Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

download Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

of 47

Transcript of Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    1/47

    The fuj)ark

    and ogam): order

    as a key to

    origin

    1. Introduction

    Do the Germanic runes derive fromone orotherof the

    Mediterra-

    nean

    alphabets?

    1

    If so,

    which one:

    Greek, Latin, north Italian,

    Etrus-

    can?And why are therunes, whenset out as a

    series,arranged

    in non-

    alphabetical

    order,

    starting

    with

    the

    charactersfupa

    r k

    (Fig.

    1),

    which

    have given the series its traditional name?

    Previous

    attemptsto

    determine

    the

    origin

    of the

    runes have mostly

    concentrated on the shape of individual runes and the similarity to

    particular letters.

    2

    The unorthodox

    order

    in the fuj>ark has been con-

    sidered

    interesting but not of

    primary

    concern

    in

    determining

    which

    alphabet, ifany,mayhave served as amodel. After all,it is not known

    whether the runes were arranged in

    fufcark

    order

    from their inception

    or whether they were re-arranged some timelater.The earliest known

    1

    The word alphabet is reserved here for a sequence ofletters in alphabet order.

    This is important in

    view

    of the distinction that came to be made between, on

    the one hand, Germanic words for letters of the (Latin)

    .alphabet,

    as used in

    "books"

    - AS.

    bocstcef,

    ON.

    bokstafr,

    OS.bokstaf, OHG.

    buohstap,

    NHG.

    Buch-

    stabe

    -

    and,

    on the

    other,

    OS.stab,AS. (run)stcef ON.

    (runa)stafr

    for

    runes,

    thus indicating that the two systems were probably perceived as functionally di-

    stinct. The first record of the word

    alphabetum

    seems to be that by Septimus

    Florens Tertullianus (d. AD 220). Beforethat the Latin word for the series of

    letters

    was

    elementa.

    It is

    noteworthy that

    the

    Latin alphabetum uses

    the

    Greek

    names of the first two

    letters,

    reflecting the

    predominance

    of

    Greek

    in

    things

    grammatical. The pure Latin

    abecedarius

    or abecedarian is later

    (fifth

    century

    AD). The Greek seem to have been distinguished from in

    the same way that Priscian distinguished between

    elementa

    (or

    elementa litterarum)

    and

    litterae, i.e. "elementary sounds"

    and

    "letters".

    On the

    other hand,

    means

    "in the

    order

    of the

    letters",

    i.e.

    alphabetically, following

    the

    basic m eaning

    ofof

    "one

    of a

    series,

    a

    component part,

    an

    element",

    cf.

    themeaning ofGothic stabeis(nom.plur.)= Grundstoffe (raw materials, ele-

    ments).

    2

    Morris (1988) gives

    a

    good overview

    of

    earlier discussions

    of the

    origin

    of the

    runes. Currently the most favoured prototype amongst

    runologists

    is the Latin

    alphabet in one form or another, see, for example, Williams (1996) and Quak

    (1996).

    Indogermanische Forschungen, 104. Band 1999

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    2/47

    Thefujmrk (and ogam) 165

    inscription to list a

    fujmrk

    is in fact quite late, probably not earlier

    than

    the first

    half

    of the

    5th century

    AD

    (Fig.

    la),

    compared

    to the

    date

    of the first

    known inscriptions using runes,i.e.

    the 2nd to 3rd

    centuryAD.

    Recently,

    however, Elmar Seebold has not only tackled the question

    ofsign order but has used it to substantiate his matching of the runes

    with

    a"proto-tyrrhenian"alphabet (Seebold,1993:417).

    3

    Amajorpro-

    blem

    with his

    approaches

    that he has to resort to manipulations that

    tend to

    weaken

    his

    argument;

    For

    example,heassociates

    theta

    with

    a

    "tektaler Reibelaut",

    namely

    the

    rune,

    and

    then

    has

    to

    reverse

    the

    order of eta ana

    theta.

    He also has to introduce a extra, "dummy"

    sibilant

    after sigma,otherwisehis

    method

    of counting willnot work.

    A

    key

    step

    in the

    procedure

    he

    proposes

    is to

    writeout

    the

    alphabet

    so

    as

    to formpairs of letters which are then systematically re-arranged

    to yield the

    fufcark

    order. Such considerations fit in with his suspicion

    that therewasprobablya"magisch-mantisch

    begrndete"

    basisto the

    re-arrangementof thealphabet and itsuse, evenif themethodhesug-

    gests

    is

    "nicht

    weiter begrndbar" (Seebold, 1993:

    420-1).

    4

    The bulk of the present

    article

    waswritten intotalignorance of See-

    bold's

    work

    5

    ; but in fact it

    will

    be

    seen

    to

    have

    a

    broad similarity

    in

    approach. Nevertheless the basic idea of matching letter pairs to rune

    pairswasarrivedat independentlyand isthus workedout quite

    diffe-

    rently. The result, it is hoped, avoids some of the

    difficulties

    of

    Seebold's

    thesis. At the same time it goes beyond Seebold'sconclusion in that it

    3

    PreviouslySeeboldsuggested

    (1991: 30) that the

    model

    was

    "ein

    hoch-archaisches

    Alphabet

    aus dem

    Kreis

    der lateinisch-faliskischen Schrift"- Themodel

    alphabet

    heproposes reads:

    ab

    k d e f z ? h i

    g

    l

    m

    nng o p

    w

    r s

    (s) 114

    j th.

    The ?

    sign

    is a"tektaler Reibelaut"

    originally

    in the theta

    position before

    it and

    eta were reversed.

    Theyand thare

    supplementary

    characters

    added

    to the

    end;

    the (s) is an

    extra

    sibilant.Theng

    takes

    theplace?

    of

    xi,thek the

    place

    of

    Italic

    C,

    the

    g

    the

    place

    of

    "redundant"

    kappa.

    The

    w

    takes

    the

    place

    of

    qoppa.

    4

    Numerousarrangements of thealphabetfor

    purposes

    ofcultormagic are dis-

    cussed by

    Dornseiff

    (1925).

    5

    I am

    grateful

    to

    Sean Nowak

    for

    drawing

    my

    attention

    to

    Seebold's work

    as

    well

    as fo r substained encouragement and critical assessment. wouldalso like

    to

    thank Professor Klaus Dwel

    for his

    support

    throughout the

    genesis

    of

    this

    article,

    theessenceof

    which

    was

    presented

    to the

    Seminar

    fr

    Deutsche

    Philologie

    at

    Gottingen

    University

    on 28January, 1997,

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    3/47

    166

    Alan Griffiths

    indicates

    how the

    question

    of

    order

    may

    also help

    to

    elucidate some

    of the rune shapes.

    Sincethis

    emphasis

    on

    order

    differs

    from

    previous

    approaches,

    which

    have

    largely

    reliedon a comparisonof the shapesand sounds ofindi-

    vidualcharacters rather than considering

    the

    alphabet

    and the

    fufcark

    as integrated systems,webeginby consideringwhy runologists have

    so far

    failed

    to providea convicing solution to the problem of runic

    origins

    (2.1)

    and

    what

    can be

    done about

    it(2.2-2.4).We

    then list

    the

    earliest fu]parks and the principal alphabets that may be considered

    eligible as a source for the runes

    (3.1-3.2).

    When these are matched

    on the

    basis

    of

    order,

    it

    will

    be

    found that one,

    and

    only one,

    of the

    alphabets corresponds.sufficiently closely to the oldest

    form

    of the

    fuj^ark to merit

    further

    consideration (3.3) despiteanumberofdiscre-

    pancies. This encourages

    an

    attempt

    to

    demonstrate

    how the fufcark

    order can be systematically derived from a form of thealphabet that

    is

    closer to the Greek alphabet than to the Latin (4). Several difficulties

    relatingto the equivalence of sound values can be explained in relation

    to the way the

    devisers

    of the

    Gothic alphabet

    and the

    Irish ogam

    derived their schemesfrom a Greek alphabet (5). Infactthe ogam order

    of

    signs

    can be

    shown

    to

    have originated

    from the

    same alphabet

    as

    the fufcark, using a similar method (Appendix 1). By looking at the

    runic sound values and sign shapes from the point of

    view

    of a Greek

    alphabet order we discern patterns in both the phonetic structure (6)

    andgraphic

    "design"

    (7) of thesystem, which suggestacertain rationali-

    zation

    of the

    source alphabet before

    its

    re-arrangement

    as the

    fujmrk.

    Theconclusion to bedrawn

    from

    this analysis(8) isrestricted to the

    type

    of

    Mediterranean alphabet that formed

    a

    basis

    for the

    runes

    and

    the question ofhowthe fuj)ark mayhave been created. It leaves open

    thequestionswhen,whereand why,about whichit isstill only possible

    to speculate, even given

    SeebokTs

    arguments in favour of a

    "magisch-

    mantisch

    begrndete" motivation.

