         "       "     followed by  .
                , =     "         "       "        "     "   ,
                : =     "         "       "        "     "   :
                ; =     "         "       "        "     "   ;
                ! =     "         "       "        "     "   !
                ? =     "         "       "        "     "   ?
                * = nothing

     Field 6: space taken up by ASCII string

  E. Words (a form of sub-object attached to symbol)

     Field 1: Identifier = W
     Field 2: horizontal displacement of word(s), relative to
                 directive or symbol object
     Field 3: vertical displacement
     Field 4: font number
     Field 5: ASCII string (including font designations)

  F. Attributes (a form of sub-object attached to notes/rests)

     The purpose of attributes is to communication information about
     an object that is not represented precisely by purely graphical
     information.  An example is the duration of tuplets.  The tuplet
     super-object (H . X) references only the end points of a tuplet;
     there is no precise way to determine the remaining objects whose
     duration might be effected by the tuplet.  When converting page
     specific i-files to SCORE pmx files, we need to know the precise
     duration of all notes.  The only foolproof way to convey this
     information is in an attribute.  Attributes are ignored
     in the printing process.

     Field 1: Identifier = A
     Field 2: type of attribute

     1. D = duration

        Field 3: numerator of duration
        Field 4: denominator of duration

            Duration is represented as a proper fraction.
              1/4 = quarter note
              1/2 = half note
              etc.
            In terms of divisions (from stage2 source representation)
            the numerator is the number of divisions, and the denominator
            is four times the divisions per quarter.

        Field 5: tie (if present)

          (no field) = situation not determined (older files)
            0 = no tie
            1 = tie to following note
            2 = tie appended to selected pitches (multiple pitches only)

            Note: In the case of a rest, field 5 may be used (optionally)
                  to indicate that the next duration in this track is
                  also a rest (a tied rest, in effect)

        NOTE: Field 5 was introduced in Feb-2006 and might be absent
                from some older i-files.

     2. P = pitch

        Field 3: track number (normally 1)

        Field 4: base-40 pitch number (C4 = 163)  (0 = rest)

        Field 5: tie (in case of multiple pitches and ties)

            0 = no tie
            1 = tie to following note

        NOTE: In case of multiple pitches, there will be multiple
                A P ...  records.  A P ... records should not precede
                A D ...  records


  G. Super-objects (these are things whose shapes depend on the
           locations of more than one object.  They generally follow
           after all of their associated objects have been described

     Field 1: Identifier = H or P

                "P" is used in place of "H" to indicate that the
                super-object should be displayed in color (in whole
                or in part) rather than black.  This feature applies
                only to Ties and Beams and is used for highlighting.

                If "P" is used, then field 2 must start in column 3.

     Field 2: super-object number
     Field 3: type of super-object

                B = beam
                T = tie
                S = slur (dotted slur)
                X = tuple/bracket
                W = wedge
                D = word plus dashes
                E = ending
                R = long trills
                V = octave transposition
                F = figure extension
                N = null super-object (no action)

     Fields 4--: depend on the super-object type

     G-1. Beams    (type = B)

        Fields 4 and 5:

          Case I: Non-page-specific i-files

            Field 4:  0 = beam above first object
                      1 = beam below first object

              Note: If the stage2 source file contained a print suggestion
                    for modifying the length of the first stem of a beam,
                    this suggestion will be communicated to the mskpage
                    program in this field.

                    a) If the stem is to be lengthened (regardless of
                       whether it goes up or down), the suggested amount
                       of the change (in units of 1/10 of a staff line
                       distance) is multiplied by 100 and added to Field 4.

                    b) If the stem is to be shortened (regardless of
                       whether it goes up or down), the suggested amount
                       of the change (in units of 1/10 of a staff line
                       distance) is multiplied by 10000 and added to Field 4.

                    These print suggestions apply only to non-page-specific
                    files and are removed by mskpage (and mkpart) after they
                    are utilized.  They do not appear in the page-specific
                    output.


