The origin of the runes

September 26, 2006

How and why did the runes come to be?

No definitive answers have yet been given to these fundamental questions though not for
lack of attempts. Perhaps more than any other aspect of runology, this area is dominated by
misconceptions brought about through supplementing the limited evidence available with
romantic ideas, ideological agendas and new age pipe dreams _ or replacing it altoghether.
I hope I can fare a little better with this attempt, but be aware that the lack of evidence
makes a certain amount of conjecture inescapable.

The elusive 'otherness' of the runes

Within the family of alphabets, the descent of each variant from an earlier one is normally
easily established; particularly so within the branch descending from archaic Greek, to
which the runes without any doubt belong. What sets the runes apart is their real or
perceived 'otherness'.
• Shapes. This is the most obvious, but yet most super_cial feature of a writing system.
Graphs have a strong tendency to adapt to changing writing implements _ cuneiform
developed from pictographical hieroglyphs because their wedge-shaped elements were
easily impressed into wet clay. The change from drawing with ink onto smooth surfaces (or
with a stylus into a layer of soft wax) to cutting into a textured surface (wood) alone goes a
long way towards explaining the di_erences in shape between the runes and possible
ancestor alphabets.
• Repertoire. The application of a script to a di_erent language with other phonemes
entirely explains that the set of graphemes di_er. This kind of development almost always
happens in the early phase of the adoption of a writing system, and does not constitute any
fundamental 'otherness' either.
• Names and order. The lack of a near one-to-one correspondence between runes and the
letters of the parent script would make it di_cult to adopt the original names and ordering.
However, it is worth noting that like their ancestral scripts the runes had both a _xed order
and names with the associated sound as the most prominent feature. The presence of these
features are not universal among writing systems, and thus represent a 'sameness' rather
than an 'otherness'.

Why?
There are two sides to this question: _rst, why did some Germanic tribes start to write at
all; second, why didn't they just use the Latin letters, which the runes are ample evidence
that they knew well.
It might seem super_uous to ask why someone adopts something as tremendously useful
as a writing system. The question arise because the early inscriptions we now know do not
contain much that could be considered very useful.
The scarcity of extant early material that is clearly meant to be read by an intended living
recipient have made room for many odd theories. Some try to explain this _ as well as other
real or imagined 'othernesses' _ in various ways:
 • Magic. There is a persistent myth that runes are not alphabetic letters representing the
sounds of language, but rather magical symbols. There is really nothing to support the
notion that the runes have ever been thought inherently more magical than alphabetic letters
by their users. Both kinds of writing have been used in ways considered 'magic' by the
superstitious, but only through its content, not merely by being writing.
• Religion. The notion that runes were used for communication with the gods or the
deceased, and not the living, does not have any explanatory power. Only if one is well
aware that writing makes it possible to communicate with a living person not present at the
moment of writing does it make any sense to try to communicate with those permanently
not present in a similar way.
• Secrecy. That the runes should be a secret way of communicating for an initiated few
seems higly doubtful, as too many signs are too similar to corresponding Latin letters.
• Prestige imitation. This is the idea that Germanic chieftains decided that they should
have a writing system not to cover a particular need, but merely because the Romans and
the Greeks did. Just like the Romans used one kind of letters and the Greeks another, they
too had to have their own kind made. This last proposition is harder to counter, but can
safely be discarded along with any other attempts to explain why there wasn't made more
runic inscriptions with communicative purposes by the following argument.

The bias of survival

The main purpose of the runes must have been to write the native language of the users,
with the intent that the message should be read and understood. It must have succeeded in
this from early on, but the vast majority of inscriptions have disappeared along with the
wood they were carved into.
It may seem frivolous to base a theory on large amounts of tracelessly disappeared
material, which in addition has properties di_ering from that which survive. However, this
is the only way to explain a range of facts:
• The set of graphemes in the older runes closely match the set of phonemes of the
language they denote. This would only be achieved if a less closely matching set of
graphemes turned out to be insu_cient to facilitate reading.
• The runes show a rapid spread across a vast area without any centralised government or
other common institutions. This would only happen if they covered an actual need.
• After an initial period of some plasticity, the system stayed almost unchanged for
centuries. This is a clear indication of widespread Reading skills.
• Runes were later used extensively for communicative purposes. If this was not true also
for the earlier period, Latin letters would almost certainly have been adopted when the need
arose.
Once it is accepted that writing on wood was commonplace, it becomes obvious that the
surviving inscriptions on more durable material should di_er strongly from what would be
expected from a more or less literate society. If the literary production of the twentieth
century was to be judged based on what was carved in stone during this period, it would not
have been done full justice _ not even the newspapers.