    2.Method:past

    failure, new rigour

    2.1 The runologists conundrum

    Williams

    (1996: 21Iff.)has pointed outthattwo unsolved problems

    that haveso farmadeit

    difficult

    toconvinceamajorityofrunologists

    of any one particular theory on the origin of the runes are:

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    4/47

    The fujpark (and ogam) 167

    1.

    the poor fit between theform of many of the runes and the sound

    values

    they would be expected to have

    on

    the basis of the corre-

    sponding shapesof thelettersin thealphabets (e. g.Latin P [p] and

    Greek P[r] fitrunic

    w

    inshape but not sound);

    2.the

    inadequacy

    of the

    inventory

    of

    letters

    in the

    supposed Medi-

    terranean source alphabets

    to

    supply

    all the

    runes

    in the

    fu{>ark;

    this means assuming thatnewrunes were invented(e. g.runicd) or

    that

    one

    letter gave rise

    to two

    different runes,

    or

    that

    two

    alphabets

    were

    mixed to provide the full complement of runes

    (e.g.

    runicw

    from

    Latin

    P

    and

    runic

    z

    from

    Greek

    ).

    These,however,are realities and not the real problem, which

    is

    the

    degree

    of freedom

    allowedif one accepts the practice ofpickingfrom

    this alphabet or that and resorting to "inventions" if all elsefails. It

    may

    very well

    be

    that

    the

    creation

    of thefufcark was

    ad

    hocand

    arbitrary;

    indeed the apparently

    archaic

    nature of the oldest inscriptions would

    appear to support the idea of arbitrary borrowings and wilful adapta-

    tions

    by

    "barbarian"

    people.

    On the

    other hand, anyone

    who

    has

    stu-

    diedthefufcarkin anydepth must concede that, when takenas awhole,

    itis anintricateand sophisticated systemon a parwith Mediterranean

    alphabets. This

    is

    implicit,

    for

    example,

    in

    theories whichpostulate

    the

    application of certain generalprinciplesin the form ation of the shapes

    oftherunes. Such principlesas "no

    horizontal

    strokes"-presumably

    in

    connexion

    with

    writing

    on

    grainy wood-

    do

    not, however, bring

    usmuch closer to origins, evenifthey do seem to apply.

    6

    The apparently unavoidable need to

    "pick

    and mix"

    from

    this or

    that alphabet

    has

    bedevilled

    the

    study

    of

    runic origins

    and

    even today

    there

    is a

    lack

    of

    methodological rigour.

    To

    quote Williams (1996: 212)

    in extenso:

    "Theenigmatic

    rise

    of theruneshaslong

    baffled

    the

    runologists.W hy

    is it that

    this intricate, interes tingand important question has solong

    escaped

    its solution?

    I think

    we

    must

    go

    back

    to the first

    scholarly treatment

    of the

    problem,

    by

    Ludwig

    Wimmer

    in

    1874.

    In his

    m onograph

    on the

    origin

    and

    development

    of

    the runes

    he

    establishes

    a

    method which

    has

    since been followed

    by

    later runo-

    logists: when a rune coincides in shape and sound value with a letter in the

    supposed source

    alphabet,

    itsorigiiiis

    considered

    certain.When thisis not the

    6

    An exception to thisparticular"rule'*wouldbe

    the

    formof theerune with the

    straight horizontal bar,although it might beargued thatachange from this to

    the

    M shape infactproves therule

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    5/47

    :,

    ih

    tiw

    origin,

    and

    in

    shape

    arxgm-j) s.om;. MiuaJly

    ^uiUi

    fancifuJ and unproveorevenynprovitbte,expia-

    2,2 Suhttaweand circumstance

    Tite

    f)wcc^n> WjlJiarnu -dciycribes is familiar enough, but there

    is

    an

    even more fundamental problem lurking

    in that

    expression "the sup-

    posed

    source

    alphabet",

    since

    it

    raises

    the

    question

    as to how the

    sup-

    position

    isarrived at,

    inessence there

    are fiveaspects

    that

    are

    usually

    taken

    intoconside-

    ration in

    exploring

    the

    origins

    of the

    fujwk. These

    can be

    subdivided

    into

    three

    aspects

    that relate to the linguisticsubstanceof the fufcark

    and any

    potential source alphabet (substantial aspects)

    and two that

    relate

    to the

    circumstances

    in

    which

    the fu^ark may

    have

    come

    into

    being (circumstantial

    aspects),

    The

    three

    substantial

    aspects are:

    ,

    the

    shapes

    of

    individualrunes

    and of the

    letters

    of

    candidate

    source

    alphabets,

    2. the

    sound

    valuesofsame,plusthe

    phonemes

    ofcandidate

    German

    dialects where the fu^ark may bedeemed to have been formed or

    introduced,

    3. the

    sequence

    of the

    runes

    in thefujwk,as firstattested in

    inscrip-

    lions

    l ike

    those

    found

    at

    Kylver,

    The twocircumstantialaspectsare;

    1. the geographical and chronological contexts,

    2, thepossible

    processes

    oftransfer intoa Germanic context

    -

    such

    us

    d i f fus ion or implantation,

    Eiich of these aspects is in itself composed of a cluster of factors,

    which

    we need not analyse in detail here, The importance in drawing

    attention to

    them

    is that

    thepriority given to any oneover and above

    theother h t i K

    generally

    led toinvestigators

    excluding

    certain

    possibilities

    almost

    it

    priori,

    thereby biasing the direction of their investigations.

    An example js

    the

    priorily

    often given

    to

    the

    circumstantial aspects

    and in particular theimportance attached to theidea,whetherexplicit

    or i m p l i c i t , t h a i

    th

    use of runes reached the area where most of the

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    6/47

    The fufrark (and

    ogam)

    169

    earliest inscriptionsin theolderfufcark have beenfound,i.e. Denmark

    andNorway,

    by a

    process

    of

    diffusion;

    that

    is

    to

    say,

    by a

    process that

    would

    implygeographical contiguity between cultura l groups involved

    in

    the transfe r and a chronologically continuous evolution of sign form s

    rather than,say, theimplantationof an intention ally modified alphabet

    onaparwiththeintroductionof the Gothic alphabet in theBalkans

    byWulfiia.

    Without any documentaryevidence,"intentionalmodification" and

    thedeliberate introduction of asystemare ofcourse difficult to prove

    circumstantially.

    O n the other h and we should not allow circum stantial

    considerationsto blindus to the substantial,

    i.e.

    linguistic, characte-

    ristics, which may point in directions we are as yet unable

    to

    prove

    chronologically

    or geographically. Even more importantly, we must

    not allow ourselves to be biased by circumstantial considerations into

    favouringone source over another

    from

    the outset. This seems to have

    been

    thecase with,for

    example,

    M oltke(1985:51),whodismissed"the

    Greek

    hypothesis"

    because

    he

    associated

    it

    with,

    in

    particular,

    von

    Frie-

    sen, whose theory"has

    now

    been abandoned

    for

    methodological

    and

    chronologicalreasons",

    andalso with Aage Kabell, whose contribution

    "floats well

    aboveground-level".

    The

    Etruscan alphabet

    was

    similarly

    excluded on chronological grounds- "thelaterwedate theoriginof

    the

    runes, the further intothe distance Etruscan recedes"(ibid.:61).

    Thisleft theLatin alphabet, whichheclearly preferred because ofsuit-

    able geographical and chronological contiguity to allow him to cham-

    pion

    the

    Danes

    who

    were,

    in his

    view,

    sufficiently

    close

    to but

    also

    independent of the Romans to

    have

    developed a separate system of

    writing(ibid.:64).

    7

    Williams (1996:213)even goes

    so far as to

    suggest

    that

    the circum stantial evidence is such that we may"guess"that Latin

    capital

    letters were the source of

    the runes,

    8

    Admittedly he does at

    7

    "Always

    independent

    of

    Rome,

    the

    Danes

    nevertheless

    maintained

    lively

    connec-

    tions

    with this greatcivilisation and its

    trading-posts

    and military camps along

    the

    Rhine,

    as the

    archaeological

    evidence

    demonstrates.

    If anyon e wanted to get

    to knowthe Romansand to learn

    f&ek

    writingsystem (with itsobvious advan-

    tages), he had no

    need

    to travel to Rome. The Rhine region, the

    Netherlands

    and Gaul

    aillay

    close

    at

    hand,and England

    was

    there

    aswell"

    8

    "Whynot try theeasywayout?Therunes migh t

    certainly

    beborrowed500years

    before

    they first occur on actua lly

    foundobjects.

    The runes might also be borrowed

    from

    an

    obscure,

    even

    yet

    unknown, variant of the

    three Classical

    alphabets.

    The

    runes

    might

    even be borrowed from

    different systems

    of script, picking and

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    7/47

    170 Alan Griffiths

    least treat

    the

    Latin alphabet

    as a

    complete system

    of

    signs

    in a

    given

    order.

    But his

    circum stantially driven choice

    of the

    Latin alphabet leads

    him

    to force the fit to such an extent that he has toassocate familiar

    signswith sound values totally unrelated to those onemight linguisti-

    callyexpect.

    9

    Of

    course, the

    problem

    of how the

    run es transferred

    to

    Scandinavia

    cannot be ignored. But rather than allowan external factor like this

    toprejudicethe candidature ofsourcealphabets, a"cleaner"approach

    might

    be to consider the

    fu^ark

    as such and workfrom the inside out.

    In

    other words,

    let us put the

    emphasis

    on the

    linguistic characteristics

    of

    the fujmrk as a set

    sequence

    of

    signs with

    fairly

    well (though

    not

    alwaysprecisely)definedsound values and then (and only then) consider

    the implications any conclusions might have for the problem of how

    the

    fu^ark might have reached Scandinavia. There

    is,

    after all, more

    linguistic evidence

    to go on

    than historical.