            Field 5:  0 or 1 = stem directions are all the same
                    > 1 = binary representation of stem direction.
                          If field 4 = 0, then bit set = stem up;
                          if field 4 = 1, then bit set = stem down;
                          Example: if field 4 = 1, then 101001010 =
                          down, up, down, up, up, down, up, down, up

          Case II: Page-specific i-files

            Field 4:  length of first stem (positive = stem up)

            Field 5:  slope of beam

        Field 6: font number
        Field 7: number of associated objects
        Fields 7a, b,...   beam codes

            beam code = up to 6 digit number

               0 = no beam
               1 = continue beam
               2 = begin beam
               3 = end beam
               4 = forward hook
               5 = backward hook
               6 = single stem repeater
               7 = begin repeated beam
               8 = end repeated beam

                    1's digit = eight level beams
                   10's digit = 16th level beams
                  100's digit = 32nd level beams
                 1000's digit = 64th level beams
                10000's digit = 128th level beams
               100000's digit = 256th level beams

     G-2. Ties    (type = T)

        Fields 4--6:  locations of note heads to be tied

          Field 4: vertical position of tied notes (position
                     measured in dots (300/inch) from top staff
                     line)

              For ties on a virtual (second) staff, 1000 will be added
                to this value.


          Field 5: horizontal displacement of first note head from
                     associated object (non-zero only with chords)
          Field 6: horizontal displacement of second note head from
                     associated object (non-zero only with chords)

        Fields 7--9:  Meaning depends on value of reset parameter (Field 11)

          If reset parameter is 0

             Field 7: (from print suggestion) horizontal displacement of
                        the left end of the tie relative to the default
                        computed position.  Tie length will be changed.
             Field 8: (from print suggestion) vertical displacement.
                        In most cases, displacement will be relative to
                        the default computed position.  Adding 10000 (ten
                        thousand) to the number will make the displacement
                        relative to the tied note, itself.

             Field 9: (from print suggestion) horizontal displacement of
                        the right end of the tie relative to the default
                        computed position.  Tie length will be changed.

          If reset paramter is > 0

             Field 7: horizontal displacement from note head one
             Field 8: vertical displacement from note head one
             Field 9: font number

        Field 10: situation flag

           interference     tips down       tips up
           stem   staff   space   line    space   line
           -----  -----   -----   ----    -----   ----
            no     yes      1       5        9     13
            no     no       2       6       10     14
            yes    yes      3       7       11     15
            yes    no       4       8       12     16

        Field 11: reset parameter flag (used in cases where
              post-editing has specified non-standard parameters)

            value    action
            -----    ------
              0      recompute position and font (normal setting)
              1      do not recompute position
              2      do not recompute font
              3      do not recompute position or font


     G-3. Slurs (dotted slurs)    (type = S)

        Field 4: situation flag

                       bit clear            bit set
                     --------------       -------------
            bit 0:   full slur            dotted slur
            bit 1:   stock slur           custom slur
            bit 2:   first tip down       first tip up
       (*)  bit 3:   second tip down      second tip up
       (+)  bit 4:   compute stock slur   hold stock slur
            bit 5:   print slur           don't print slur

            (*) used on custom slurs only
            (+) used on stock slurs only

        Field 5--8:  position fields:

            For stock slurs, these are displacements added to
               the note positions before the slur number is
               calculated.  Displacements are measured in dots
               (300/inch).  Horizontal positive is forward to
               the right; vertical positive is down.

            For custom slurs, these displacements are the actual
               starting point and ending points of the slur
               relative to the first and second objects.

            Field 5: (extra) horizontal displacement from
                        associated object one
            Field 6: (extra) vertical displacement from
                        associated object one
            Field 7: (extra) horizontal displacement from
                        associated object two
            Field 8: (extra) vertical displacement from
                        associated object two

        Field 9--10:  parameter fields:

            Stock slurs:  Field 9:   post adjustment to curvature
                          Field 10:  beam flag  1 = set slur above beam
                                                2 = set slur below beam

            Custom slurs:  Field 9:   vertical height parameter
                           Field 10:  length of flat portion as a
                                      percent of span between end
                                      points

