Here I will make an additional post for those who want an expanded commentary on the post below from the linguistic perspective in particular. Originating as an outer garment (thick, sleeveless felted cloak of religious and cultural significance) that is loaned and then historically adopted as a tribal name for a particular group of Iranians in the Zagros by 220 BCE, and later transforming into an ethnonym in the Islamic age, all the while still maintaining a secondary outer cloak garment meaning. I will walk through the derivation stage by stage, then go through the related forms found across the Iranian world, both ancient and modern. *kr̥t-í- (Proto-Indo-European). This is the root *kr̥t- plus the suffix -i-. The root is *(s)kert-, the t-form of *(s)ker- “to cut” (Pokorny 941–942). One might ask whether the root had a laryngeal, which would change its Iranian outcome. It did not: Sanskrit kṛntáti “cuts” has a short ṛ, whereas *r̥H would have given īr or ūr. *kr̥ti- (Proto-Indo-Iranian). The form is unchanged from Proto-Indo-European. The meaning is “cut”: something that cuts or is cut. On the Indic side it gives Vedic kṛtí- “dagger”, a weapon of the Maruts in Rigveda 1.168.3, which keeps both the r̥ and the accent. It is a different word from *kr̥ti- “making”, from *kʷer- (Avestan ākərətiš): the shape is the same, but the root and meaning differ, so the two must not be mixed. *kr̥ti- (Proto-Iranian). The form is again unchanged: r̥ survives as a syllabic r, which Avestan writes ərə. The main meaning stays “cut”, and a secondary sense appears, probably early in Iranian: a garment. The step from “cut” to “garment” is a natural one, since a garment is a piece of cut or tailored cloth, and the Iranian sources make the link themselves. The Pahlavi commentary renders Avestan kərəti- as kirrēnīdag “cut”, and Bailey derives the Khotanese garment word from kart- “to cut, tailor”. *kur(ə)ti- (the u-forms). Here r̥ became ur. That is not the expected outcome, but it is not a problem, for two reasons. First, the u is not reconstructed out of nowhere. It is attested, in Kýrtioi by 220 BCE and in Parthian kórtēn. Second, ur after k is a known Iranian outcome: Balochi has kurt “did” < *kr̥ta-, and ur is the usual outcome in Parachi, Sanglechi-Ishkashmi and Shughni (Clayton 2022). So the u comes either from a western dialect of the Balochi type, or from an eastern u-form that spread west with the garment, as the borrowed name of a traded item. This word family shows that kind of travel twice: the *kurət- forms in Ossetian, Bactrian and the Tarim Basin are loans, and later Sogdian kurθ- spread west as Persian kurta. Persian itself cannot be the source, since it gives ur only after labials (murd “died”, burz “high”) and ir after k (kird “did”, kirm “worm”). *Kurt(i)- (garment word and tribal name). Polybius records the Kýrtioi (Latin Cyrtii) as slingers for the Median satrap Molon in 220 BCE. Livy has Cyrtian slingers at Magnesia in 190 BCE, and Strabo places them in Media Atropatene and Persis. The match is sound for sound: κ = k, υ = u, ρ = r, τ = t. The weak ə after the r, if it was there, has been lost. Greek had δ for d, so the τ shows a real t at this date. With no suffix after the -i, the i never became y, so no θ arose as it did in Sogdian. The word kept its garment sense alongside the name, because its descendants still carry it at both later stages. Parthian kórtēn shows it at the t-stage (below). Arabic kurdī, a plain wool wrap recorded from the narrator Ibn Abī al-Zinād (d. 790 CE) in Abū Dāwūd 915, shows it at the d-stage, though it can also be read as “the Kurdish one”. A garment becoming a tribal name is nothing unusual: the Kurdish Sanjabi are said to be named after squirrel-fur tunics, and the Turkic Qarapapaq and Karakalpak both mean “black hat”. *Kurt (early Western Middle Iranian). The final -i is lost. Parthian has kórt-ēn, which Hesychius explains: “The Parthians call by this name a garment that boys receive when they reach manhood.” Schmitt lists it among the Parthian garment words in Hesychius. Latte’s edition traces the gloss to Diogenianus, who wrote under Hadrian, so it dates to 138 CE at the latest, with the t still unvoiced. Two details belong only to the Greek: the -ēn is the Greek accusative ending, and the