    At the

    same time,

    if the

    linguistics point to an historically unfamiliar solution, we must be wary

    of being drawn into

    fanciful

    explanations such as

    coloured

    Bugge's

    Greek theory (which supposed

    an

    Arm enian teaching

    Goths

    the

    runes

    used

    in Galatia ).

    2.3

    Three

    principles

    To sum up sofar, the thesis that followsisbased on three principles:

    L Substancebefore

    circumstance

    Assumptions about circumstantial aspects must

    not be

    allowed

    to

    override evidence discernible

    in the

    linguistic substance

    of the

    fu]park.

    For example, there mustbe nopre-disposition towardsone orother

    choosing

    at a

    leisurely fashion.

    But

    this

    is not the

    easy

    way

    out.

    "The least forced starting point

    for a

    discussion

    of the

    origin

    of the

    runes

    is to

    assume

    that they

    were

    invented some time around

    the

    birth

    of

    Christ, give

    or

    take a century or so. We may also assume that the dominant culture of the

    time, the

    Roman empire,

    is the

    most likely area

    to

    search

    for the

    inspiration

    of

    theGerman letters, ifnothing makesuslook elsewhere.We may finallyguess

    that

    the

    capital letters

    of the

    Latin alphabet

    in its

    entirety

    is

    what

    we

    should

    use as a sourcecandidate." Cf. Quak (1996: 172): "Wenn man die Fundlage

    und

    die Chronologie

    der

    Funde sieht, kommt

    von der

    Logik

    der

    Zeit

    und der

    Geographie her eigentlich nur das lateinische Alphabet als Ursprung in Be-

    tracht."

    9

    Such

    ase

    derived from

    M andp

    from

    K.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    8/47

    The fJDark (and ogam) 171

    potential source alphabet because of a preference for the ideas of

    diffusion

    and

    evolution,

    and

    hence theprobability of geographical

    contiguity

    and

    chronological continuity

    in the

    process

    of

    transfer.

    It

    could

    be

    that

    the

    linguistic substance such

    as

    sign shape

    and se-

    quencepoints in adirection that does not fit any such predilection.

    If

    so,

    then the linguistic evidence mustbe

    given

    the

    benefit

    of the

    doubt

    and

    received opinion

    on the cultural

    circumstances

    may

    have

    to bequestioned.

    2.

    Acceptance of

    realities

    It has to be

    accepted

    as a

    reality that

    no

    one alphabet

    on its own

    can

    account

    for all the signs of the fu|>afk.

    This implies that even

    if

    one

    alphabet

    is

    seen

    to

    correspond

    in the

    main

    to the fujmrk

    inventory on the basis of substantial

    characteristics

    -

    including

    the

    sequence

    of

    signs

    - it has to beaccepted

    that some signs

    will

    have been

    intentionally

    adapted

    or

    even borrowed

    from elese-

    where.

    3. A consistent

    system

    Admission

    of the

    possibility

    of intentional

    adaptation

    or

    borrowing

    must not degenerateinto"an easy way outwhenall elsefails".Any

    divergencein

    shape,

    inthe

    correspondence between shape

    and

    sound

    value

    or in the sequence of signs as

    compared

    with the source al-

    phabet towhichthemajorityof runes correspond m ust not be waved

    awaywith

    afanciful

    explanation

    or

    accepted

    as an

    arbitrary altera-

    tion.

    It

    mustbedemonstrated

    to be

    either consistent with

    the

    con-

    ventions

    displayed

    by

    other instances

    of

    alphabet transfer

    from one

    languagetoanother,or atleastnot theresultof anarbitrary, one-off

    intervention.

    As far as

    sign sequence

    in

    particular

    isconcerned,an

    extension

    of the

    above principleofsystemand conventionis toassume that the fuj^ark

    orderwas nolessindependentof theoriginalalphabetical order than

    the

    ru ne shapes w ere

    independent of the

    original letter shapes.

    In

    other

    words, the

    fufcark

    order was the result of a re-arrangement of the al-

    phabetical

    order.

    There-arrangement,

    moreover,

    is

    unlikely

    to

    have

    been arbitrary ,

    which

    conversely implies that

    it was

    deliberate

    and

    pro-

    bably

    methodical.Arequirementtoexplainhow an alphabetical order

    was

    methodically converted into the fu]>ark order alsoadds rigour to

    the

    test

    ofacceptabilityof any

    thesis

    on theoriginof the

    runes.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    9/47

    172

    Alan Griffiths

    2.4 The relevance of ogam

    In the course of our argument we shall be involving the script known

    as ogam or ogham that is

    found

    on monuments in Ireland

    from

    the

    5th century

    AD on

    (see Appendix

    1).

    Among

    the

    reasons

    for this in-

    volvement

    is the

    fact that ogam

    and the

    runes have

    at

    least

    six

    features

    incommon:

    10

    1.They

    are

    both

    first

    documented

    in

    areas just across

    the

    northern

    frontiersof the

    Rom an Empire:

    2nd to 3nd

    century

    AD inDenmark/

    S.

    Norway

    for the

    runes;

    5th

    century

    AD in

    Ireland

    for

    ogam.

    2.

    Both systems assign meaningful indigenous names

    to

    their

    signs,

    whereas

    Greek adoptsandadapts theSemitic namesandLatin uses

    aphonetic

    scheme

    (a, be,ce,el,em,

    etc.) which

    it

    apparently inherited

    from

    Etruscan.

    3.Both systemsorder their signs

    differently from any of the

    Medi-

    terranean alphabets.

    4.

    Both systems divide their series into groups or "families", al-

    though there are

    four

    groups (known as

    aicmi,

    singular

    aicme)

    in

    ogam compared tothree (knownascettir,singularcett) in the fu^ark.

    5.There is aspecific runic signfor

    ng,

    not found in Mediterranean

    alphabets, although Greek practice

    was to

    account

    for the

    nasali-

    zation by writing as well as , and . Traditionally there

    wasalso

    an

    ogam sign

    for

    ng,although McManus (1986; 1991) claims

    that this sign may originally have represented [g

    w

    ].

    10

    Features 2 and 4 are cited by Marstrander (1928) and features2 to 5 are cited

    by

    Pedersen

    (1920-24), as indicating Celtic influence on the formation of the

    fuj)ark. McManus (1991: 23), also citesfive other "common features, givenby

    among others Arntz(1935):1. Both systems appear

    to be

    designed

    for

    epigraphic

    useon hard materials (wood orstone),their latter-day use in manuscrip ts being

    a mere antiqu arian curiosity. 2. Both scripts had particular magical associations.

    3.Vocalic[u] and consonantal [w] are distinguished in both fuf>ark and ogam.

    4. The

    fact

    that ogam originally

    had no

    sign

    for

    [p]

    might

    be

    compared

    with

    the

    limited use ofrunicp. 5.Runic

    h

    was notmerely graphicand

    z

    was notrestricted

    to foreign words, as in Latin. As McManus points out, however, the first two

    ofthese

    do not

    distinguish ogam

    and

    fu^ark from other alphabets, since Roman

    capitals also

    had a

    specificallyepigraphical function, while

    the

    magical properties

    of

    the writtenwordareunversal (and often overstated in thecase ofogamand

    runes).

    The

    other

    three features are related to Arntz's desire to demonstrate

    that ogam wasderived from the fufcark rather than the Latin alphabet.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    10/47

    The

    fujsark

    (and ogam)

    173

    6.

    The

    ogamic system

    of

    strokes across

    or

    just touching

    a

    running

    line is highly reminiscent of later

    cryptic

    runes based on asystem

    of

    counting strokes.

    When

    expressed in such general

    terms, none

    of these observations is

    particularly

    helpful in

    establishinganunequivocal relationshipbetween

    the twosystems.Thespread inbothtime andplace of theearliestfinds

    speaks against

    the

    idea

    of one

    scheme being derived

    from the

    other

    ratherthan infavourof it.However,there is one overriding methodo-

    logical reason

    for

    considering ogam

    beside the

    fuj)ark:

    ogam

    is a

    prime

    exampleof asystem wheretheorder of the"signs"has been intention-

    ally changed and, sincethe form of these signs isunrelated to letter

    shapes, their relation to any alphabet can be established only on the

    basisof

    sound values

    and

    their orderin

    the

    inventory.

    In

    other words,

    no

    problem

    of

    signshapemuddies

    the

    waters

    of

    comparison. Moreover,

    ifit islegitimatetocompare ogam with Mediterranean alphabets, then

    it

    is

    also legitimate

    to

    compare ogamwith

    the

    fuj)ark,

    if

    only

    to

    de-

    termine

    how the two

    treated their sources similarly

    or differently.

    By

    treating

    the

    fuj>ark

    in the

    same

    way as we

    have

    to

    treat

    the

    ogam

    inventory,namelyas aseriesof sound valuesin agiven order, rather

    thanas

    characters,

    we can

    stand back from

    the

    problem

    of

    comparing

    individual

    sign shapes andconcentrateon the system as a sequence of

    sound

    values.

    11

    As a consequence the"derivation"of sign shape ac-

    tuallybecomes easier.

