 Notes:

      [2]  Syllabaries in general do not distinguish between vowels and
      consonants. They do not have a regularity, symmetry, or any kind of
      matrix. The ancient Cypriot syllabary, the newly invented (1821?)
      Cherokee syllabary, and even the modern Japanese Katakana system have
      very little regularity.  The syllables are arbitrary marks that contain
      the consonant and vowel in one symbol that does not allow any disassembly
      into constituent phonetic parts.  This aspect of syllabaries is
      sometimes used to show that they are a less developed form of writing
      system.

      Fidel, the Ethiopic syllabary, on the other hand, is highly regular and
      has a quite clearly defined set of vowel markings which were *added* to
      a base consonant/radical.  In fact, just looking at the Fidel one can
      see a clearly outlined alphabetic underpinning.  It seems that what we
      have in the Ethiopic system is not a sort of pre-alphabetic syllabary,
      but an outright further development of an alphabet--the Sabaean alphabet
      (1) which, like Hebrew and Arabic had no vowel representation.  It is more
      plausible to see in Ethiopic a move from the Sabaean alphabet back into
      a syllabic form, but with a standardization based on awareness of the
      explicit relationship between consonants and vowels.  To dismiss
      Ethiopic as an underdeveloped writing system is to overlook an example
      (perhaps the only one) of post-alphabetic development in writing
      systems.

      As Coulmas Observes :

      "     Old Ethiopic was first written in the typical Semitic manner
      without vowel signs, but in the fourth century AD the writing of
      Ethiopic underwent drastic changes.  The resulting script is interesting
      because of its unusual and quite un-Semitic vowel indication. " (2)


      [3]  The traditional matrix of the Ethiopic script is in rows of seven
      columns with the base "W" form (here column 8), and the other "W" forms
      (here column 9-12) presented separately at the end of the list as
      extensions to the core set.   We felt a 37 x 12 matrix would be a more
      natural arrangement both because of the similarity of the characters
      as well as the efficiency it provides for machine access and
      manipulation of the characters.