When and where?

The oldest known runic inscriptions date from the second half of the second century and
are found in southern Scandinavia (present day Denmark and southern Norway). While the
oldest continental _nds are only slightly younger, the scant material extant from the early
period also favours southern Scandinavia numerically.
It does also exhibit exactly the kind of graphical variation to be expected in a formative
period, strongly suggesting that the history of runic writing does not extend signi_cantly
further back in time. It is thus highly probable that the runes were 'invented' in southern
Scandinavia _ or possibly in the neighbouring areas on the continent _ in the second
century, or at earliest the _rst.

How?
The natural way for a writing system to cross a major language boundary is for bilingual
persons literate in the source language to start using the same system to write the target
language. This would be done according to the conventions used for the source language, as
the writing system would be considered the same regardless of which language it was
applied to.
When this 'naïve' approach reveals that crucial distinctions in the target language cannot
be made with the given set of graphemes, new ones are added.
New graphemes are commonly made by modi_cation of the closest existing one or, if
available, taken from other writing systems. Existing graphemes that are not needed are
rather discarded than reused with di_erent values, since the _rst users will consider the old
and the new system essentially as the same, and thus not see those letters as available for
reuse.
If a writing system was to be constructed in southern Scandinavia or the adjacent part of
the continent in the _rst or second century, Latin letters would be the most probable source
of inspiration. This is also the alphabet with the closest similarity to the runes.

Sounds
The Proto-Norse language of southern Scandinavia in the period had twentytwo
phonemes: /a/ /b/ (with allophones [b] and [¯]) /k/ /d/ (with allophones [d] and [ð]) /þ/ /e/ /f/
/g/ (with allophones [g] and [°]) /h/ /i/ /j/ /l/ /m/ /n/ (with allophones [n] and [-]) /o/ /p/ /r/ /s/
/t/ /u/ /w/ /r/. Length was phonemically signi_cant in almost all of these.
The Latin alphabet of the period consisted of twenty-three letters: A B C D E F G H I
(representing [i] and [j]) K L M N O P Q R S T V (representing [u] and [w]) X Y Z.
However, some letters were not denoting a single unique sound, and were thus not natural
to use when trying to write another language.
K and Q both represented the same sound as C, and X represented a sequence of two
sounds. Y and Z weren't used in Latin words, only in Greek names and loan-words.
Writing Proto-Norse with the remaining eighteen letters would be straightforward.
Since I and U in Latin were used both for the vocalic allophones [i], [u] and the
semivocalic [j], [w], it would be natural to use these letters for the phonemes /j/, /w/ as well
as for /i/, /u/. Similarly, D would probably be used for the phoneme /þ/, while /r/ could have
been written R, S or possibly Z.
Reading Proto-Norse written this way would have been possible, but not quite
straightforward. The ambiguity between three or four pairs of phonemes, all frequent in
in_ectional endings, would have made comprehension considerably more di_cult than what
was the case for Latin.
Shapes
In order to _nd out which runes that can be said to be an adaptation of which letters, the
limitations on the shapes due to the material must be established.
Both practical considerations and evidence from later times suggest that the preferred
medium was squared o_ sticks of wood where each line of writing spanned the entire height
of a facet. This would make any horizontal elements be hard to distinguish from the natural
grain of the wood, and the lack of such is also the most distinguishing feature of the runic
forms.
The early runes consist of graphical elements which can be divided into three groups:
staves are perpendicular lines from edge to edge of the surface; branches are oblique lines
from one stave or edge to another; twigs are short diagonal lines which need not end at an
edge or a stave. While staves are always straight in well executed inscriptions, branches
may be straight or moderately curved. Twigs are generally straight except when connected
to other twigs; then they together may form a continuous curve.
A characteristic feature of the twigs is that they tend to be rather short, and do not scale as
expected with the overall size of the runes _ being proportionally large if the runes are
cramped into a small area and proportionally small on large runes not constrained by a
limited writing area. This may indicate that they were produced in a di_erent way than
staves and branches, perhaps by simply pressing the point of the knife into the wood?
Forms tend to be oriented so that branches and twigs never are placed to the left or at the
foot of a stave unless there is also features at the right or at the top.
The later development of the runes show a tendency towards forms with at least one stave.
This tendency seems to have been at work from the very beginning, as the ratio of runes
with staves to ones without in the earliest period is 18:6, while the corresponding ratio for
Latin letters are a more balanced 13:10.

