History of Elven Writing Systems/Age of Starlight

Quenta Eldatencelion
Age of Starlight
Early Tengwar
*An alternative mode
Certhas Daeron
First Age
Mode of Beleriand
Runes of Gondolin
Second Age
Quenya Mode
Angerthas Daeron
*Angerthas Moria
Third Age
Classical Quenya Mode
"General Use"
*"General Use" Westron
*"General Use" Sindarin (Sindarin Tehtar Mode)
"Full Writing" with "General Use" Consonants
*Sindarin "Full Writing" with "General Use" Consonants
*Westron "Full Writing"
Angerthas Erebor



Rúmil of Tirion was a Noldorin scholar and the composer of Ainulindale. Before that he was the inventor of the first elvish writing system in the Valian Year 1179. It has been suggested that his system was based on an even more ancient, unrecorded writing system of the Noldor, but this is hardly probable. The Sarati (“significant marks”), as he called his letters, were ideal for writing both on stone and on paper, but we have only seen what seems to be their “calligraphic” versions.

Several different writing-directions were used with the script. Rúmilian texts were written in rows or in columns like Japanese, LTR like the western scripts, RTL like eastern scripts or boustrophedon (RTL–LTR–RTL etc.) like some ancient Greek inscriptions.

Rúmil decided that the vowels should be written with diacritics and not with independent letters, since, according to the linguistic perceptions of the Valinorean loremasters, vowels were considered as “colours” of the consonants.

Todate various interinconsistent modes of Sarati have been published, from various stages of Tolkien's youth and development. They are mostly used for English or early Qenya words and phrases. Unfortunately there is no definite, compatible to Lord of the Rings-Quenya mode of Sarati known.

Early Tengwar


Fëanáro Curufinwë (better known as Fëanor) invented the tengwar (“signs representing audible phonemes”), strongly influenced by the Sarati, in the Valian Year 1250. From that time on, the term Sarati, from the general notions of “letter”, only refers to the Rúmilian writing system.

Rumilian Influence


With the first sight one might observe the stylistic influence of the Sarati. Especially two of the sarati (namely P p and B b) are visually similar to the shape of a tengwa, since their morphology is consisted of a “line” and “curves”. The organization of the tengwar follows the aforementioned concept, where the form of the letter reflexes its sound. Fëanor decided that the doubling of a “curve” should add voice to the “basic” sound, and generally the way the “curves” are doubled is similar to the “p” and “b” sarati, so these two sarati must be his main inspiration.

Fëanor therefore, by reducing the number of variable elements, produced a simpler and more consistent set of characters. The two basic elements of the tengwar were a “stem” (telco) and a “bow” (lúva), which could be combined and modified in a number of different ways. Fëanor decided that the tengwar whose telcor and lúvar were organized in a particular way, represented a certain group of related sounds: The telcor determined how the sound was articulated, and the lúvar where in the mouth it was made.

The greatest difference between the systems was that the tengwar were written from left to right.



It is more probable that the Tengwar system was not especially based on the needs of Quenya, Feanor's mother tongue; this is obvious when we compare the Quenya mode table with other, later modes, where the phonemes fit to the Tengwar table more consistently. It is therefore possible that the Tengwar started initially more like a generic phonological 'template', on which all later derived modes and spellings could be based. The dialects of Aman as well as ancient reconstructed words could then be written with an adaptation of that generic system.

The following table is based on a theory by Jim Allan, published in An Introduction to Elvish. It shows a reconstruction of the original theoretical 'template' of the Tengwar, perhaps never used in any language.

The only mode known closest to the theoretical values would be the General Use mode, of which we first hear millennia later.

Note that the second-to-last series represents voiceless consonants.

Labial Dental Velar Labiovelar
q p 1 t a k z kw
w b 2 d s g x gw
Q ph ! th A kh Z khw
W bh @ dh S gh X ghw
e f 3 th d ch c chw
r v 4 dh f gh v ghw
y hm 6 hn h n hñw
t m 5 n g ñ b ñw

Quanta Sarme


Unlike Rúmil, Fëanor considered vowels as independent sounds and not just “colours” of the consonants, so he devised the “full writing” (Quanta Sarmë).