o is a normal Greek spelling of a short u. Greek had no letter for a short u, so older writers used υ (Kýrtioi; Kŷros for Old Persian Kuruš), and later ones often used o (Πάκορος for Parthian Pakur, Ὀσρόης for Husrōw). *Kurd (later Western Middle Iranian). t became d after r, a regular change in Parthian and Middle Persian (compare Middle Persian kird “did” and murd “died”). It came after 220 BCE, when the Kýrtioi still had t, and was roughly complete by the 3rd century CE. The name did not stay *Kurt, as Balochi kurt did, because the Kýrtioi lived in Media and Persis, inside the area where Parthian and Middle Persian voiced t after r. This gives Kurd, the form found in Middle Persian (Pahlavi kwlt’) and from there into the early Islamic period. In Arabic, kurdī is loaned to mean a wool wrap in Abū Dāwūd 915, from the narrator Ibn Abī al-Zinād (d. 790 CE), and Ibn al-ʿArabī (d. 1148 CE) repeats it as kurdiyya, quoting the same narration. In Ottoman Turkish, Meninski (1780) gives kürdiye as “cloak, short garment”. Redhouse (1890) gives kürdiyye and kürdekī as “frock coat with short sleeves” and kürdī as “woollen cloth or garment of Kurdish make”. Lady Mary Wortley Montagu (1717) describes the curdee as “a loose robe”. Now for the other Iranian forms: Avestan kərəti-. This is a religious outer garment worn over the sacred shirt and belt (Nērangestān 91); the belt counts only if it is tied on the shirt beneath it. Avestan ərə writes r̥, so this is *kr̥ti- itself. The meaning is secure at §91. A second instance at §87 depends on Waag’s emendation, and Bartholomae and Bulsara read that passage as a different word, but nothing here depends on it. The ending -īš could also fit an ī-stem, but Vedic kṛtí- and Old Ossetian *karti- point to an i-stem. Ossetian kærc “fur coat, sheepskin coat”. It comes from Old Ossetian *karti-, which Abaev says “corresponds perfectly” to kərəti-. The sound changes are regular: r̥ > ar, a > æ, and final -ti > -c. Khotanese kīḍ- (in kīḍakyä “garment”). The changes are again regular: r̥ > ä, rt > ḍ, and the vowel becomes ī before the -i- of -ti- (Bailey 1979). A Late Khotanese document has kīḍakyä begaṃnä śau, “one garment, woven”, written before Khotan fell around 1006. Khotanese cannot show whether its vowel came from zero-grade *kr̥ti- or a full-grade *karti-, but the zero grade is the one Avestan and Vedic show, so it is the simpler assumption. The *kurət- loans: Ossetian kuræt (Iron kʷyræt) “beshmet”, Tarim kuratu “shirt” and Bactrian kuratīg “shirt”. These cannot be inherited. A native Ossetian t would have become d, and native -ti gives -c, as in kærc. Native Bactrian would have ιρ and δ (compare κιρδο “made”). So they are borrowings from an unknown donor, and their vowel after the r shows that the u-form had a weak ə there. Each language adapted it in its own way: Ossetian ə > æ (and in Iron u > y, hence kʷy-), Tarim ə > a with a local ending -u, and Bactrian ə > a with ο for u and the suffix -īg. Bactrian kuratīg appears in a letter about “linen shirts from Balkh”, and Tang glossaries of the 8th century give the Tarim form as the “Sanskrit” for Chinese shan “shirt”. *kr̥θy- (the θ-derivative). A suffix beginning with a vowel turned *kr̥ti- into *kr̥ty-. Since ty > θy is a Proto-Iranian change (Avestan haiθiia- < *satya-), this derivative is old. Sogdian has both kərθ- (krδʾkh, in the early 8th-century Mugh documents) and the u-form kurθ- (kwrδʾk, qwrθy): the regular vowel and the u-form side by side in one language. Both lost the y after θ, as in xēpθ “own” < *xwaipaθya- (Yoshida 2016). kərθ- shows the ordinary Sogdian r̥ > ər, and kurθ- the u-form. That θ cannot come from t, because Sogdian keeps old t. Yaghnobi kŭrta and kŭrsa “shirt” inherit the word, with θ > t in the west and s in the east. The later kurt- loans. Persian kurta, kurtak and kurtī, Kurmanji kurtek “robe”, Tajik kurta, Pashto kurta, Parachi kurtī, Shughni kurta, Armenian kʻurtik and Arabic qurṭaq (“arabicised from kurta”) all come from Sogdian kurθ-. They have no vowel after the r (unlike Bactrian kurat-), their θ became t, and that t was never voiced, so they were borrowed after t > d. This is also why Kurd has d while kurta has t: Kurd was already in Western Iranian before t > d, while kurta arrived from Sogdian after it.