    3. Correspondences

    3.1 The

    runic inventories

    We

    begin with

    the

    earliest

    forms of the fuj>ark. In the

    oldest known

    fujmrk inscription, found at Kylverand dated to the firsthalf of the

    fifthcentury AD(Fig. la),the 24

    runes-are

    presentedas anunbroken

    series. However,

    on bracteates

    found

    4 Vadstena

    (Fig.Ib)

    andGrum-

    pan

    (Vstergtland), both dated about

    fifty or

    more years later, there

    It

    is

    interesting

    to

    note

    thatin

    teaching

    the

    alphabet

    Greek and

    Roman

    tutors,

    at

    least

    up to the

    second century

    AD

    began

    by

    introducing their pupils

    to the

    names

    andorderof the

    letters before showing them

    their

    form;

    see

    Carcopino

    (J985: J2 i )citing

    Quintilian,

    I3,

    16-J7/Apparently

    Qiiintillian strongly con-

    demned thisapproach.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    11/47

    174 Alan Griffiths

    are punctuation marks at the two points in the sequence where the

    fufcark

    was

    split into three groups

    of

    "families", later known

    as

    cettir.

    Othernotable

    differences

    betweentheKylverfufrark and,forexample,

    the

    Vadstena inscription are:

    1.Kylveriswrittenleft toright,theVadstena (and

    Grumpan)

    version

    right to left.

    2.

    Kylver's

    p

    has the

    usual form, Vadstena's

    is

    more like

    a

    b.

    3.Kylver'ss zj and rj differ from Vadstena's.

    4.

    Kylver's

    /'

    comes

    after

    /?,

    Vadstena's

    before

    it.

    5.

    The

    Kylver order

    of

    d

    is

    reversed

    in

    Vadstena (assuming

    the

    finalrune is a d).

    6.

    The

    a

    and

    b

    inKylverare in mirror-image.

    Weshall revisit mostofthese

    differences

    in thecourseof ourdiscussion.

    3.2 The abecedaries

    As far as the

    order

    of

    runes

    and

    letters

    is

    concerned

    (as

    distinct from

    shape),

    we

    should

    in

    theory need only

    take

    intoaccount

    two

    basic types

    ofMediterranean alphabet, namelytheLatin and the Greek, sincethe

    north

    Italian,

    Lepontic and Etruscan versions are in essence Greek.

    The

    main

    differences

    between Etruscan

    and

    Greek alphabets

    are

    that

    Etruscan had no separate character for /o/, only used the characters

    BandDin model (learners') alphabets, included forms for both sibilants

    son

    and

    sigma

    (whereas Greek alphabets used either one or the other),

    and sometime before530 BCaddeda figure-of-eightsign for /f/ at the

    end of its alphabetseries.

    1

    ^However, because Etruscan has played such

    a

    prominent role

    in

    previous discussions

    of

    runic origins

    we

    include

    it

    along with Latin

    and

    Greek (see Table 1).

    :

    Wulfila's Gothic alphabet

    is also included since we shall be referring to the way this alphabet

    treated its Greek

    model.

    13

    (The fu|>ark and ogam values have been

    included and willbeelucidated later.)

    12

    In ignoring the Etruscan alphabet 1 concur w ith Morris (1988: 6).

    13

    The

    version

    of the

    Gothic alphabet given here

    isfrom theCodex

    ArgenteusHand

    1, (as reproduced in Philippa and Quak, 1994: 23). The Etruscan alphabetis

    reproduced from Bo nfante (1990) (butseealso

    Caffarello,

    1975: 71) and follows

    the

    form

    taken in the fourth to first centuries BC,except for the qoppa,which

    isearlier.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    12/47

    The

    fufrark.(and

    ogam) 175

    Admittedly

    the lists give a very broad

    picture,

    since they largely ignore

    local variations as well as detailed variations overtime,but they provide

    a

    baselinefrom which

    tostart,

    without

    prejudgingthe

    issue

    as to

    date

    and

    place.

    In

    fact

    one of th

    advantages

    of

    concentrating on order

    rather

    than sign

    shapes

    is

    that alphabetical

    order changed much less

    radically than letter shapes with

    respectto

    time

    and

    place. Details

    of

    specific

    alphabets

    from specific

    dates

    m ay

    needto

    be filled inlater, but

    withregard

    to

    order

    the

    main

    differences

    between

    the

    Greek, Etruscan

    and Latin alphabets are:

    1.

    Originally Latin Chad a similar sound valueto

    that

    of the Greek

    gamma,nam ely [g],

    but as

    early

    as the

    time

    of the

    kings

    the

    Romans

    began using

    C for [kj in

    place

    of K

    (possibly through Tuscan

    or

    Sabineinfluence). Etruscan appears tohave used the third letterin

    their series for a [k]before[a].

    2. Classical Greek had no use for thedigammaas a letter but retained

    it

    as a numeral in the sixth position; the original pronunciation of

    the digamma was probably

    like that

    of

    [w], which

    was

    also

    the

    sound denotedby theEtruscan letter (usually transcribed as

    V):

    To

    represent

    the sound

    [fj

    (not present in Greek), the Etruscans ini-

    tiallyused

    a

    combination

    of the

    digamma

    and H, i.

    e.FH,

    but

    tran-

    cribedas

    VH.

    Sometime before530 BCthey replaced thisby a figure-

    of-eight sign which they listed at the

    end

    of their inventory. The

    Rom ans began

    by

    copying

    the

    Etruscan

    use ofFH for [fj but

    even-

    tually

    dropped the

    H.

    For [w]they made use

    of

    the vowelV(see

    point

    7).

    Roman

    grammarians,

    however,

    "misledby

    shape, ascribed

    to F the

    power

    of the digamma;

    thus:

    F Aeolicum

    digamma,

    quod

    apud

    antiquissimos

    Latinorumeandemvimquamapud Aeolishabuit,

    eumautemprope sonum

    quernnunchabet,

    significabat

    p cum

    aspira-

    tione;

    sicut

    etiamapud veteres

    Graecospro

    F,p,et h,

    Prise.

    1,4,

    p. 12"

    (Lewis,s.v. F).

    In

    fact

    Latin

    F

    never represented the sound

    of di-

    gamma

    but a value

    approximating

    1

    '"that

    ofGreek

    ,

    but rougher,

    Quint.1,4,14;

    12,10,29;cf. Prise. 1,4,p.14"(ibid- seealso point7

    below). v

    3.

    Greekhad a Z

    where

    Latinhad G; theLatin Gwassupposed to

    have

    been introduced

    by a freedmaa of the

    consul

    Spurius Carvi-

    lius Ruga sometime after 250

    B.C.;

    before then Latin had a form

    of the Greek Z. The

    sound

    values of the Greek sibilants, inclu-

    ding

    zeta,

    are not always clear and varied from dialect to dia~

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    13/47

    176 Alan

    Griffiths

    lect

    14

    Beforethe

    re-introduction

    of Z to

    transcribe Greek

    names,

    in

    Cicero's

    time, Latin represented

    zeta

    as

    S

    initially

    (as in

    Sethos,

    sona,

    etc.) and asSS medially (asAtticisso,badissas,etc.); seealso

    point6.

    4.Latin had no equivalent for Greek and a sign with asound

    value

    equivalent

    to , i.e. X, was

    placed towards

    the end of the

    Latin

    series,

    not

    after (see point

    8).

    5.Classical Greek

    had no use for as a

    letter

    but

    retained

    it as a

    numeral

    6.

    Where Latin

    had an

    S

    between

    R

    and

    ,

    some versions

    of the

    Greek alphabet had ,whileothers,particularlyin thewesternco-

    lonies, had no but a son,

    shaped like

    an

    M, between

    and

    (rho);

    there were hardly

    any

    Greek alphabets that included both

    son

    and sigma- Argoswasone;thevalueofsonappearstohave been

    a

    voiced [z].

    The

    Etruscan alphabet included both

    sanand sigma,

    but

    there

    appears

    to

    have been

    a

    difference

    in

    their usage between

    the

    dialects

    of

    north

    and

    south.

    15

    7.

    Latin

    V

    doubled

    as

    vowel

    /u/ and

    semivowel /w/. There

    was an

    attempt

    by the

    emperor Claudius

    to

    introduce

    a

    separate sign

    for

    the

    Latin semivowel, namely

    an

    inverted

    and

    reversed

    F, but

    this

    was

    never

    widely

    adopted. This Roman modification

    of a form of

    the

    digamma

    for

    /w/,however,

    reflects

    their continued awareness

    of

    the

    value

    of the

    Greek

    and

    Etruscan letter.AfterupsilonGreek placed

    its

    aspirated

    ,i.e.

    [ph], originally represented

    in

    Latin

    byP(H) or

    B\

    by the

    second century

    BC the

    value

    of was

    probably that

    of

    a bilabial spirant;notuntilc. 400 ADwere Latin

    PH

    (for) and

    F (alabiodental) treated asidentical. Later Etruscan had aphifor

    [ph]and a figure-of-eight

    sign

    at the end of its

    inventory

    for [f].

    14

    InJeffrey

    (1990, Plate

    20:16)

    there

    is

    even

    an

    example

    of a

    Corinthian alphabet

    inwhich

    xi

    isplaced in the

    san

    position afterpiand

    san

    appears in thesigma

    position

    after rho.

    15

    Etruscan had difficulty indistinguishing theshapes ofM[m] andM

    (san)

    and

    at one

    point attempted

    to

    solve this

    by

    using

    a

    half-obliterated M,rather like

    a

    Greek

    in

    form. This letter

    shapewas

    revived

    by

    Verrius

    Flaccus

    (4th century

    BC) to represent the hardly audible [m]intially before Latin vowels.The date

    of this"revival"excludesanydirect relation to therunesbut theprocess

    illu-

    stratesa

    tradition that

    may

    have some relevance

    to the

    formation

    of

    runic

    m

    and w(see Section 7.2, point3).