        Field 11--12:  more parameters

            Stock slurs:  Field 11:  post adjustment to x-position
                                     negative = left; positive = right
                          Field 12:  post adjustment to y-position
                                     negative = up; positive = down


     G-4. Tuplets/brackets    (type = X)

        Field 4: situation flag

                       bit clear                  bit set
                     --------------             -------------
            bit 0:   no tuplet number           tuplet number
            bit 1:   no bracket                 bracket
            bit 2:   tips down                  tips up
            bit 3:       0                      tuplet aligns with beam
            bit 4:   tuple near notes           tuple near beam
                     (bit 4 has meaning only if bit 3 is set)

            bit 5:   break bracket for number   bracket is continuous
                     (bit 5 has meaning only if bit 0 is set)
            bit 6:   number outside bracket     number inside bracket
                     (bit 6 has meaning only if bit 5 is set)
            bit 7:   bracket is square          bracket is curved
                     (bit 7 has meaning only if bit 1 is set)

        Field 5:  tuplet number:  value = 1000 * [second number] + first
                    or single number

        Fields 6--10: parameters

          Case I: bit 3 of sitflag = 0 (tuple not aligned with beam)
               or bit 3 of sitflag = 1 and bit 4 of sitflag = 0
                  (tuple aligned with notes of beam)

            Field 6:  horizontal displacement from
                            associated object one
            Field 7:  vertical displacement from
                            associated object one
            Field 8:  horizontal displacement from
                            associated object two
            Field 9:  vertical displacement from
                            associated object two
            Field 10: not used

          Case II: bit 3 of sitflag = 1 (tuple aligned with beam)

            Field 6:  horizontal post adjustment to tuple

            Field 7:  vertical post adjustment to tuple

            Field 8:  horizontal post adjustment to tuple

            Field 9:  vertical post adjustment to tuple

            Field 10: beam super-object number with which tuplet
                            aligns

     G-5. Wedges    (type = W)

        Field 4: wedge size at left end

        Field 5: wedge size at right end

        Field 6: horizontal displacement from
                        associated object one
        Field 7: vertical displacement from
                        top of staff
        Field 8: horizontal displacement from
                        associated object two
        Field 9: vertical displacement from
                        top of staff

     G-6. Word(s) plus dashes    (type = D)

        Field 4: horizontal displacement from
                        associated object one
        Field 5: horizontal displacement from
                        associated object two
        Field 6: vertical displacement from staff lines

        Field 7: spacing parameter

        Field 8: font designator


     G-7. Endings    (type = E)

        Field 4: situation flag

                  0 = ending with no number
                  1 = first ending
                  2 = second ending
                  3 = third ending
                  4 = fourth ending

             The display/printing of ending superobjects is normally
             suppressed for all staff lines below the top staff line
             in a score.  That is, the superobjects may be there, but
             they are not displayed/printed.  Adding 10 to the situation
             flag will force the display/printing of an ending
             superobject.  For example:

                 11 = force the display/printing of first ending
                 12 = force the display/printing of second ending

        Field 5: horizontal displacement from
                        associated object one
        Field 6: horizontal displacement from
                        associated object two
        Field 7: vertical displacement from staff lines

        Field 8: length of left vertical hook

        Field 9: length of right vertical hook


     G-8. Long trills    (type = R)

        Field 4: situation    1 = no trill
                              2 = trill with no accidental
                              3 = trill with sharp
                              4 = trill with natural
                              5 = trill with flat
                              6 = trill with sharp following
                              7 = trill with natural following        "
                              8 = trill with flat following           "

        Field 5: horizontal displacement from
                        associated object one
        Field 6: horizontal displacement from
                        associated object two
        Field 7: vertical displacement from
                        associated object one


     G-9. Octave transposition     (type = V)

        Field 4: situation    0 = 8ve up, 1 = 8ve down
                              2 = 15 up,  3 = 15 down

        Field 5: horizontal displacement from
                        associated object one
        Field 6: horizontal displacement from
                        associated object two
        Field 7: vertical displacement from
                        associated object one
        Field 8: length of vertical hooks


    G-10. Figure extension    (type = F)

        Field 4: level   1,2,3 or 4

        Field 5: horizontal displacement from
                        associated object one
        Field 6: horizontal displacement from
                        associated object two

        Field 7: additional vertical displacement from
                        default height

  H. Q-records (linear only)

     Field 1: Identifier = Q  (one field only)

        A Q-record is used to force MSKPAGE to make a page
        break.  At the moment, these are not incerted by
        AUTOSET; they must be added by hand to a linear
        i-file.