Basic correspondences
Only two graphs are common to the Latin alphabet and the early runes, the vertical line
and the symmetrical cross of two oblique lines. The former is also the simplest sign in both
systems, and represent the same sound, the wovel [i].
The Latin letter is also used for the semivocalic allophone [j], while the rune is not used
for this sound as it constitutes a distinct phoneme in Proto-Norse.
By substituting diagonal lines for horisontal ones or for curves, close correspondences can
be seen between the letters B H R T and runes with the same sound values _ the only
di_erence is that the phoneme /b/ in Proto-Norse had an allophone [¯] not found in Latin,
but which was (naturally) denoted by the same rune as [b].
When invoking the principle that twigs tend to be short, it becomes clear that the strange
small rune denoting the phoneme /k/ is a rendering of the Latin letter C denoting the same
sound. This may also explain how the letter S could give rise to the various forms of the
rune denoting /s/. These consist of a varying number of diagonal bars (with a minimum of
three) making up a zig-zag line spanning the entire line height. Probably the number of bars
was originally three _ i. e. an S with diagonal lines instead of curves _ but if this
wasn't su_cient to span the entire line due to the limited length of each twig,
the zig-zag was just continued the rest of the way.
Allowing rotation or _ipping in order to achieve the preferred orientation, the letter L
gives rise to the rune with the same sound value. Similarly, the rune denoting /u/ is in one
of its early forms just an inverted V, though a form where the left leg had been made into a
stave was equally common already in earliest period and eventually became dominant.
Two more matches can be made by assuming variant forms for the Latin letters A and E.
A not uncommon form of the letter A has a 'crossbar' which did not connect to the left 'leg'
but went parallel to this from the baseline to the center of the right leg. Interpreting the right
leg as a stave and the left leg and the crossbar as twigs, and turning the result to the
preferred orientation yields the shape of the rune denoting /a/. The rune denoting /e/ may be
based on a common form of the letter E consisting only of two disconnected vertical lines
spanning the entire line _ visually identical to the letter I written twice. In the oldest forms
of the rune, the two staves are connected by a horizontal stroke at the top, but this stroke
quickly develops into a pair of twigs.
The resulting forms for /a/ and /e/ are similar to what one would expect for runes based on
the letters F and M. However, the phonemes /a/ and /e/ were considerably more frequent
than /f/ and /m/, so it is the latter two that would be expected to develop divergent forms to
maintain the distinction. These runes are still clearly closely related to the corresponding
letters, but their forms can not be derived through the simple rules above alone.
Something similar may be the case for the runes denoting /n/ and /o/. The tilting of the
crossbar of the letter H made the corresponding rune very similar to the predicted form of a
rune derived from the letter N. They are both occuring with a medium to high frequency,
but H might have gained the 'privilege' of keeping its form unchanged through being more
frequent in an initial position even though sligtly less frequent overall. The shape predicted
for a rune based on the letter O is used for the super_uous rune denoting the allophone [-].
What this is based on is hard to tell, and its frequency is considerably lower than that of /o/.
The /o/-rune is also problematic in that its shape does not fully conform to the stave-
branch-twig decomposition as described above _ its lower part contains strokes which are
often longer than what is typical for twigs, and are not connected to an edge or a stave at
the upper end. The same discrepancy is found in many forms of the /r/-rune, where the
lower diagonal stroke does not connect to the stave, and can therefore not be interpreted as
a branch. In both of these cases, it is probably signi_cant that the 'non-conforming' twigs
end on an edge, but not at points de_ned by other features _ i. e. that there is Little need for
precision with regard to their _nal endpoints.