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    14/47

    Thefufrark

    (and

    ogam) 177

    8.

    Latin

    X

    Was

    added

    to

    its

    original inventory

    to

    account

    for

    [ks]

    in Greek words.It isphonetically equivalenttoclassical Greek

    and some western Greek valuesof thecharacterXbutnot toclassical

    GreekX [kh],

    9. Latin and

    were

    similarly added at the end of the series to

    take care of Greek sounds.

    10.

    Greek andwere also later

    additions;

    aform similar to in

    westernGreekandEtruscan alphabets

    -

    asign likeadownward-poin-

    ting

    arrow-

    had a sound value similar to that of classical Greek X

    [kh].

    To sum up, only two alphabets need be considered if we take the

    order

    of

    characters

    as a

    prime criterion

    in

    m atching letters

    and

    runes.

    These

    are the

    Greek

    and

    Latin alphabets. Admittedly,

    the

    Latin alpha-

    bet was based on the Greek, most probably via an Etruscan model,

    but itshows distinctive featuresin itsomissionof theGreek characters

    theta

    and

    xi,

    as

    well

    as its

    redistribution

    of the

    phonemes

    /k/

    and

    /g/

    andits use of the

    digammafor

    /f/ and of

    V

    for

    both

    /u/

    and

    /w/.

    The

    order

    in

    other Italian alphabets

    is

    hardly distinguishablefrom that

    in

    theGreek, EtruscanorLatin.

    33 Thematch

    The firststepwetake in matching the fuj>ark runesto the Latinor

    Greek alphabet isfairly radical: weremovethe vowels. There islittle

    a

    priori

    justification

    for this other than to observe that the Irish ogam

    treated

    its

    vowels separately from

    its

    consonants

    and

    that there

    are a

    number

    of

    later

    fujrark

    inscriptions

    from

    which

    itcan

    be deduced that

    the runic consonants were sometimes learnt separately

    from the vo-

    wels.

    16

    We

    might also note

    in

    this connexion

    the

    example

    of a

    third-

    century BCVenetic writing tablet,now inEste,onwhichtwoconsonant-

    only series of letters are inscribed, presumably for learningpurposes,

    one

    of the series being in

    purely

    alphabetical

    order.

    17

    Principally, ho-

    16

    Knirk

    (1992:

    192-7).

    17

    The

    tablet

    (3rd

    century

    BC,presentlyin the

    Musep

    NazionaleAtestino,Este,

    Italy)

    gives

    a

    consonantal

    syllabary round

    the

    edges

    and on the

    bottom row,

    reading right to

    left,

    the

    series

    of

    consonants

    inVenetic

    alphabet order:

    (v) (z) (h)

    (q)

    k lmnps' (?)

    thefirst

    four

    charactersand the

    last

    one not

    beinglegible.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    15/47

    178 Alan Griffiths

    weverJustification

    fortreatingtheconsonants separatelyis to befound

    posteriori, in the end

    result, i.e.

    the

    almost

    perfect

    match between

    runic

    and

    alphabet consonant orders.

    Having

    removed the vowels, we now write out the consonants of

    theearlier and later Latin alphabets and the archaic and classical Greek

    alphabetsin pairs (Fig. 2a to 2e). Comparison between the later Latin

    and classical Greek series (Fig. 2a and 2c) is disturbed by the omissions

    from these

    two

    alphabets: theta,xi

    and

    from

    the

    laterLatin;digamma,

    qoppa and eta (a vowel)from the classical

    Greek.

    As is only to be

    expected,

    the correspondence between the earlier Latin series and the

    archaic Greek series using

    sigma

    and etaas a consonant (Fig. 2b and

    2c)iscloser,but thisis not thepointof theexercise. W hatisremarkable

    is

    that when

    the

    fuj)ark consonants

    are

    written

    out in

    boustrophedon

    (Fig.

    2f),

    there is aparticularly close correspondence between pairs of

    runicconsonantsandpairsofalphabet consonants in thearchaic Greek

    serieswhich includes

    san

    rather than sigma (Fig. 2e).

    In

    fact none

    of

    the other

    four

    forms of the Greek or Latin alphabet compares with

    the runes so closely.

    The correspondence can be seen even more clearly if the rune pairs

    are

    arranged

    in the

    same order

    as the

    Greek alphabet (Fig. 2g).

    It is

    not exact: apart from obvious

    difficulties

    in reconciling some of the

    signshapes, the association of certain runes with letters having totally

    different sound values alsopresentsproblems.

    For

    instance

    the

    Greek

    digamma represented

    [w] and not

    [f],

    and xi did not

    represent

    [j]. And

    what about the value [] for the rune positioned in association

    with

    kappa, and [k] for the rune associated with archaic qoppal

    The discrepancies between the two sets of characters are of three

    types:

    1.

    Order,although certain pairsofrunic consonantscan be

    identified

    withpairs of

    alphabet consonants,

    the

    runicpairs

    have

    obviously

    been re-arranged. Is this re-arrangement as arbitrary as it at first

    appears

    or can we

    discern some system

    in it?

    2.

    Sound values: how do weaccount for the values assigned to the

    runes/, w,;;and kaswellas 7,sand z?

    3.Signshapes: how do

    we

    explain the

    shapes

    of

    runicg,

    k, d,p,#,

    7,z,p and w, to name but the most obvious?

    We

    deal witheach

    of

    these

    in

    tu rn, which means

    that

    while

    we

    tackle

    the question of order and systematic re-ordering, for example, we must

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    16/47

    The f park-(and

    ogam)

    179

    assume

    that the problems of sound value and sign shape are resolvable.

    We

    deal with

    the

    question

    of

    order

    and

    system

    first

    because

    this

    plays

    a

    role

    in our

    resolution

    of theothertw o

    problems.

    Thus

    it is the

    order

    ofsound valuesinboth Wulfila'sGothicalphabet and theIrish ogam

    when

    compared with their source alphabets that helps to corroborate

    the posited association of Greek letters with the runes mentioned in

    point 2 above. Finally it is also order that may provide the

    key

    to a

    system

    in the shaping of certain runes tha t are not imm ediately recog-

    nisable

    in the Mediterranean

    models.

    4.

    Order

    and re-ordering

    4.1

    Thefupark consonants

    Numerous

    different

    ways

    of

    arriving

    at the

    same result

    are

    conceiv-

    able. Without documentary evidence wehaveno way oftelling exactly

    whatmethod was used. The two suggestions described below are cer-

    tainlynot the

    only possibilities. W hat

    is

    important

    is to

    recognise that:

    1. the

    starting-point

    in

    each case

    is an

    arrangement

    of

    eighteen Greek

    consonants

    in

    alphabetical order

    and

    divided into three groups-

    two setsofeight(B to K and to P) plusthe last

    pair

    (T and ):

    A M N S I I S ' Q P

    2.

    the subsequent recombination of these sets of consonants is trans-

    parentlysystematic evenif thereason for it is not immediatelyclear.

    Eachof the twomethodsisbasedon adistinctive principle.The first

    uses

    a

    system

    of

    columns

    and rows, and is

    suggested

    by the

    fact

    that

    the fujsark itself maintains pairs of consonants recognisable from a

    Greek alphabet as shown above (Fig.

    2f).

    The second method is based

    onthe knownpracticeof inscribing a learning alphabet round the edge

    ofa

    writing

    tablet

    (cf.

    the Este

    tablet

    mentioned inNote 20, and the

    well-knownMarsillianad'Albegriatablet).

    18

    In both cases Greek

    kappa

    isassociated

    with

    runic TJ and qoppa with

    runic

    k,which for the time

    being

    mayseem strangebut willbeexplained later (Section 53).

    18

    See,

    for

    example, Caffareilo (1975:56-58),

    and

    Diringer (1968; Fig.22.2a).

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    17/47

    180

    Alan Griffiths

    Suggestion

    1

    based

    on

    columns

    androws

    The first set ofeight consonants, from B to K, isarranged in two

    rows

    reading

    from

    left

    to

    right, then right

    to left, as

    shown

    in

    Fig.

    3a.

    The

    second

    set of

    eight,

    from

    to P, is

    arranged

    in two

    columns,

    first

    top tobottom, then bottomtotop,asshowninFig.3b.These columns

    aresplit inhalf horizontally, and the last pair ofconsonants, and

    , is

    inserted

    in the

    split (Fig. 3c).

    The

    rowsfrom Fig.

    3a are

    then sys-

    tematically

    spliced pairwise into

    the

    columns,

    the first two

    pairs

    of

    each

    rowoneither sideof and , and thesecond twopairsat the top of

    the columns, asshownin Fig.3d. If we now read down one column

    and up the

    other(Fig.

    3e) and

    subsequently write

    the

    runes

    out in

    this

    order

    but in

    three rows

    in

    boustrophedon

    (Fig.

    3f),

    we

    have

    the

    fujrark

    order.

    19

    , .

    Suggestion 2 - usingawriting tablet

    Here the first eight consonants are written out as in Fig.4a. The

    second

    set of

    eight consonants

    is

    inscribed round

    the

    three sides

    of a

    writing

    tablet

    (Fig. 4b),

    the

    fourth, open side being

    the

    side with

    the

    handle.