Quanta Sarme had distinct letters for vowels (ómëar or óma-tengwi), and was used by the Loremasters for special purposes only[1]. No texts of this system are known to us, and its usage and characters are unknown, but Tolkien says that it was used in Middle-earth for other languages, like Sindarin, where the tehta-mode was inconvenient.

This system is not heard much thereafter and it seems that its usage hasn't endured into the following Ages. However by the above information, we can easily assume that it was the predecessor of the Mode of Beleriand, and the letter Osse ] is maybe the only remnant of that (otherwise extinct) system.

Original Quenya mode


Despite Fëanor is said to prefer the Quanta Sarme, he also used a more 'conservative' system which (judging by the majority of the subsequent samples) seem to have been proved far more popular. Fëanor held the idea of syllabic analysis of the words by the Sarati, and made also use of signs (tehtar) for vowels (instead of the full letters), placed over the preceding tengwar/consonants, indicating their “colour”. A consonant followed by a vowel was considered as a “full letter” (ñávëa or ñáva-tengwë). The sarat was imported from the previous system, and when a vowel had no preceding consonant, it was used as a carrier for convenience in writing. This system however was used by Fëanor mainly for tradition and brevity, favouring the Quanta Sarme instead.

Apart of the standard consistent characters, there were also additional tengwar which don't fit in the structure. These are usually modifications of the standard tengwar. Hyarmen is a modification of Charma, Úre is of Wilya, and Anna derived from an earlier sarat.



According to Appendix E (mixed with other recently published sources) and our knowledge on Quenya phonology evolution[2], we can reconstruct a Tengwar table, hopefully as close to the original Feanor's concept as possible:

Tincotéma Tyelpetéma Parmatéma Calmatéma Quessetéma
1 t
q p
a k
z kw
! th
extended grade
extended grade
Q ph
extended grade
A kh
extended grade
Z khw
extended grade
2 nd
w mb
s ng
x ngw
3 th > s
Thúle > Súle
e f
d ch
c chw
4 nt
r mp
f nk
v nkw
5 n
t m
g ñg-
b ñgw-
6 ɹ
y v
h 3 > -
n w
Additional Tengwar:
7 r
u rd
j l
m ld
8 s
I s
Silmë nuquerna
k z > r
Ázë > Árë
, z > r
Ázë > Áre nuquerna
9 hy
l *y
. w
½ h
` *(short carrier)
~ (long carrier)



Tengwar names: The names given in Appendix E were based on the 3rd Age table composed in Gondor. It is not known if this arrangement was given then or existed since Fëanor’s days. It is possible, however, that some of the known names may have replaced earlier, unrecorded forms. The best that can be done here is to give the oldest known names (e.g. Charma instead of Aha). This is our evidence:

Anna originally represented 3, a sound from primitive g (cf. *galadâ > *galda > *3alda > alda). The 3 sound was early lost, and Anna was used as an initial vowel carrier to indicate an assuming “vanished” initial consonant wherever words begun with a vowel (note that its "ancestor", the sarat I, was also used for 3 before being used as a carrier. It was inherited as a carrier by Fëanor, and later it replaced initial Anna). This usage tried to explain the relation of words between Noldorin words starting with a vowel where Telerin had g- (cf. Quenya alda vs. Telerin galla), but did not explain all the cases of words without an initial consonant (alca, according to this, should be spelled *hDjaE that time, although it had never been **galka before). The problem with the name is that Anna is given as derived from the root AN, and not *GAN (although there is some evidence that the latter root is the correct, and therefore its archaic form was *3anna).

Halla represented an archaic breath h, that survived from primitive H only in Amanye languages, while Charma represented ch, that derived from Primitive KH. The problem is that the h of the word halla evolved from the latter sound: KHAL > *khallâ > halla, therefore pronounced *challa in Fëanor’s time. We are lead to the conclusion that there must be another ancient unrecorded name for this letter that had the breath sound h!