  I. Tag records (linear only)

     Field 1: Identifier = Y

       Tags are incerted into linear i-files by AUTOSET.  Their purpose
       is pass instructions and information to MSKPAGE.  They are not
       propogated into page-specific i-files.

     Field 2: type of tag     P = print abbreviated part name
                              U = line control

     I-1. Part name (type = P)

        This tag signals MSKPAGE that from this point in the score
        (until cancelled), MSKPAGE is to append the designated
        abbreviated part name to the left of the staff line containing
        this part.

        Field 3: font number  (0 = turn this feature off after
                                   typesetting the current line)

        Field 4: ASCII string (includine font designations)

     I-2. Line control (type = U)

        This tag sends MSKPAGE instructions to help it determine
        whether or not to DELETE this part from a particular system.
        MSKPAGE starts by typesetting (logically) all parts in the
        system.  It then uses these tags to determine which parts
        in a particular system are "extraneous."

        Field 3: control code   0 = nullify the previous control code
                                    (i.e., return to normal rules)
                                1 = tag the current and all subsequent
                                    measures as type-1
                                2 = tag the current and all subsequent
                                    measures as type-2

          Type-1 measure:  Delete a line of music if every one of its
                           measures is type-1.  This is used to suppress
                           a "dominant" representation of a part.

          Type-2 measure:  Delete a line of music if any one of its
                           measures is type-2.  This is used to suppress
                           a "non-dominant" representation of a part.

            The "dominant" version of a part is the one which contains
            the most or greatest variety of information.

            For example, when a section (e.g., Violin I) divides into
            two sections (divisi) and is represented on two staves,
            this becomes the "dominant" version of the part, because on
            a system-line basis it over-rides and must replace the single
            line version of the part.  The divisi parts will be flagged
            as type-1 up to the point where they diverge (or where the
            music engraver definitely wants them on two staves).  The
            single line part will be normal (considered printable) up
            to the point where the divisi parts take over and will be
            flagged as type-1 thereafter.  At the point where the
            divisi goes away, the divisi parts are again flagged as
            type-1 and the type-2 designation for the single line
            is cancelled.



V. Record Formats for Page Specific Record Types (page-specific,
    scrolling page-specific, and multi-page)

  A. Systems

     Field 1: Identifier = S
     Field 2: 0
     Field 3: x co-ordinate of system on page
     Field 4: y co-ordinate of system on page
     Field 5: horizontal length of system
     Field 6: vertical length of system
     Field 7: number of lines in system
     Field 8: control string in double quotes (null string allowed)

            The control string describes how the lines of the system
            are bound together on the left, and where bar lines stop
            and start between the staff lines.  The control string
            consists of the following elements:

               [  =  start bracket
               ]  =  end bracket
               {  =  start brace
               }  =  end brace
               (  =  start bar line
               )  =  end bar line
               .  =  part with a single staff line
               :  =  part with a grand staff (piano)

            What defines a system (holds it together) is a single
            vertical line on the left.  To the left of this line can
            appear an arrangement of brackets and braces, which help
            the reader group the lines in the system.

            The bar lines in a system have their own (vertical) pattern
            of starting and stopping.  These are indicated in the control
            string by the placement of the '(' ')' characters.  The '('
            character indicates the start of a bar line on the top staff
            line of the part indicated by the next dot (.:,;); and the
            ')' character indicates the end of a bar line on the bottom
            staff line of the part indicated by the previous dot (.:,;).