Sound shifts?
Some runes appear to have inherited the shape of a Latin letter, but not its sound value.
The reason for this is of course not otherwise unknown sound shifts altering the
pronounciation of some runes after their inception, but rather a combination of necessity
and coincidence.
In what was termed the 'naïve approach' above, it is assumed that some Latin letters
originally might have been used for two di_erent Proto-Norse phonemes until the necessity
of separate graphemes became apparent. In this approach, both /d/ and /þ/ are expected to
have been written with the letter D. It turns out that /þ/ is indeed written with a rune that
looks exacly as the 'transformation rules' above would predict in that case, whereas /d/ itself
is written with an entirely di_erent sign with no obvious source.
My theory is that very early in the development of the runes, /d/ and /þ/ were written as an
angular D (that is a triangle with a vertical line to the left and two slightly shorter diagonal
lines to the right). Already before the shortening of lines not connected in both ends to an
edge or a stave _ i. e. the special development of twigs _ the problem of distinguishing
between /þ/ and /d/ was solved by 'doubling' the sign when it was to be read with the latter
value. This doubling consisted in appending a mirror image of the symbol to its original
form, yielding the 'bowtie' shape of the /d/-rune. Since the combined diagonals now reached
from stave to stave (and also edge to edge), they were not shortened when their
counterparts in the /þ/-rune later were (in the later Anglo-Saxon runes, they are actually
shortened).
Doubled letters and ligatures are among the most common ways to express additional
sounds in Latin-based scripts. Obvious parallels are the digraph DD representing /ð/ in
Welsh, and the well known ligature W _ originally a digraph VV.
This theory also explains why the phoneme /g/ is written with a graph identical to Latin X,
while the form predicted to result from the letter G is used for the phoneme /j/. The
necessity of distinguishing /i/ and /j/ must have led to the semivowel being written with the
letter for the most similar consonant available, G, already from the very earliest phase. This
meant that the earliest runes probably su_ered from the same de_ciency as early Latin: the
phonemes /k/ and /g/ had to be written with the same sign. In the case of the runes, this
would have been two diagonal lines from the center of the line to either edge of it. Not only
the problem, but also its solution was the same _ adding a distinguishing feature to this sign
when it was to be read with the latter value.
In the case of the rune, this feature was again a mirror image of itself, only this time
appended to the left. The e_ect was the same as above: the diagonals of the /g/-rune now
reached from edge to edge and remained unchanged, while in the /k/-rune they had to be
considered twigs and therefore became shortened.
The alternative explanation that the /g/-rune simply re_ects the letter X is untenable.
Though it is possible (though improbable) that a letter should be reused with a di_erent
value, X would then of course be used for the problematic /j/, and /g/ would be written G.
Because of the simplicity of this shape, it is not problematic to assume that the visual
identity of the letter X and the rune denoting /g/ is coincidental. This assumption does
however imply that X (and the other letters not used for the basic correspondencies) were
not known to those who _lled in the gaps.
The last rune which apparently seems to be based on a letter with a di_erent sound value is
the one denoting /w/, whose form is that predicted from the letter P. Now /p/ is a phoneme
of Proto-Norse, but with good margin the least frequent. Its frequency is actually so low
that even though it once had a grapheme of its own, this was at an early time lost in
Scandinavia and the sound came to be written with the rune normally denoting /b/. The
shape of the /p/-rune cannot be derived from the letter P, so it must be assumed that this
letter was not known to those who invented the sign for /w/. This sign does not seem to be
related to the letter V denoting the corresponding allophone in Latin either, but since the
semivowel /j/ turned out not to be based on the similar vowel I but rather the consonant G,
it is tempting to compare the /w/- rune with B, the most similar consonant. And the
similarity is indeed striking; the /w/-rune being a 'subset' of the /b/-rune, having one of the
two pairs of twigs removed so that /b/ ends up being denoted by a doubled /w/-rune. Like
the /g/-rune, the /w/-rune has a shape that is so simple that its similarity to an unrelated
letter is unproblematic.
Since the /p/-rune is not based on the letter P, it is probably derived from another rune
after the earliest stage, when the in_uence from the Latin alphabet had ended. The fact that
this rune was later replaced by the /b/-rune suggests that its form might have been based on
this. Graphically this is conceivable, as no other rune or letter is more similar to the /p/-
rune.

'Why?' revisited
Why did some Germanic tribes start writing? Because they saw the romans doing it and
realised that it was a good idea. And why didn't they just use the Latin letters? They did, as
far as it was possible given the di_erences in writing implements and phonemic structure.
The fuþark turns out to be a legitímate child of the Latin alphabet, its 'otherness' arising
from its environment and its 'sameness' from inheritance.