Óre: Óre is given by Tolkien as /ɹ/: “weak (untrilled) r, originally occurring in Quenya” (Appendix E). This quote suggests that Quenya originally contrasted /ɹ/ and /r/, since there would be no need for independent letters if the two sounds were merely allophones of each other. In this case, however, the original distribution of the letters Óre and Rómen must have been different from the one we observe at the end of the Third Age, where both letters are in perfectly complementary distribution, no contrast being possible. If Óre is the original name of this letter, this could mean that it is one of the words with original /ɹ/.

The problem about the sound of /ɹ/ is that there is no other evidence apart from the mentioned quote.

Extended stems: As we are told in Appendix E, the original Fëanorean alphabet contained a Grade of “extended” stems, both raised and lowered. The usage of those tengwar (whose names, if they had, are of course unknown) was to represent spirant sounds. As we know, no Amanya language possessed spirant sounds, since the Primitive spirants th, ph, and kh had already became th, f and ch in Quenya. It is possible that they were used for recording Valarin or maybe archaic forms.

Tyelpetéma: Appendix E informs us that Quenya also made use of a palatal series, the Tyelpetéma. Christopher Tolkien, who made the names known, notes that the names are given in a number of different formulations, and he cannot determine which were the standard ones. The only difference from the Tincotéma was the underposed (for tengwar with raised stems) and overposed (for tengwar with lowered stems) dots. Those tengwar, not included in the table of the Appendix E, must have been ommited from the standard letters.

Áze: The /z/ sound later changed to /r/ only in the Noldorin dialect, and this letter took the name Áre.

Azya: This tengwa is given as Arya, and is attested either as a Rómen or as an Óre with the overdots. Since the sound ry derives from original sy > zy (still present in Vanyarin), the form of the letter should be Áze with the dots, changed after z merged with r. But then this tengwa should be placed among the Additional Tengwar, for it is not consisted of a telco and a lúva.

Yanta: Yanta's shape reminds of the Rúmilian letter for y. It might have represented this sound too (Yanta probably written *lD4#, while this word should read **ainta in 3rd Age spelling). In our attested examples of 3rd Age texts, it occurs only as the -i of dipthongs (in lE, lH ai, oi etc.). Maybe e.g. tuilë was originally spelled *1UlUj$ (tuyulë) like in Rúmilian orthography, before simplified to 1lUj$. Cf. Úre.

Úrë: Besides that in the attested examples of 3rd Age, Úre appeas as the -u of the dipthongs (.D, .F au, eu etc.), its original use and etymological relation to the word úrë itself are unknown. Since it’s attested in diphthongs in our samples like Yanta is, we can suppose it was used for intervocalic w, which later became v. Maybe taurë was originally spelled *1D.D7R (tawarë) before simplified to 1.D7R, while for initial w, only Wilya was used. Jim Allan suggested that úre should be written as .J7R, a function that reminds of the Rúmilian spelling of long u… – or maybe it was used as u in Quanta Sarme, like in the Mode of Beleriand? Cf. Yanta.

Thúlë: In Noldorin Quenya the sound th had merged with s since very early. The word thúlë thus became súlë.

Ñoldo/Ñwalmë: Those sounds were found only initially. Sometimes written ng- and ngw-, but not to be confused with Anga/Ungwe, which are used only medially.

Nuquernë Tengwar: The table of Appendix E gives us the inverted forms of Silmë and Ázë, which were used when followed by a vowel (since they were too tall to receive a tehta). It is not known if these forms were invented by Fëanor or by later users, but Silmë Nuquerna was used in the Mode of Beleriand (as a vocalic y), which makes us think it was already present in Valinor. If we are to suppose that the original form of Arya was an Ázë with dots, there must also have been an inverted form of it (Arya Nuquerna).