    The

    four consonants,

    to K , are

    then inserted

    as in

    Fig.

    4c

    and the

    four from

    to F as in

    Fig.

    4d,

    while

    the

    pair,

    and , are

    set at the

    corners (Fig. 4e). Finally

    the

    whole sequence

    fromF to is

    read in

    boustrophedon

    as in theprevious suggestion inFig.3f.

    4.2 Thefupark vowels

    Theplacingof thevowels amongst theconsonants is almost sym-

    metrical

    but

    unfortunately

    not

    quite.

    A

    particular problem

    is

    created

    by

    thealternative positions of

    and

    L

    InFigs

    5a-5b

    theKylverand

    Vadstenafu{)arks(Figs

    la-lb)

    arewritten

    in

    boustrophedon.Thecom-

    monest order,however, combines

    the

    o/upositions

    of

    Kylver

    and the

    /// 'positionsofVadstena (Fig.5c).

    20

    Thesituation iscomplex. In the

    19

    Significantly

    this involves dividing

    the

    series into three sets

    of

    eight, thus

    fore-

    shadowing

    the

    three

    families, or

    cettir,

    of the

    later fufcark inscriptions such

    as

    that on theVadstena bracteate.

    20

    The

    vowels

    in the

    Anglo-Saxon Runic Poem, which lists

    the

    rune names,

    are

    arrangedin theVadstena order.TheCodexSalisburgensis

    140

    givesAnglo-Saxon

    futhorcs,

    one

    with

    theVadstenaorder, theother with the commonest order.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    18/47

    The fut> rk

    (and

    ogam) 181

    "commonest"order (Fig,

    5c)

    the

    arrangement

    of u o

    /and

    is in

    fact

    virtuallysymmetrical,with andi'asthe finalrunesin thebottom line

    and

    u

    and

    i

    the

    penultimate

    runes

    in the top

    line.

    It

    would appearthat

    the

    vowels

    in

    both

    the

    alphabet

    and the fufcark

    adjoin

    similar consonants.

    To see

    this

    we

    need

    to

    read

    the

    consonants

    of the three

    cettir

    of the

    fu{>ark

    in

    boustrophedon

    from the bottom up:

    fprkgw

    > :

    h

    nj

    p

    ^ 5

    .

    . < . " ' "

    '

    .

    t

    bmlg d

    whichwritten

    in

    line read:

    > >

    . '

    >

    d linb t h n j p z s

    w

    g k r p-f.

    Taking

    the

    most commonlyfound positions

    for

    runic

    and

    Tin

    the

    fujwk,we now find for

    the

    Greek alphabet and the

    fuj>ark:

    In the

    alphabet:

    A

    precedes

    precedes

    precedes

    F all

    three consonants

    In the fujsark: e precedesb precedes/? precedes/ are

    labials.

    Inthe

    alphabet:

    follows follows I

    follows

    Two

    consonants

    are

    In

    the

    fu^ark:

    precedes

    d

    i precedes^

    a

    precedes/? dentals

    +E/j.

    With the

    fufcark

    as

    baseline, this

    can be

    illustrated

    as

    follows:

    Alphabet

    vowels:

    dtjlm\bthnj\pzswgkrp\f .

    Fufrark vowels: e i u

    Alphabet vowels:

    I

    t

    d 1

    ylm

    bthnjj swgkrl

    p1

    f.

    FuJ)ark

    vowels:o t .

    where

    thevowels above the line

    are

    thoseadjoiningtherelevant con-

    sonants in thealphabetand the vowels below the

    line

    are thoseadjoining

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    19/47

    182

    Alan

    Griffiths

    the runicconsonants.The change of vowels from alphabet to fufcark

    is

    also coherent:

    Alphabet

    replaced

    by

    fufcark /

    or

    Alphabet

    replaced

    by

    fuj)ark

    u

    or

    o,

    thusgiving

    the

    symm etrical distribution

    of

    Fig.

    5.

    As

    regards alphabet A and I it seems that alphabet A could not

    simply remaina,whichleft only

    one

    option:

    Alphabet A replaced by fu]park e

    Alphabet

    I

    replaced

    by

    fufcark

    a.

    The omissionof C//Y from thealphabet set ofvowelsinthis scheme

    is

    consistent with

    the

    ideathat Greek

    tallies with runic w,

    on a par

    with the

    Gothic solution.

    (In the

    ogam setup

    the U

    also appears

    to

    have

    a

    separate status,

    see

    Appendix

    1.)

    A

    system along these lines wouldgosomeway toexplainthevariation

    in

    the

    positions

    of the and

    runes since

    it

    could lead

    to

    confusion

    about whether these vowels should

    be

    inserted before

    or

    after

    the

    and drunes.

    5. Order and the assignment of sound values

    5.1 New phonemes, old signs

    When

    a

    form

    of the

    alphabet

    was

    applied

    to a

    "new

    59

    language

    and

    new

    phonemes had to be

    accounted for,

    it was not usual to invent

    totally

    new

    signs. Mostly,

    the new

    phonemes were assigned

    to

    existing

    signs which represented phonemes that did not occur in the

    "new"

    language, or extra signs were borrowed

    from

    other alphabets. Borrowed

    signs

    were usually placed at the end of the chosen abecedary (as Et-

    ruscan 8 =

    [f]),

    whilean existing sign usually remained in its original

    position

    in the

    abecedary

    (as

    LatinFwithout

    the

    H).

    A

    third possibility

    was

    for an existing sign to be slightly modified and then either be kept

    next

    to the

    original sign

    (as

    /beside

    /

    or Wbeside V

    in

    m odern

    al-

    phabets) or be slotted into a position previously occupied by a sign

    thathad become redundant (as theLatin Gderived from Cin thezeta

    position).Occasionally ligatures wereformed

    and

    consolidated into let-

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    20/47

    The

    fu^ark

    (and ogam) 183

    ters in

    their

    own right, aswith the Cyrillic

    representation

    of

    u], [ja],

    etc. from uncial Greek

    , ,;placedat the end of the

    alphabet along

    with

    the

    ligatures

    for

    [eng],

    [ong],

    etc.

    from

    iv,

    ov.

    21

    It is in this light that we

    iriust

    consider the discrepancies wehave

    noted above inmatchingthesound valuesofGreek lettersand those

    ofrunic/, w,j\ and k as

    well

    as thesibilantssand z.

    Comparison

    with

    the way theGothic alphabet and the

    Irish

    ogam inventory dealt

    withthesame sound values ishelpful here.

    5.2 Runicf , w andj: comparison with the

    Gothic

    alphabet

    Comparison with

    the

    Gothic alphabet indicates

    a

    similar approach

    between

    Wulfila's

    solution inadapting the Greek alphabet to a Ger-

    manic language

    and

    that

    adopted

    earlier

    in the

    creation

    of the fuj}ark.

    Table

    1

    lists

    the

    fujrark beside

    the

    Gothic,

    Greek,

    Latin

    and

    Etruscan

    alphabets, based on theassumedcorrespondences

    between

    the fuj)ark

    andGreek. It willbeseen that theGothic list introduces adaptations

    to theGreek consonants at twoplaces wherethe

    fu^ark

    also deviates

    from

    the

    Greek

    alphabet, namely

    at the

    archaic

    digamma

    and the

    xi

    positions.

    The

    digammaposition

    is not left

    vacant but

    is filled

    with

    a

    character

    for

    /kw/, while

    thexi

    position

    is filled

    with

    a

    sign, seemingly

    derived

    from

    a

    runic source,

    for /J A

    Theplacing of

    /j/

    in the xi.

    position

    in the Gothic alphabet corro-

    borates our assignment of

    they

    rune

    to

    the

    Greek\-xi

    position in the

    model for the fujmrk: thedeviser(s)of

    both

    the fuj?ark and the

    Gothic

    alphabet treated the

    /j /

    phoneme in the same

    way.

    They apparently

    needed

    to

    represent

    /j /

    by

    separate sign rather than

    follow

    the

    Latin

    practiceofusingasingle sign,/, for both [i]and

    [j].

    There beingno

    exact equivalent

    in

    Greek,

    the

    Germanic sign

    was

    given

    the

    place

    of

    xi,for

    which there

    was no

    use.

    Superficially,

    this seems

    to be an

    example

    ofa.straightforward

    re-occupation

    of the

    site

    of a

    redundant sign.

    The treatment

    of

    the

    digamma positionin Gothic is

    curious.

    It was

    notfilled by a

    sign

    to

    represent

    the/w /

    phoneme, which

    had

    disappeared

    from

    classical

    Greek,

    Instead,

    a

    q

    *=

    [kw]

    was

    introduced,

    while

    the

    21

    Cyrillic

    also

    shows an interestingmodification torepresent[j]

    ([tj],

    [djj) from

    (placed

    beside

    [i])aswellas the

    [zh],

    [dz]

    pair

    from

    ta

    and

    (beside[z]).The

    origin

    of the

    sibilants

    [sh],

    [sht], [ts]

    and

    [c],which follow[o:]*

    is too

    controversial

    for

    commenthere.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    21/47

    184

    Alan Griffiths

    long

    redundant qoppa was

    only

    retained as a numeral, as in Greek.

    The /w/ wasrepresentedby an equivalentof

    upsilon,

    positioned

    after

    /,

    which

    is

    precisely

    the

    position

    we

    have presumed

    for it in the

    model

    alphabet

    for the

    fufcark.