            Examples:   (.)     creates a bar line through one set of
                                  staff lines
                        (:)     creates a bar line through a grand staff
                                  (two sets staff lines)
                        (..)    creates a bar line which starts at the top
                                  of the first staff line and ends at the
                                  bottom of the second staff (i.e., two
                                  independent parts with a bar line
                                  connecting them together).
                        (.)(.)  creates a bar line through each of the
                                  staff lines separately, with no
                                  connection between them.

            All segments of a system bar line must be specified.  The
            control string "{..}" will produce a a brace on the left, but
            no bar lines will appear.  In this case, one must either
            specify "{(.)(.)}", meaning two bar lines with no connection,
            or "{(..)}", meaning one bar line crossing both staff lines.

            The control string "(....)(.):{..}..." would produce the
            funny looking result of having no bar lines in the piano part
            or in the (string) parts below them.

            As a short cut for convenience, the '[' character also doubles
            as a '(' character, and the ']' character also doubles as a ')'
            character.  This allows one to write "[....]" as a shortcut for
            "[(....)]".  Shortcuts are handy, but they can have unintended
            side affects.  In the example above, modifying control string
            with outside square brackets (i.e., "[(....)(.):{..}...]" )
            will produce bar lines for all staff lines, but may not connect
            the staff lines in a manner the user intends.  This control
            string will produce a solid bar line from the fifth staff line
            through to the bottom.  If this were the control line for a
            score of a piano concerto with four woodwinds, timpani, piano
            and strings, it might have the following form.
                           "[(....)(.)(:)({..}...)]"

            The staff lines in a system are normally of the same size, but
            it is possible to combine staff lines of different sizes in the
            same system.  When this is done, the print and display software
            must decide what font size to use for printing the various bar
            lines, braces, and brackets.  At the moment, the decision is
            made in the following way:

            1. The left line, and all braces and brackets are printed
               using the same font size.  The default is to use the font
               size from the bottom staff line.  If the size from a
               different staff line is desired, this should be indicated
               by replacing the dot for that staff line with a comma, or
               in the case of the grand staff replacing the colon with a
               semi-colon.  If more than one comma/semi-colon appears, the
               largest of the indicated sizes will be used.

            2. As stated above, bar lines can run through one or more staff
               lines, and they can be printed in different sizes (widths).

               (a) If a bar line runs through only one staff, then it is
                   printed in the font size of that staff.

               (b) If a bar line runs through multiple staves of the same
                   font size, then it is printed in that font size.

               (c) If a bar line connects staves of different font sizes
                   the default size for the bar line is given by the
                   size of the bottom staff.  If the size of another
                   staff is desired, this is indicated by replacing the
                   dot for that staff line with a comma, or in the case
                   of the grand staff replacing the colon with a semi-
                   colon.  If more than one comma/semi-colon appears,
                   the largest of the indicated sizes will be used.

            3. These rules will work in most situations, but there are some
               limitations.  In cases 2(a) and 2(b) you cannot use a font
               size different from that of the staff size to print bars.
               You cannot, for example, use one font size to print all
               segments of a bar line in the case where the segments span
               staff sets of different sizes.  It is possible to remove
               these limitations, but this would require a more elaborate
               definition of the control string (such as allowing the
               font size for each bar segment to be specified explicitly).
               At this stage of development, such an elaboration seems
               excessive and unnecessary.


  B. Music Lines (page specific)

     Field 1: Identifier = L (or small l = single line staff, e.g., for triangle)
     Field 2: y off-set in system
     Field 3: text off-set(s) from line   (separated by |)
     Fields 4--7:  parameters carried over from previous line

       Field 4: dyoff(s)
       Field 5: uxstart(s)
       Field 6: backloc(s)
       Field 7: xbyte(s)   (length of field = number of text lines)

     Field 8: y off-set to virtual staff line (0 = none)
     Field 9: notesize (optional; 0 = none)
     Field 10: additional vertical off-set for figured harmonies (0 = none)
                 Note: Field 10 cannot exist without Field 9


  C. End of Music Line

     Field 1: Identifier = E
     Field 2: xbyte(s) at end of line (length of field = number of bytes)