Like the Sarati, the Amanya system of the tengwar is also unknown at this point. According to the theories mentioned above, they will be tentatively reconstructed, showing some tengwar transcriptions of various Quenya words as pronounced in Fëanor’s time. For better understanding the words and the connection between spelling and etymology, alongside are given the etymological roots or the better attested later Quenya forms:

  • dE6U, *cheru < KHER
  • 1E.6R, taurë
  • 1`C.D6R, *tawarë > tavar
  • `C.D,Fj, *awazel > aurel
  • hEm#, *3alda > alda
  • 1Ulj$, tuilë < TUY

Possible examples of Quanta Sarme


We know absolutely nothing about the Quanta Sarme orthography and the following are purely conjectural, based solely on what we know about 'classical' tengwar.

  • ],Ö], azya
  • l]y]55], Yavanna
  • ½]5., hanu < HAN

An alternative mode


For variant spellings, the general hypothetical 'template' above could have been implemented. For example, we have no clue on how the Vanyar and Teleri used the known Tengwar, since their dialect possessed sounds not found in Noldorin, like independent d and b. The Vanyarin poem’s name, Aldudénie, is known from the Silmarillion, and the word ulban, “blue”, is also adapted from Valarin by the Vanyar. They could probably write these words by modifying this flexible system according to their language (like the Noldor and the alien speakers did later in Middle-earth), but some sources state that the Vanyar kept the Sarati[citation needed]. Tolkien notes that the system of Fëanor provided the means of expressions for more sounds than those occuring in Quenya and Telerin, but he doesn’t mention them or the reasons.

In Addenda et corrigenda to the Etymologies, an old version of the tengwar by Tolkien appears. Some tengwar have different names and values. It is not known if this mode coexisted with the one presented above, or if one evolved after the other.


q p
a k
z kw
w mp
s nk
x nqu
3 s
e f
d h
c hw?
4 st
f ht
v skw
t mb
g ng
b ngw
6 n
y m
h ñ
n w
Additional Tengwar:
7 r?
u rd
j l
8 s
i r?
9 hy
o hw?
l y
. w?/v?
` (carrier)
Vocalic Tengwar:
` a
~ á
`V e
`B i
~V é
*Ére?? *Éra??
~B í
] o



During that time, the Sindar of Beleriand began developing an alphabet for their language. Its letters were entirely made for carving on wood, stone or metal, hence their angular forms and straight lines. These letters were named cirth (sing. certh).

The assignment of values was unsystematic. The form of a certh was consisted of a stem and a branch. The branch was found usually on the right side of the stem, and sometimes on the left, but with no phonetic significance. Therefore 3 would just be an alternative form of 1.

The following cirth were given by Tolkien as the most ancient of the later Angerthas. They had possibly different values at that time than the ones given below. Since we lack any other evidence though, we will use the known standard values of the letters to have a framework of discussion and make our comparisons; this is not so far-fetched as it seems: it's possible that at some later stage, some sound values were 'stuck' to some of the cirth, setting some stable basis for Daeron to make his final standardization.

If we are to accept the known values to these old Cirth, we will observe that they don’t cover all the sounds of Sindarin, since we are missing rh, lh, mh, v, y, œ. The most possible explanation is that those early Cirth were used for the “Old Sindarin” tongue, and many of the above mentioned sounds indeed didn’t exist in that language. However this doesn't explain the absence of sounds w and a. This indicates that some ancient, unknown cirth could have existed, but didn’t make it to the later systems and therefore aren't given by Tolkien; a fuller table therefore can't be reconstructed.

As for the vowel usage, perhaps the certh for u possibly was used for w (like in early latin orthography). The certh for a can’t be guessed, so maybe this sound was “meant” (like in some Quenya Tengwar Modes). More possibly it was one of the two cirth whose value can’t be determined, or it was one of some other cirth that did not survive till the later Angerthas.

Long vowels were evidently indicated by doubling (cf. the known certh for long o, n, is a simplified form of bb).

Note: The term “Old Sindarin” is used hypothetically to facilitate understanding of Sindarin etymology, although we don’t know if it actually existed, and when. The term “Old Sindarin” is not a term used by Tolkien, but the term, as well as the language, are based on “Old Noldorin” as known by him.