MAJOR development. After several years of on and off attempts, I have finally solved the etymology of the ethnonym Kurd, which currently has no solution in academia. It comes from Proto-Indo-European *kr̥t-í- “cut” → Proto-Indo-Iranian *kr̥ti- → Proto-Iranian *kr̥ti- → *kur(ə)ti- (“cut”, but also an outer “garment”) → *Kurt(i)- (adopted as a tribal name) → *Kurt → *Kurd (by the 3rd c. CE) → Kurd (from a tribal identity to an ethnonym). The first image, an etymological tree, sets out and justifies each sound change, if you want to see how the derivation is linguistically proven. It also contains the historical details on each form where attested. X reduces image resolution, so to read it clearly, expand the image and open it in a new tab. The second image shows the garment in question: the thick, tent-like traditional shepherd’s cloak made of dense felted wool, or wool mixed with goat hair. Specifically, the name Kurd comes from this outer-garment sense, which was later adopted as a tribal name. Among Iranian languages, the garment word first appears in Avestan as kərəti-, in instructions on how Zoroastrians should wear it together with the sacred shirt and belt. In Parthian, kórt-ēn was the garment boys received as a coming-of-age gift when they reached manhood (which is linguistically a close match to the word Kurd, as the stem's predicted outcome of the Parthian would eventually become Kórd/Kurd since that gloss predates t > d). The word has cognates across the Iranian world, living and extinct, such as Khotanese, Ossetian, Sogdian and Bactrian. There is historical evidence that Kurd still carried a secondary garment meaning in the early Islamic period, as the word was borrowed into Classical Arabic for a type of garment: kurdī, from the narrator Ibn Abī al-Zinād (d. 790 CE in Baghdad), in Abū Dāwūd 915 (before 889 CE), and kurdiyya in Ibn al-ʿArabī (d. 1148 CE), among others. The hadith reads: “…from ʿĀʾisha, with this report. He said: and he took a kurdī that belonged to Abū Jahm. It was said: O Messenger of God, the khamīṣa was better than the kurdī.” This is important because it shows the t > d form unlike Kurt-(ek), and other forms that passed through west Iranian languages later on via Sogdian. They are still from the same root however. Tribal names and clans very often come from places, animals, objects and so on. The Kurdish tribe Sanjabi is named after the squirrel fur, Herki is named after a citadel, etc. Kurds are first attested in history specifically as a tribe by Eratosthenes, Strabo and others. The older reading of Kurd as “nomad” has no etymological basis in any Iranian language, and many Kurdologists/Iranologists today regard it as an anti-Kurdish narrative with no basis in the historical record (see Boris James). Some earlier academics unknowingly built on Arabic mistranslations and deliberate misreadings, which are now corrected. There were also at least 11 Iranian nomadic tribal organisations on the plateau before Islam with no Kurdish identification, and 5 Iranian tribal organisations on the plateau that never had a Kurdish tribal affiliation which completely contradicts this narrative. The main sound-change question left open in my tree is how r̥ became ur, since it normally gives ar or ər. There are several resolutions, but it is not yet clear which one it is. The u-form *kur(ə)ti- may have spread by borrowing, as a garment from the East Iranian lands: ur is the usual outcome in Parachi, Sanglechi-Ishkashmi and Shughni (Clayton 2022), and Sogdian kurθ- probably passed into Persian kurta. Or it may be a regular change in some western dialect, as in Balochi kurt “did” < *kr̥ta- (Korn 2003). Several scholars already tie the garment word to the root for “cut”. Abaev derives Ossetian kærc “fur coat” from Old Ossetian *karti-, which “corresponds perfectly” to Avestan kərəti- “garment” (that is, *kr̥ti-), perhaps from *(s)ker- “to cut”; he calls Ossetian kuræt a loan. Bailey derives Khotanese kīḍakyä “garment” from kart- “to cut, tailor”, comparing kərəti-. Henning took Sogdian kurθ- “garment” as the same word as kərəti-, and Benveniste linked Persian kurta to the Sogdian. Sims-Williams, publishing Bactrian kuratīg “shirt”, calls it “a word of unknown but probably non-Iranian origin”, which would leave room for a substrate source such as the BMAC (2007: 224). Although Sims-Williams did not gloss through the full extent of the Iranian forms, or look too deeply into it. I have opted otherwise, and agree with the other scholars who do derive it through Iranian, because of the following reasons. Avestan kərəti-, Ossetian kærc and Khotanese kīḍ- show the regular outcomes of *kr̥ti- at opposite ends of the Iranian world. The root has an Indo-European etymology, *(s)ker-t- “cut”, shared with Vedic kṛtí- “dagger”, and the Pahlavi commentary renders kərəti- as kirrēnīdag “cut”. The forms Sims-Williams compares are the u-forms, and they can all be accounted for within Iranian. The Sogdian θ comes from the regular change ty > θy. Persian kurta, Arabic qurṭaq, Pashto kurta and Ossetian kuræt are loans. A word with a regular Iranian etymology doesn’t need a substrate source. To clarify for anyone who doesn't know, the stem is important for derivation as it carries the evidence of descent, while the the suffixes don’t necessarily. A stem passes down through Iranian forms and goes through its regular sound changes, so kurt- with t, kurθ- with θ and kurd with d each show where a form sits on the tree. Suffixes are added afresh by each language and are often swapped or borrowed depending on the application and context. Given this etymology, the identification of the Cyrtians with the Kurds is even stronger than it already was, as the forms are an exact match. Identifications with the Carduchi, Qardu, among other folk proposals have become even more improbable, both linguistically and historically, than they already were.