  D. System Bar

     Field 1: Identifier = B
     Field 2: bar type     Single bars   Double Bars
                           -----------   -----------
                           1 = simple    5 = double light
                           2 = heavy     6 = light-heavy
                           3 = dotted    9 = heavy-light
                                        10 = heavy-heavy
                                        25 = heavy-light with
                          99 = no bar            repeat dots

     Field 3: x off-set from beginning of line
     Field 4: number of breaks in line (to avoid interference with
                  text and other designations)
       Fields 5--: parameter pairs, beginning and ending of breaks


  E. Page Text

     Field 1: Identifier = X
     Field 2: font number
     Field 3: x co-ordinate of beginning of text
     Field 4: y co-ordinate of text line
     Field 5: ASCII text

     The ASCII text may contain font change instructions.  The sequence
     is the "!" character, followed by a two digit number (which at the
     moment must be between 30 and 48), followed optionally by the "|"
     character, which serves as a terminator to the font change instruction.
     The need for a terminator arises because in the future we may not want
     to restrict the font change number to exactly two digits.  If the
     display software finds a "|" following a string of digits following
     the "!" character, it will consider this a terminator and remove
     it from the text along with the font change command.  To print/display
     the "|" character immediately following a font change, use the "|"
     twice.

     Special cases:

     (1) If the page text record has only two fields, the letter X and
         a number, then the number is interpreted as a new value for
         notesize.

     (2) If the number in Field 3 (x co-ordinate) is followed directly
         by the letters "C" or "R", this means (C) center the text on the
         x co-ordinate, or (R) right justify the text to the x co-ordinate,
         respectively.


  F. Tagged Page Text

     Field 1: Identifier = Y
     Field 2: font number (if this = 0, then no print/display)
     Field 3: x co-ordinate of beginning of text
     Field 4: y co-ordinate of text line
     Field 5: ASCII text

     In addition to the font change instructions in "X" records, these "Y"
     records also support a system of tagged fields.  Listed below are
     the tagging sequences and their functions.

        \< = begin tag structure.  A tag structure has two components:
               the tag, and the data field being tagged.  The tag comes
               first and may contain any ASCII characters except the "|"
               character, which is interpreted as "End of Tag."  Immediately
               following the "|" is the data field.  This is terminated
               by the "end tag structure" sequence.
        \> = end tag structure.
        \] = break tag structure.  It is possible for a tag data field to
               continue over to a line or page boundary.  In this case,
               the tag structure must be broken at the end of the line
               and restarted at the beginning of the next line.
        \[ = restart tag structure.  A tag structure can only be restarted
               if it was broken on the previous line (or last line of the
               previous page).  It is a requirement that the tag (first
               component of the tag structure) be restated.  In theory,
               the software could reconstruct the tag, but in the case
               where the data field extends over several lines, this
               could prove cumbersome.  Therefore, following the "\["
               sequence comes the tag, followed by the "|" character,
               followed by a continuation of the data field.

     Under this structure, it is possible to have nested tags.  This
     feature could prove useful.  For example, suppose we are interested
     in tagging names, and suppose we are also interested in knowing
     the first name and the last name.  We could implement this with
     three tags: name, fname, and lname.  Now suppose we want to tag
     the name in the line of text "According to Lowell Lindgren,"
     The Y record might appear as follows:

     Y 37 50 140 According to \<name|\<fname|Lowell\> \<lname|Lindgren\>\>,

     This includes both the "outer" or "primary" tag, with its endpoints
     and "inner" or "secondary" tags with their endpoints.

     If this structure were broken across a line boundary, it might
     look like this

     Y 37 50 140 According to \<name|\<fname|Lowell\>\]
     Y 37 50 180 \[name|\<lname|Lindgren\>\>, there was a cantata

     What this allows the software to know is that the first name "Lowell"
     and the last name "Lindgren" are part of the same name, even though
     they occur on separate lines.


  G. Page delineator (Multi-page files only)

     Field 1: Identifier = P

     To distinguish a page delineator from a "P" type super-object,
     one of the following conditions must hold.