    A striking

    difference

    between th e Gothic

    al-

    phabet

    and our

    assumptions regarding

    the

    fufcark

    is in the

    placing

    of

    /in

    the

    digamma

    position in our presumed model for the

    fufcark,

    whereas

    the Gothic series obviously treats

    the / as

    paralleling

    with respect

    to

    position (and perhaps also

    phonetically,

    since

    the

    Greek sound value

    by

    this time approximated

    a

    bilabial spirant)

    but

    resembling

    a

    runic

    (orLatin)/graphically.

    22

    5.3 Runic

    g, k and

    s,

    z:

    comparison with

    the

    Irish

    ogam

    The assignment of runic rj and k in relation to Greek kappa and

    qoppa can best be appreciated by taking into account the assignment

    of

    sound values

    in the

    Irish ogam. Here

    the

    later manuscript tradition,

    to which we owe our knowledge of such assignments, transcribes sign

    No. 9,[k],as a C(as in the Anglo-Saxon furore) and signNo. 13as

    NG,presumably with

    the

    value [q].According

    to our

    analysis

    of the

    fujsark

    so far therune representing[]appearstooccupyasite originally

    occupiedby

    kappa,

    whilethe

    k

    rune appearsin theredundant

    qoppa

    site

    (see Figs 2, 3 and 4). When we subject the Irish ogam to a similar analy-

    sis(see Appendix 1) we find that ogamC

    =

    [k] occurs in thekappa site

    while

    ogam sign No.

    13,

    transcribed as

    ng,

    occurs in the

    qoppa

    position.

    McManus (1986: 18ff.; 1988:

    157-9;

    1991: 38, 144) suggests that

    ogamicNG

    was one of the

    characters which

    a

    later revision

    of

    ogam

    aimedat "rescuing from redundancy"by

    assigning them cosmetic

    va-

    lues

    that were reflected

    in

    modified forms

    of

    their names. Thus, accor-

    ding

    to him, sign 13originally represented Primitive Irish /g

    w

    /, which

    when

    subsequently delabialized merged with the value of sign 12, /g/.

    In

    order

    to

    differentiate

    the

    two,

    the

    name

    of

    sign

    13 in the

    manuscripts

    wasemended

    from getal

    to ngetal,with an initial NGwhich is not a

    22

    The

    classical

    grammarians' association of

    Latin F

    with the Aeolian digamma

    and

    Greek F(see

    Section 3.2,point2) may even be

    reflected

    in thepositionof

    /andwin thefu^ark in

    relation

    todand/:

    runic

    d

    and/stand

    at theterminals

    of

    the'fut>ark

    and in the

    eventualdivision

    into

    three

    setsof

    eight

    (cettir) the

    w

    stands

    opposite/in the firstcett,whilethe

    t

    stands opposite

    d

    in the

    third

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    22/47

    Thefu{>ark (and ogam) 185

    radical initial in Irish and cannot therefore be authentic. (Other signs

    involvedin

    this revision were those

    for

    H, Q,

    and Z; see

    Appendix

    1.)

    What

    is also of interest in this connexion is that the glosses on the

    kenningsfor

    NG

    (GG)

    make

    specific

    reference

    to a

    relation between

    K

    andNG.

    23

    This could simplyrefer to a semantic relation between the

    names chosen

    for

    these characters (cath,

    "battle", and

    getal,"act

    of

    wounding"?)

    or to an

    awareness

    of the

    voiceless/voiced relation

    ofk/g,

    but if so, whyshouldthepointbemadein theglossesonNG/GG rather

    than in thoseon Cor (7?

    The

    assignment

    of

    ogam

    NG,

    alias/g

    w

    /,

    to theqoppa

    site

    and of /k/

    to the kappa siteaswellas the Irish glosseson the relation ofK and

    NG,

    go

    some

    way to

    support

    the

    idea that runic

    and

    k

    were also

    seenasrelated andthat they were originallyassigned to thesame sites

    as

    their ogamcounterparts

    (g/qoppa, and

    k/kappa)

    but

    that

    for

    some

    reason the fujrark characters were subsequently interchanged. Sucha

    conclusion would tally with an interpretation of the shapes of the k

    and (7

    runes

    as

    having been adaptedfrom

    kappa and

    qoppa,

    respectively,

    by

    removal of the upright stave or stroke in each

    case,

    to create a system

    in

    which

    all

    palatal

    and

    velar signs were staveless (see

    Section

    7.2,

    point4).

    As forsand z, it isalso evidentfrom

    the

    matchingofogam sound

    values to

    those

    of the

    supposed source alphabet that,

    as in the

    fuj>ark,

    /s/

    tallied with

    Greek

    zetaa n d / z / with Greek

    san

    (Appendix1, Figs

    1-2and1-3).Moreover, theparallel between

    san,

    runicz andogam

    Z isagain supported by

    McM anus'(1991:

    38)suggestion thatthe ma-

    nuscript transcription

    ofogamZ is a cosmetic one and is"clearlyde-

    23

    There

    are

    threeseriesof kennings, belongingto the OldIrish period Briatharo-

    gaim("Word-ogams"),which provide informationon themeaningofIrish letter

    nomenclature.For theseketmiiigsandtheglosses onthem,

    together

    withacom-

    mantary

    on

    each

    letter,see in

    particular

    McMatius

    (1988).

    The

    relevant

    glosses

    onNG/GG,

    with

    McManus*

    translations,

    are:"getalsen .i. ar is

    luth

    lasna

    leigib

    7

    co ibnius etir

    cath

    7getal":i. e.thatis[akenningfor]

    Getal

    \.e.foritissustance

    with

    the

    leeches

    and

    thereiskinshipbetween battle

    andgetal(

    4

    act

    of

    wounding'?).

    "Ngedal

    .i.

    gilcach

    nuo

    rait,

    ut

    dicitpr

    luth lego ngedal

    .i

    gilcach

    norait,ar is

    luth laisna legaib7coibnes iter K 7 NG, no miodach ice a. ar imad a iceno/

    A.

    gilcach no

    rait":

    Ngedal Le.

    reed

    orbog-myrtle,

    ut

    dicitur 'sustenance of a

    leech'

    is

    ngedal i.e. reedor

    bog-myrtle,for it is

    sustenance with

    theleechesand

    there

    is

    kinsh ip between

    kand

    ng

    or a

    healing physician

    i.e.on

    account

    of the

    abundance

    of its

    healing (powers 7)or/i.e. reed

    or

    bog-myrtle.

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    23/47

    186

    Alan Griffiths

    signed toeffect an

    independent status

    for

    what

    was by

    then

    a

    redundant

    symbol",

    itsvalue having

    merged

    with

    that already represented

    by5,

    /s/.The initialof theIrish manuscript nameof Z,

    straif,

    indicating "an

    intermediaterealization

    of

    PCI.*st->

    Olr.

    s-,

    probably[st]"(Schrijver,

    1997:

    310, n.

    32),

    in

    fact

    parallels

    the

    presumed value

    of

    runic

    z,

    which

    was

    probablyhalf-waybetween modern-English

    rand

    modern English

    z"(Elliott,1980: 16).

    6.A phoneticpattern?

    From this match

    of

    sound values

    it

    would appear

    that

    the

    model

    for

    the

    fujwk consonants

    was the set of

    eighteen Greek consonants

    that included

    digamma,

    consonantal

    eta,

    qoppa

    andson

    instead

    ofsigma:

    8'(2

    The Germanic phonemes /j/ and //were assigned to the sites of xi

    and qoppa, /s/and

    /z /

    tozetaand

    son,

    /{/and /w/ to

    digamma

    and

    phi:

    bgdfshpkl

    mnjp

    r

    tw

    Inthis"pre-fu]3ark"seriesofeighteen consonantsthe /k/ and/ng/ were

    interchanged:

    bg df

    shpylmnjpzkrtw

    and

    from this sequence

    the

    fufcark order

    was

    derived along

    the

    lines

    suggested

    in

    Figs.

    3 and 4.

    Therationale behindtheswitchofkand;;in the"pre-fufcark" series

    is

    not

    immediately obvious

    but an

    analysis

    of the

    sound values according,

    to a Greek division of consonants into ,i. e.mutes

    (voiced,

    voicelessand

    aspirated),

    and,i.e.

    semivowels

    and

    spirants,

    re-

    veals

    a symmetrical phonetic pattern (Fig.6).

    The first sixconsonants of thepostulated"pre-fujsark" series divide

    into agroup ofthree

    ,

    b g

    d,

    followedbythree

    ,/

    s /?,

    whilethelastsixconsonants comprise alternately

    arid,

    p

    z k r

    Mv,

    with

    the

    three mutes

    p k t in

    the

    same

    phonetical

    sequence

    as the first

    three mutes

    b g d

    (Fig.6a).

    The

    middle

    six

    consonants

    of

    the "pre-fu{)ark",p l m nj are all,withthe firsttwo,p

    /;,

    and last two, n j, phonetically related to thesecond andpenultimate

    mutes

    of the

    whole series, g

    and

    t(Fig.6b).Finally, the trio

    /

    s h

    is

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    24/47

    Thefuf>ark (and

    ogam) 187

    phonetically related to the

    tri

    p z

    k,while

    m may be seen to have a

    nasal-mutation relation

    to

    b

    comparable

    to that of togand

    of n

    to

    t familiar inCeltic sound-systems

    (Fig.