Publication in the journal of Iranian Studies hosted by Cambridge University Press, Boris James debunks the long-held view that “Kurd” in early medieval Arabic sources simply meant “nomad”. He argues that this view was a “hyper-reaction” against pro-Kurdish essentialism. It was reinforced by a misreading of Minorsky, and by misappropriated historical data. He answers it by examining both modern scholarship and Arabic sources from the 9th to the 15th century. He shows that “Kurd” was already the name of a people in the earliest sources, first defined in relation to Arabs and non-Arabs, and that from the 11th century this Arab centric historicity dies down, and Kurds are defined independently from this categorisation. He states it remains impossible to know, for technical reasons, the precise moment when individuals from a specific territory began to see themselves as Kurds from a medieval point of view because the sources offer very little in that regard. The clearest historical evidence of that in the medieval age begins in the 1100s CE due to the Ayyubids. James concludes that in medieval Arabic sources “Kurd” was a genuine ethnonym. He builds this from medieval authors writing between the 9th and 15th centuries. For the early period he uses al-Jahiz, Ibn Durayd, al-Ya’qubi, al-Tabari, Hamza al-Isfahani, al-Mas’udi and Ibn Hawqal. For the later period he uses Ibn al-Sam’ani, al-Wahrani, ‘Imad al-Din al-Isfahani, Ibn Shaddad, Yaqut, Ibn al-Athir, Ibn Khallikan, Abu Shama, al-Khazraji, al-Dhahabi, al-‘Umari, Ibn Khaldun and al-Maqrizi. His target is a modern orthodoxy resting on Vladimir Minorsky. In response to some pro-Kurdish authors who treated Kurdishness as a fixed essence, a number of scholars accepted that “Kurd” eventually became an ethnonym but treated its early medieval meaning as simply “nomad”. James’s examples are Basil Nikitine, Wadie Jwaideh and Martin van Bruinessen. He also challenges Jean Aubin, who read “Kurd” as describing a pastoral, tribal way of life, and Hakan Özoğlu, who saw it as an outsiders’ term. In James’s account, three things drove the “nomad” reading. It was a “hyper-reaction” against pro-Kurdish essentialism. Fixing the term’s meaning resolved the puzzle of why groups such as the Lurs and Gouranis were called Kurds then but not now. And these scholars mixed early and late evidence, with Nikitine and Jwaideh stretching Minorsky’s narrow remark about 10th-century western Iran into a general rule. To correct this, James draws on Jan Retsö’s parallel work on the word “Arab”, Lakoff and Johnson’s view that concepts are defined through one another, Fredrik Barth’s theory of ethnic boundaries, Catherine Quiminal’s “contexts of meaning”, Ian Hacking on social construction, and Gabriel Martinez-Gros’s reading of Ibn Khaldun. He also uses Stephan Conermann’s study of Kurds in Mamluk eyes, while rejecting Conermann’s criticism of Ibn Khaldun. Historians such as Anne-Marie Eddé, Reuven Amitai, Jean-Michel Mouton and Percy Kemp supply further evidence. James separates two historical phases and shows that “Kurd” took its meaning from its shifting relationship with other ethnonyms (‘arab, ‘ajam, turk), not from any fixed trait. His starting point is how ethnonyms work. Unlike words such as “Muslim”, nothing in the word “Kurd” decides who belongs, and Mas’udi’s legend deriving it from “driven out” merely confirms an existing negative image. Being associated with nomadism no more makes “Kurd” mean “nomad” than calling the Laz “the Belgians of Turkey” makes “Laz” mean “stupid”. In the first phase (8th to 11th century), authors defined Kurds through two other categories. An Arab element, expressed through Bedouin genealogies and shared pastoralism, linked them to Arabs. A non-Arab (‘ajam) element, chiefly language, set them apart. This was a comparison between peoples, not an equation. In the second phase (11th to 14th century), “Kurd” stands on its own. Yaqut distinguishes Kurds from Bedouin, and Ibn Khallikan lists Kurdish