         (a) Record length = 1
         (b) "space" character in column 3
         (c) The first 4 columns must be the word "Page"

     The last record in a multi-page file must be the four
     character string "/eof".  Multi-page files are designed
     primarily for use as (one form of) input to the muse2ps
     conversion program.


  H. Meta-data records

     Field 1: Identifier = @

     This line is ignored by the display/print software, but may contain
     important information related to the description or structure of the
     data.

     Some uses of @-records:

     1. Source ID.

        Field 2: "SOURCE:"

        Field 3: <File ID>

        Filed 4: <[md5sum: ... ]>

        The idea is to represent all of the sources that contributed to
        making this page (or musical example).  The Source ID records
        should normally follow the Z-header records.  In cases where
        there are no Z-header records, the Source ID record should
        appear before the first system (and before any @ SYSTEM: records).

     2. Movement ID. + Info

        Field 2: "MVT:"

        Field 3: Type of data

          N = name  Field 4: = movement name

          P = part  Field 4: = part number
                    Field 5: = <sub-part numbers> enclosed in carrot brackets
                                numbers separated by commas, null-set allowed
                    Field 6: = part name

          L = line  Field 4: = line index number
                    Field 5: = <part numbers for each track> enclosed
                                numbers separated by commas, no null set allowed
                                If the first field inside the <..> string is an
                                "x", this indicates a keyboard part, which is
                                non-divisible and multi-track.  The numbers that
                                follow, e.g., "<x,#,#,..>" are the part numbers
                                assigned to the various keyboard tracks.
                    Field 6: = line name

          Note: "part" means any part or group of parts.  If there are two
                   oboes, then there are likely to be three parts: oboe-1,
                   oboe-2, and oboes.
                "line" means any line compiled by AUTOSET for purposes of
                   making a score.  For the bass strings, there are often
                   three lines: cello, basso, and cello + basso.  These
                   three lines never appear in the same system; either
                   the cello and basso appear as separate lines, or they
                   appear combined on the same, single line.

     3. System ID.

        Field 2: "SYSTEM:"

        Field 3: ASCII string

           The purpose of this record is to identify the composer,
           work, and movement to which this system belongs.  This
           information (if present) will be used by the search program
           to identify "hits."  This record should be located just
           above the System record to which it applies.

           There is an option to include a bar line number in this
           ASCII string.  If this is to be used, the number must
           occur at the beginning of the string, and must be enclosed
           in < > brackets, e.g., <28>, with no space following.

     4. Line ID.

        Field 2: "LINE:"

        Field 3: line index number (for joining lines between systems)

        Field 4: ASCII string

           The purpose of this record is to identify the instrument(s)
           or the voice represented on this staff line.  This information
           (if present) will be used by the search program to identify
           "hits."  This record should be located just above the Line
           record to which it applies.

     5. Line Text(s)

        Field 2: "TEXT:"

        Field 3: ASCII string

           The purpose of this record is to provide a searchable ASCII
           string containing the text (underlay) represented on the line.
           This record should be located just below the Line record to
           which it applies.  In cases where there is multiple text
           underlay (e.g., some Bach chorals), there will be mulitple
           @ TEXT: records.

     6. Comment.

        Field 2: "COMMENT:"

        Field 3: ASCII string

     7. Grid data. (scribe program)

        Field 2: "GRID:"

        Field 3: Command Set (includes grid units per measure)

        Field 4: Notation represented by one grid unit, e.g. 7 = quarter note

        Field 5: Minumum grid spacing

           The purpose of this record is to provide the scribe program with
           parameters it needs to implement cursor movement along the grid
           This record, if it exists will alway be at the top of the file.

     8. Pointer data. (scribe program)

        Field 2: "POINTERS:"

        Field 3: pointer to the next note (or rest) object (index into pointers array)

        Field 4: pointer to the previous note (or rest) object (index into pointers array)

           The purpose of this record is to store various pointers used by the
           scribe program.  The values have meaning only in the context of that
           program and of the working pointers it creates.  The @POINTERS: record,
           if it exists will always following directly after the object record to
           which it refers.