1 p 8 t e c R r l i
2 b 9 d r g a l S u/w?
5 m? @ n u ñ g s/h z e
6 ? # h/s % h/s h ss? b o



5: The original value of this certh wasn’t given by Tolkien, but he mentions that in later Angerthas, it took the value hw after 6 became m (for reasons explained below) while he doesn't give us the early certh for M. We can infer that this was the certh for M judging by both its 'labial' shape, and the symmetrical shape used for nasals, like @, u.

6: As mentioned, this certh will have the value m in later Angerthas, for reasons explained below. Its original value can’t be guessed (maybe the certh for a?), but judging from the “labial” shape, it could be w.

#/%: As Tolkien mentioned, the reversal of the stem didn’t had any phonologic significance. These Cirth were interchangeable and used for h or s depending on when g was s or h. This would perhaps apply to these early Cirth as well

g: Similarly, this Certh was h when #/% were s and vice versa.

h: This Certh will have the value ss in later Angerthas. It must had another unknown value before (maybe the certh for a?).



As mentioned, it is probable that this primitive Certhas were used “Old Sindarin”, and we can try to see how some words could be written. The spelling exemplified here assumes that like in Latin, a single certh was used for w and u, and similarly for k, t and p for kh, th and ph. Since we can’t guess the form of a, we will consider that it was actually ommited.

  • g82Rb@, STBRON (*sthabron), “carpenter”
  • Sbbl, WÔI (*wôia), “envelope”
  • bReaa, ORKLL (*orkhalla), “superior”
  • ua52z, ÑLMBE (*ñalambe), “barbarous speech”
  • lSRl@z, IURINE (*yurine), “I run”
  • Sz9, UED (*weda), “bond”

Certhas Daeron


In the Valian Year 1350[3], Dairon (better known as Daeron), the Minstrel and Loremaster of Doriath, reorganised the cirth and added new ones, making the extension of the cirth known as Angerthas Daeron (“Long Runerow of Daeron”), used for inscribing names in Menegroth. The Dwarves working for Thingol liked them and adopted them, making them known also in the East, beyond the Mountains.

The nature and the timeframe of the standardization however are obscure; some point the similarities between q et 1 with 1 and 8 as signs that Daeron was influenced by the Tengwar, therefore this mustn't had occurred before the return of the Noldor, which means during the First Age. Unlike the previous system, the reversal of the certh had a phonemic significance: reversed cirth were softer versions of their originals. This also gives us another information: perhaps lenited consonants must have started to occur in Sindarin around that time.

Tolkien mentions that at that time (or maybe later) a sound of mh was needed. The most appropriate solution was to revert the certh for m to indicate its softening, but 5 (we assume it was this certh) couldn’t be reverted, so m was given to 6 (which until then had a value unknown to us), and 5 got the value of hw. The same process took place with r, l etc., and maybe at that time the distinct cirth for semivowels and umlauts, like w, y and œ, were employed.


1 p 8 t e c R r
2 b 9 d r g T rh
3 f 0 th t ch a l
4 v ! dh y gh s lh
5 m?>hw @ n u ñ g s
6 ?>m # h h ss
7 mh
l i, y- S u z e c a b o
D ú x é v á n ó
F w m œ
G y



There are no Angerthas Daeron texts by Tolkien either, so we cannot know how the text forms can look like. By some English specimina we see at least that words are separated by dots, phrases by colons and longer periods by 3 or 4 dots. We don’t know if Angerthas Daeron, like Tengwar, used other diacritics for modificators or combinations like nd, mb, dw, gw or diphthongs. It seems, however, quite straightforward, and for now we use them in close correspodence to the latin letters. The following examples attempt to show words as would be in the Age of the Starlight (Daeron’s time), so the first and last examples should be Gelydh and Elvellon in the 3rd Age.

  • c@@b@IluIumaG!, Annon-iñ-ñGœlydh, “Gate of the Noldor”
  • lItlR0I#l@, I-chirth hin, “these runes”
  • rSRz@I2x@Iz@@l, Guren bêd enni, “my heart tells me”
  • za7zaab@, Elmhellon, “Elf-friend”

Go to:


  1. Appendix D to Quendi and Eldar
  2. The Evolution of Primitive Elvish to Quenya: Valinorean
  3. The Grey Annals