origin separately from Arab and ‘ajam. Al-‘Umari describes Arabs who “adopted the lifestyle of the Kurds” and ‘ajam who “have Kurdified”. Maqrizi reverses Mas’udi’s weighting, rejecting the Arab genealogy and stressing the non-Arab element, which shows the term had its own dynamic. James suggests three causes for the change: new contacts with Turks, Kurdish and Daylamite dynasties, and the Turco-Mongol invasions; closer knowledge of Kurds once they joined the armies and cities of the great empires; and rivalry with the Turks. He then identifies six frames in which “Kurd” was used. Two continue from the first phase: Arab versus non-Arab, and Arab tribal genealogy. The third is Ibn Khaldun’s Bedouin cycle, in which Kurds rose with the Zangids and Ayyubids and returned to the margins under the Mamluks. The fourth is a civilisational stereotype of Kurds as violent, rebellious, heterodox and poor in Arabic. The fifth is geographical, tying Kurds to named territories such as Bilād al-Akrād (“land of the Kurds”) as their tribal territory shifted westward. The sixth casts the Turk as the Kurd’s “Other”. Finally, against Aubin and Özoğlu, whose views leave no room for Kurds’ own role in shaping the term, he finds evidence that Kurds identified as such themselves. ‘Isa al-Hakkari urged a rival to back Saladin as a fellow Kurd rather than leave power to the Turks. Abu al-Hayja’ wrote that Kurds and Turks would never obey each other. Al-Mashtub cut off his son’s Turkish-style braid because Kurds did not wear their hair that way. This rivalry peaked as Ayyubid rule ended, and the creation of a Mamluk “dynasty of the Turks” allowed later writers to depict the Ayyubids as the “dynasty of the Kurds”. Being identified as a Kurdish dynasty also weakened the Ayyubids’ claim to rule universally, beyond any single people.

This map plots every passage in which Ibn Miskawayh writes about Kurds in his history Taǧārib al-Umam. There are 35, describing events from 685 to 980 CE, coloured by period and placed at the towns and districts the text names. They stretch from Mosul, Šahrazūr and Āḏarbāyǧān across the Zagros to Ḫūzistān, Fārs and Kirmān. Most come from the 10th century: the Kurdish ruler Daysam in Āḏarbāyǧān, Ḥasanūya in the Ǧibāl, and Kurdish troops, raiders and chiefs dealing with the Buyid court. The later continuations printed with the book are left out, as they are by other authors. Abū ʿAlī Miskawayh (c. 932–1030) was an Iranian historian and philosopher from Rayy. He served as librarian to the Buyid vizier Ibn al-ʿAmīd and later worked for ʿAḍud al-Dawla. His Taǧārib al-Umam (“The Experiences of Nations”) is a universal history from the ancient Iranian kings to 979–980 CE. For the earlier centuries he abridges al-Ṭabarī. From the 950s on he writes as an insider, often from what he saw himself or heard from Ibn al-ʿAmīd. This makes the book one of the main contemporary sources for the Kurds of the 10th century. Open map in a new tab for better resolution.

I’ve tabulated the Iranian tribal groups of the plateau that historical sources describe as nomadic. Here you can see many examples of non-Kurdish Iranian nomads, which refutes the notion that “Kurd” was merely a generic term for Iranian nomads. The sources treat the Kurds as a distinct group and name other nomads separately. Take for example Sasanian-era Syriac texts distinguish the nomadic Kermanis (Qarmānāyē) from the Kurds, and Yāqūt describes the Baloch of Kerman as “a people like the Kurds” rather than calling them Kurds, which directly implies a distinction. The notion that “Kurd” means Iranian nomad stems from a misreading of cherry-picked comparative remarks in Arabic and Arabic-influenced historiography, such as “the Kurds are the Bedouin of Persia”. It projects the Arab division between the ʿarab (settled) and the aʿrāb (nomads) onto Iranian society. We know this not to be the case because Sassanian social structure is quite decently attested, and this isn't how society was divided up.