    6c).

    The

    non-nasal

    semivowels

    and

    liquids

    ], [r] and

    [wj,

    [1]

    can

    also

    be

    seen

    as

    forming phonetic pairs,

    and in a

    Celtic/Germanic context

    the

    linking of the s rune to dand p and of z toy and r isconceivablein

    termsof (a) theCelticuse of a

    barred

    >,

    DD,SSoi

    Greek torepresent

    a group

    of

    related sounds also rendered by.SS (Piggott, 1974:37-8),

    and (b) the later development in Old Norse of the

    Germanic

    [z]into

    a

    strongly palatized

    [r]

    represented

    by the

    rune

    and

    usually transcribed

    as

    R.

    The

    pattern

    would not have been symmetrical if k and TJ had not

    been

    interchangedor if asibilanthadbeen retainedin the

    sigma

    position

    between

    rand t instead of the z in the san

    position.

    In

    view

    of this

    pattern

    it is

    conceivable that

    the

    alphabet which

    formed the

    basis

    for

    the runes underwent a degree of phoneticrationalization,possiblywith

    Celticmediation, before being re-ordered into the fufcark sequence.

    7.Derivation of sign shape

    7.1

    Order and tradition

    The ordering of phonemes in the

    different

    abecedaries tends to deviate

    less perceptibly

    than the

    sign shapes representing

    the

    sounds.

    If we

    play

    safe and

    compare runes withindividualGreek

    or

    Latin letters

    on

    thebasisofshapeaswellassound valuesand identifycharacters which

    have

    both equivalent shapesandeqivalent sound values, thereis agrea-

    ter correspondence between runes andLatin letters (B, C

    9

    F,H,

    /,R,

    S, T) than there isbetween

    runes

    and either classical (eastern) Greek

    letters

    (, , , ) or

    western Greek letters

    (B,

    H,

    I, R, S, T)

    which

    formed the starting-point for the Etruscan and Latin alphabets. It is

    the

    greater

    tally of correspondenceswith

    Latin

    that has been seen to

    favour the possibility of

    a

    Latin source for the

    runes.

    Above

    all,

    the

    agreement with Latin C,J%Hand fl has played an important role in

    the

    arguments, despite

    the

    western Greek

    usageof a

    tailed

    R and an

    aspirated

    [h]

    that

    was

    absent

    in the

    dialectsrepresented

    by the

    classical

    alphabet. However,thedifference in tiie talliesis not

    very

    great com-

    pared to the general lack of correspondence (in at least 16

    instances),

    whichfar outweighs thecorrespondences (in at most 8 instances).

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    25/47

    188 Alan

    Griffiths

    Even if we

    look beyond

    the

    major alphabet types

    to

    specific local

    alphabets it has to beconcluded thatnot allrune shapesare derivable

    from

    one

    single kn ow n abecedary w ithout recourse

    to

    special pleading.

    Consequently

    judgm ents regarding

    the

    plausibility

    of

    explanations

    for

    the

    source

    of

    this

    or

    that rune have

    to

    weigh

    the

    pros

    and

    cons

    of

    suggestions

    involving

    the

    modification

    of signs, the

    invention

    of new

    signs,borrowingsfrom

    different

    abecedariesordifferent stylesofscript,

    as

    well

    as the re-assignmentof sound values.The usual assumption is

    thatthe starting-point was a single known abecedary andthatthe mo-

    difications,

    inventions, borrowings

    and

    re-assignments

    took

    place with-

    in this basic framework.

    Any

    change

    to the

    basis

    has to be

    motivated,

    usually

    in terms of an inadequacy in the source alphabet to cope with

    specific Germanic phonemes.

    Thereisalso, however,thematter of

    tradition.

    It israre,for instance,

    fora sign to be a completely new invention. M ost signs that superficially

    appear to be inventions are infactm odifications of existing signs. More-

    over, although modifiedand borrowed signs as wellasexisting signs

    may beused for different sound values from

    those

    they denoted orig-

    inally, it is

    usual

    for

    them

    to be

    used

    for

    phonemes within

    the

    same

    class.

    Thus,

    it is

    unlikely

    for a

    sign originally representing, say,

    a

    labial

    to shift

    to a dental. This precludes such suggestions as runic

    yark (and

    ogam) 189

    list of

    signs whose shapes have evolved independently,

    any

    suggestion

    that graphemes may be

    related

    to each other, or are evert arranged

    according to a

    system,

    willseem

    specious

    indeed. However, given that

    the

    fufcark

    issuch aradical re-arrangmentof the alphabet order, and

    that

    this

    re-arrangement can be shownto havefollowed a system, it

    is

    not inconceivablethat there has

    been

    some deliberate modification

    of the

    shapes too.

    The use ofwedgesand

    upright staves with oblique

    branches

    point

    iii

    the

    direction

    of

    intentional modification.

    But did the

    "redesigning" go further than this? Can w e discern a logic that m ight

    explain

    shapes

    not

    encountered

    in

    other abecedaries

    for the

    equivalent

    phonemes?

    12 Theconsonants

    It is of

    course possible

    thatthe

    individual

    rune

    shapes

    and the fujpark

    order

    were simultaneous creations.

    But it is

    more likely that

    the

    shaping

    of

    the

    runes predates

    the

    re-arrangement into

    thefu^ark. If the

    origin

    was

    indeed Greek, as we have attempted to demonstrate, then it would

    not be unreasonable to suppose that the runes were first arranged in

    the

    order of

    theGreek

    alphabet

    and

    that the original order was one

    inwhich

    k

    and

    were

    in the

    kappa

    and

    qoppa

    positions,

    respectively,

    as

    suggested above.

    Keeping the

    consonants separate

    fromthevowelsand

    arranging them

    in

    alphabetical order

    in twolines,we

    arrive

    at

    Fig.

    7.

    Here

    we see the

    following

    pairings based

    on

    shape

    and

    sound values:

    1. In the top row ofFig.7a wehavethegraphicandphonetic pairing

    -

    of the

    dentals

    dandp and the velars

    g

    and

    k.

    The

    possible logic

    in

    the

    "design"of the shapes is demonstrated in Fig.7b.

    25

    Although

    the

    shape

    of g is recognisably Greek

    (Latin

    X

    being excluded

    in

    view

    of the

    process

    of

    re-ordering),

    the

    sound value

    of

    [kh] applied

    to

    the

    eastern Greek

    X and not to the

    western Greek letter with this

    shape, which had the va lue [ks] (whence the Latin use of the letter).

    However,

    it is

    well documented that

    the

    Celts used

    X for

    [kh],

    26

    25

    Antonsen's(1975)categorization ofnineshapesbymeansofstaffs,

    branches,

    pockets andcrooksis

    helpful

    in thisrespect

    26

    Jackson (1953);and

    Kluge

    (1975),s.v.Amt.

    Celtic

    use ofXas aGreekchi"is

    known

    from inscriptionsfrom Gaul (TIOCOBREXTIOon theColignyCalen-

    dar)

    and in

    Britain (in

    an

    inscriptionfrom South

    Shields)and it

    also

    appeared

  • 8/9/2019 Alan Griffiths-The Fu_ark (and Ogam)_ Order as a Key to Origin-Walter de Gruyter (1999)

    27/47

    190

    Alan

    Griffiths

    which

    could thereforebe the sourceof run ic use.The shape of the

    d

    would

    at first sight seem to be an invention, but the Celticuse of

    Xfor[kh]suggests also that Celtic use of a double>>for

    27

    may

    alsohave playedarolein theshapingofrunicd.The use of

    "singlet"

    forms

    for

    k

    and

    fr

    and

    "doublet"

    forms

    for

    g

    and

    d,

    and not the

    other way

    round,

    will no

    doubt have been

    a

    matter

    of choice, in-

    fluencedbya sense of symmetry, since the first three consonants,

    b,

    g and d, form a neat trio of"doublets".

    2. In the

    bottom

    row of

    Fig.

    7a the

    coupling,

    and t

    isindicated.

    The

    relation between

    the

    shapes

    and

    sound values

    of

    these runes

    is

    not so

    directly obvious

    as the

    dental

    and

    velar relations

    in the top

    row,

    but they are all alveolars and are the only runes (in the early

    formsof the

    fu>ark)

    w ith a central stave. In fact therelationbetween

    nand

    z

    in this

    respect

    could verywellbeparticularly close.Ifrunic

    n< N and runicz < M

    (sari),

    then both runes can be said to have

    developed similarly, with two side-staves becoming one centralstave,

    aslater found in the development of themrune. The motivation in

    both cases would have been

    the

    need

    for

    differentiation,

    the z

    from

    runiceand/or

    m,

    and then

    from

    runic

    A.

    It isworth noting thatin

    the eventual

    fuj)ark

    ordering the z and

    n

    runes remain in symmetrical

    placings about

    the

    centre.

    28

    3.Fig. 7c arranges the runes according to the phonetic pairings sug-

    gested

    in

    Section

    6

    (Fig.

    6c)

    after

    the

    switchofk

    and rj.If weconsider

    earlier as a variant spelling on some coins of the British

    'king'

    Tasciovanus

    (TAXCIAVfANOS]),of c. 20 BC to 10AD": Piggott (1974:

    36-7).

    If weexclude

    the

    so-called

    "primitives"

    (Crete,

    Thera,

    M